fix(ecstore): preserve online writes during pool retirement (#7472)

* fix(ecstore): reconcile identical scanner backlog replicas

* fix(ecstore): type invalid decommission requests

* chore(ecstore): tighten typed-error ratchet baseline

* fix(ecstore): fence late writes to retiring pools

* fix(ecstore): share healthy pool capacity during decommission

* fix: allow active multipart uploads to drain

---------

Co-authored-by: overtrue <anzhengchao@gmail.com>

.agents/references/adversarial-validation.md

@@ -0,0 +1,33 @@
+# Adversarial Review Shape
+
+Use when root `AGENTS.md` triggers adversarial validation or for a substantial PR
+review. Paths below are repository-relative. The root finding standard and
+completion rule apply; selecting a lens does not require finding a defect.
+
+Risk and review shape:
+
+- **Exempt:** documentation, comments, formatting, or typos with no runtime,
+  build, test, or agent-execution effect.
+- **Mechanical:** renames, moves, test/tooling-only changes, and agent-rule
+  changes. Run correctness and simplicity lenses.
+- **Standard:** localized behavior changes. Run one integrated final-diff pass
+  covering correctness, simplicity, and test coverage; add only domain lenses
+  matched by the diff.
+- **High risk / substantial PR review:** high risk includes locking,
+  erasure/quorum/heal, replication, multipart, RPC, lifecycle/tiering,
+  persistence/fsync, IAM/KMS/auth, cryptography, on-disk/on-wire formats, and
+  S3-visible semantics. Cover all applicable lenses using exactly two
+  independent reviewers when delegation is explicitly authorized. Split the
+  lenses between them. Otherwise perform two fresh sequential passes.
+- **Outbound client defaults:** what `TargetClient`, `PutObjectOptions`, or
+  the remote SDK configuration sends to every replication or migration target
+  is high risk for every target class even when the change fixes one. Follow
+  the SOP in `docs/postmortems/2026-09-03-replication-checksum-default-regression.md`:
+  run the outbound target matrix, document each new env knob in the same PR,
+  and list verified and unverified target classes in the PR Impact section.
+
+Available domain lenses are security, concurrency/durability, compatibility,
+and performance. Select `.agents/skills/adversarial-validation/SKILL.md` for an
+explicit adversarial request, a high-risk change, or a substantial PR review;
+then read only its matching role references. A routine standard pass does not
+load the playbook unless the reviewer needs a RustFS-specific probe.

.agents/references/implementation.md

@@ -0,0 +1,65 @@
+# Implementation Rules
+
+Applies when changing code or running artifact-heavy work. Paths below
+are repository-relative. Read only the relevant sections during read-only review.
+
+## Worktree and Disk Hygiene
+
+- Start implementation from the latest `origin/main` and confirm the requested
+  change is not already present.
+- An existing clean, isolated task worktree is sufficient. Create another
+  worktree only when the current checkout is shared, dirty with unrelated work,
+  or belongs to another task.
+- Never commit from a shared checkout.
+- Use a task-specific branch named `<type>/<topic>`, such as `fix/...`,
+  `feat/...`, `test/...`, or `docs/...`, unless the user specifies a name.
+- Do not include agent, tool, contributor, account, or organization names in
+  branch names.
+- Push to the user-requested remote or the repository's configured push remote.
+  Do not hard-code or infer a remote from an account name.
+- Check free space before artifact-heavy builds, tests, coverage, or downloads.
+  Re-check before a broad gate when space is tight.
+- Remove only task-owned temporary/build artifacts. Never delete another task's
+  worktree or uncommitted data.
+- At handoff, mention disk or cleanup details only when they affected execution
+  or artifacts/worktrees remain intentionally.
+
+## Change Style
+
+- Preserve existing control flow unless changing it is required for correctness.
+- Prefer a direct local edit over new files, wrappers, managers, or speculative
+  abstractions.
+- Add a helper only when it removes current duplication, names a real domain
+  boundary, or isolates a non-trivial invariant.
+- Remove an in-scope path superseded by the change. If compatibility requires it,
+  adapt at the boundary to one canonical core and use the repository's
+  `RUSTFS_COMPAT_TODO` policy.
+- Comments explain non-obvious invariants or reasons. Do not narrate code or
+  record change history.
+- Mention unrelated problems when useful; do not fix them in a narrow task.
+
+## Reuse and Boundary Rules
+
+- Before adding helpers, constants, fixtures, or wrappers, search the touched
+  crate, the domain-owning crate, `crates/utils`, `crates/common`, and relevant
+  direct dependencies.
+- Reuse requires matching semantics: normalization, error types, deadlines,
+  durability, and compatibility must fit the call site. A narrowly named local
+  helper is better than forced reuse with different semantics.
+- Validate untrusted input at its trust boundary, then trust the validated type.
+  Values crossing disk, RPC, persistence, or version boundaries remain
+  untrusted at every consumer.
+- Re-check boundary values immediately before destructive actions such as
+  delete, overwrite, or quorum decisions.
+- Every new branch needs a concrete triggering input/state. For decoded or peer
+  data, corruption and mixed-version input are valid triggers.
+- Required values must return a typed error when absent or corrupt; do not use a
+  default that converts corruption into a plausible result.
+- Attach error context once where it is actionable. Do not erase typed errors
+  below aggregation or quorum layers.
+
+## Naming
+
+Use Rust API naming: `SCREAMING_SNAKE_CASE` constants/statics, `snake_case`
+functions/variables, and `PascalCase` types. Do not rename unrelated existing
+violations.

.agents/references/pull-requests.md

@@ -0,0 +1,51 @@
+# Git and Pull Request Rules
+
+Applies to commits, pushes, PRs, and issue/discussion actions. Paths below are
+repository-relative. User authorization and root `AGENTS.md` still govern scope.
+
+## Final PR Preflight
+
+Before creating or updating a PR, reuse completed review and verification:
+
+- Verify the actual base (normally `origin/main`) and the complete task diff,
+  including file names and whitespace. Exclude secrets, logs, generated
+  artifacts, and unrelated edits. Retain an existing PR's base unless requested.
+- Confirm the final diff passed the root verification tier; fix task-owned
+  failures and run missing scoped checks. Report unresolved required checks or
+  authority without expanding the task. Do not start another general review.
+- Keep the English Conventional Commit title at most 72 characters. Use the
+  template headings, actual checks, material risks, and rollback notes.
+- Immediately before writing to GitHub, confirm the head and task diff are
+  unchanged. Rerun only checks invalidated by edits or relevant state changes.
+
+## Pull Request Lifecycle
+
+- Creating or updating a PR includes one immediate snapshot of checks,
+  mergeability, reviews, and unresolved threads.
+- Unless the user explicitly requests monitoring, a release workflow requires
+  it, or an automation already owns it, hand off after the PR is open with the
+  current state and next event to watch. Do not delay ordinary handoff with
+  fixed quiet-period sleeps.
+- For requested monitoring, use event-driven or bounded waits. Report only state
+  changes, actionable failures, or a meaningful prolonged delay.
+- Investigate failures/comments before changing code. Fix task-attributable
+  issues, rerun affected verification, push, reply or resolve the thread, then
+  resume the requested monitor.
+- Never merge without required reviewer approval or explicit authority.
+- After an observed merge, verify the commit reached the base, then clean the
+  task worktree/branch when safe. Preserve unmerged work for closed PRs unless
+  deletion was explicitly authorized.
+
+## Git and PR Baseline
+
+- Follow Conventional Commits; keep the subject at most 72 characters.
+- Source comments, commits, PR titles, and PR bodies are in English.
+- Keep every heading from `.github/pull_request_template.md`; use `N/A` where
+  needed and include commands actually run.
+- Use `--body-file` for multiline `gh pr create`/`gh pr edit` content.
+- PR/issue/discussion content must not contain the literal sequence `\n` or
+  hard-wrapped prose paragraphs.
+- Do not include local absolute paths or tool-specific labels/prefixes in GitHub
+  content.
+- Resolve review threads after the underlying issue is fixed. If declining a
+  suggestion, reply with a short evidence-based reason.

.agents/skills/adversarial-validation/SKILL.md

@@ -1,11 +1,11 @@
 ---
 name: adversarial-validation
-description: Review a final RustFS diff adversarially when the user requests adversarial review, the root AGENTS.md classifies the change as high risk, or a substantial PR is being reviewed. Do not use for ordinary questions, diagnosis, planning, status, documentation-only work, or routine low-risk implementation.
+description: Review RustFS diffs or designs for explicit adversarial requests, high-risk changes under the repository review policy, or substantial PR reviews. Skip ordinary questions, diagnosis, planning, status, routine low-risk implementation, and prose with no execution effect.
 ---
 
 # RustFS Adversarial Validation
 
-Use the risk tier and review shape defined in the root `AGENTS.md`. This skill
+Use the [repository risk tiers and review shape](../../references/adversarial-validation.md). This skill
 routes a review to RustFS-specific probes without loading unrelated domains.
 
 ## Select Lenses
@@ -31,15 +31,18 @@ adversarial review.
 
 ## Review Protocol
 
-1. Freeze the exact final diff/head and list the selected lenses.
+1. Freeze the exact final diff/head (or the design under review) and list the
-2. Run the review shape required by root `AGENTS.md`.
+   selected lenses.
+2. Run the review shape required by the repository risk tier.
 3. For each selected lens, either report a concrete finding or a null verdict
    naming the attacks performed.
-4. A finding needs `file:line`, a triggering input/state/interleaving, the wrong
+4. Apply root `AGENTS.md`'s finding standard. Test each candidate against callers,
-   outcome, and a focused fix or missing regression check.
+   existing coverage, and invariants before accepting it; an adversarial role
-5. Fix or rebut every finding with code-path, test, or invariant evidence.
+   does not have to produce a defect.
+5. Fix or rebut supported findings with code-path, test, or invariant evidence.
 6. After a non-trivial edit, rerun only lenses affected by that edit against the
    new exact diff.
 
 Do not turn a null verdict into a long checklist. Record concise evidence that
-the relevant failure classes were attacked.
+the relevant failure classes were attacked, then stop under the root completion
+rule. Keep the required per-lens verdicts for high-risk PRs.

.agents/skills/adversarial-validation/references/test-coverage.md

@@ -3,9 +3,10 @@
 - For every behavior claim, name the focused test/check that fails if the
   changed hunk is reverted. If none is practical, require the reason and
   residual risk.
-- Confirm tests exercise the real production path and assert returned values,
+- Confirm tests exercise the real production path and distinguish the intended
-  exact bytes, stored state, or the specific error variant—not only success,
+  behavior from the named regression. A success, `is_err()`, or no-panic check
-  `is_err()`, or no panic.
+  can be sufficient when that is the actual contract; require exact values,
+  bytes, state, or error variants when those distinctions matter to the change.
 - For new flags/modes, verify each branch and ask which test fails if the branch
   is inverted.
 - For new error propagation, inject the failure and assert the caller observes

.agents/skills/arch-checks/SKILL.md

@@ -1,12 +1,13 @@
 ---
 name: arch-checks
-description: Resolve failures from the repository's architecture guard scripts — check_layer_dependencies.sh, check_architecture_migration_rules.sh, check_unsafe_code_allowances.sh, check_logging_guardrails.sh, check_doc_paths.sh. Use when make pre-commit / pre-pr or CI fails on one of these checks.
+description: Diagnose failures from check_layer_dependencies.sh, check_architecture_migration_rules.sh, check_unsafe_code_allowances.sh, check_logging_guardrails.sh, check_doc_paths.sh, or check_no_planning_docs.sh. Use when one of these guards fails, not for every architecture question or documentation edit.
 ---
 
 # Architecture Guard Checks
 
-All five run in `make pre-commit` / `make pre-pr` and in CI. Fix the cause;
+Read only the section for the failing guard. Use `.config/make/` and the current
-never weaken a check to get green.
+workflow to verify its wiring; not every guard is part of every gate. Fix the
+cause and rerun the failed guard; never weaken a check to get green.
 
 ## `check_layer_dependencies.sh` — layer DAG in `rustfs/src`
 
@@ -54,7 +55,8 @@ Instruction docs (`AGENTS.md`, `CLAUDE.md`, `ARCHITECTURE.md`) and every
 Markdown file under `docs/` (architecture, operations, testing, index) must not
 reference repo file paths that no longer exist. If your refactor moved code,
 update the docs that point at it — the error message lists `doc -> stale-path`
-pairs. Cite paths plus symbol names, never line numbers (see `docs/README.md`).
+pairs. In durable docs, cite paths plus symbol names rather than line numbers
+(see `docs/architecture/README.md`). Review findings still need `file:line`.
 
 ## `check_no_planning_docs.sh`
 

.agents/skills/code-change-verification/SKILL.md

@@ -8,19 +8,11 @@ description: Review a commit, PR, or merged patch when the user requests ordinar
 Use this skill for an ordinary requested review. If the root policy or user calls
 for adversarial validation, use `adversarial-validation` instead of running both.
 
-## Quick Start
-
-1. Read the scope: commit, PR, patch, or file list.
-2. Map each changed area by risk and user impact.
-3. Inspect each risky change in context.
-4. Report findings first, ordered by severity.
-5. Close with residual risks and verification recommendations.
-
 ## Core Workflow
 
 ### 1) Scope and assumptions
-- Confirm change source (diff, commit, PR, files), target branch, language/runtime, and version.
+- Derive the change source, target branch, and relevant runtime/version from the
-- If context is missing, state assumptions before deeper analysis.
+  supplied diff and metadata. Ask only when missing context could change the verdict.
 - Focus only on requested scope; avoid reviewing unrelated files.
 
 ### 2) Risk map
@@ -40,43 +32,20 @@ for adversarial validation, use `adversarial-validation` instead of running both
   - unchecked assumptions and null/empty/error-path handling
   - stale tests, fixtures, and configs
   - hidden coupling to shared helpers/constants/features
-- If a point is uncertain, mark it as an open question instead of guessing.
+- Apply root `AGENTS.md`'s finding standard: try to disprove a candidate before
+  reporting it. Mention an unresolved question only when it could materially
+  change the verdict; do not fill the report with speculative possibilities.
 
-#### Rust-specific checks (apply to all Rust changes)
+#### Rust-specific checks
 
-Run the full checklist in [rust-code-quality](../rust-code-quality/SKILL.md) — the canonical Rust review checklist for the unwrap/casting/cloning/locking/recursion/error-type/serde/test rules and the reuse-and-necessity checks (duplicated helpers, defensive branches without a nameable trigger, redundant error wrapping). Do not restate those rules here; carry its P0–P3 ratings over unchanged and use this skill's output format.
+For changed Rust behavior, use the matching sections of [rust-code-quality](../rust-code-quality/SKILL.md). Reuse checks already performed by the selected review workflow. Comment-only or formatting-only Rust diffs do not require the full Rust checklist. Carry its P0–P3 ratings over unchanged and use this skill's output format.
 
 ### 4) Findings-first output
-- Order findings by severity:
+- Order supported findings by P0–P3 severity; preserve the Rust ratings above.
-  - P0: critical failure, security breach, or data loss risk
+  Include `path:line`, the failure and impact, a focused fix, and its validation.
-  - P1: high-impact regression
+- If no supported issues remain, state `No findings` with the reviewed scope and
-  - P2: medium risk correctness gap
+  any material verification limitation. Do not append optional improvements to
-  - P3: low risk/quality debt
+  make a clean review look productive.
-- For each finding include:
-  - Severity
-  - `path:line` reference
-  - concise issue statement
-  - impact and likely failure mode
-  - specific fix or mitigation
-  - validation step to confirm
-- If no issues exist, explicitly state `No findings` and why.
 
-### 5) Close
+Close after the required review. Recommend additional verification only for an
-- Report assumptions and unknowns.
+identified unresolved risk or required gate; reuse evidence for unchanged code.
-- Suggest targeted checks (tests, canary checks, logs/metrics, migration validation).
-
-## Output Template
-
-1. Findings
-2. No findings (if applicable)
-3. Assumptions / Unknowns
-4. Recommended verification steps
-
-## Finding Template
-
-- `[P1] Missing timeout for downstream call`
-  - Location: `path/to/file.rs:123`
-  - Issue: ...
-  - Impact: ...
-  - Fix suggestion: ...
-  - Validation: ...

.agents/skills/issue-triage/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: issue-triage
-description: Triage a GitHub issue — determine if it is already fixed, needs implementation, or should be closed. Searches related commits and PRs, verifies implementation status, and posts a triage comment or closes the issue. Use when the user provides an issue URL and asks whether it can be closed or needs work.
+description: Assess whether a GitHub issue is fixed, needs implementation, or can be closed by checking related work and current code. Use for issue completion/triage requests. Status questions are read-only; comment, close, or change labels only when the conversation authorizes that action.
 ---
 
 # Issue Triage
@@ -20,6 +20,8 @@ Read the issue body to understand what was requested. Extract:
 - Any linked PRs or commits mentioned in the body or comments.
 - Any checklist items or sub-issues.
 
+Resolve the issue repository and implementation repository separately (for example, `rustfs/backlog` tracks work in `rustfs/rustfs`). Pass the implementation repository explicitly to PR queries; the current checkout may belong to another repository.
+
 ### 2. Search for related work
 
 Search git history for commits referencing the issue:
@@ -29,26 +31,29 @@ git log --oneline --all --grep="<N>" | head -30
 
 Search for related PRs:
 ```bash
-gh pr list --search "fixes #<N> OR closes #<N> OR #<N>" --state all --json number,title,state,mergedAt
+gh pr list --repo <implementation-repo> --search "<issue-url>" --state all --json number,title,state,mergedAt
 ```
 
+Also search qualified issue references and subject keywords; for same-repository
+issues, include `#<N>`. Follow explicit links even without a text match. A search
+page with no match does not prove the work is absent.
+
 If the issue mentions specific PRs, check their status:
 ```bash
-gh pr view <PR_N> --json state,mergedAt,title
+gh pr view <PR_N> --repo <implementation-repo> --json state,mergedAt,title,mergeCommit,baseRefName
 ```
 
 ### 3. Verify implementation
 
-For each linked or related PR that is merged, verify the fix is actually present on the current main branch:
+Fetch the implementation repository's current base branch. For each merged candidate, verify its merge commit is present and inspect the current code for the claimed behavior; a commit message match alone is not proof:
 ```bash
-git log --oneline main | grep -i "<keyword>"
+git fetch <implementation-remote> <base-branch>
-# or
+git merge-base --is-ancestor <merge-commit> <implementation-remote>/<base-branch>
-git log --oneline main --grep="<PR_N>"
 ```
 
-If the issue describes a specific defect, check the relevant code to confirm the fix is in place:
+If the issue describes a specific defect, inspect the fetched base's code rather than assuming the current checkout contains it:
 ```bash
-grep -n "<pattern>" crates/<relevant>/src/<file>.rs
+git show <implementation-remote>/<base-branch>:crates/<relevant>/src/<file>.rs
 ```
 
 For issues with checklists, verify each item individually. If sub-items are tracked as separate issues, check those too:
@@ -58,13 +63,15 @@ gh issue view <SUB_N> --repo <owner/repo> --json state
 
 ### 4. Determine verdict
 
-- **All items fixed and merged**: Close with a summary comment listing what was fixed and which PRs.
+- **All items fixed and merged**: Recommend closing; name the verified PRs and behavior.
-- **Some items fixed, some remaining**: Comment with status of each item. Do not close.
+- **Some items fixed, some remaining**: Keep open; report each remaining item.
-- **Not yet implemented**: Comment with a summary of what remains. Do not close.
+- **Not yet implemented**: Keep open; report what remains.
-- **Superseded or no longer relevant**: Close with explanation.
+- **Superseded or no longer relevant**: Recommend closing with evidence.
 
 ### 5. Take action
 
+For a status-only request, return the assessment without GitHub writes. If commenting, closing, or label edits are authorized, perform only those actions; do not ask again for authority already given. Prepare the final assessment before asking for any missing authority. Write `rustfs/backlog` issue content in Chinese.
+
 Close with comment:
 ```bash
 gh issue close <N> --repo <owner/repo> --comment "<body>"
@@ -75,9 +82,9 @@ Comment without closing:
 gh issue comment <N> --repo <owner/repo> --body-file /tmp/triage.md
 ```
 
-Update issue labels if needed:
+Update labels only when label changes are authorized, using existing repository labels; never add tool-specific labels:
 ```bash
-gh issue edit <N> --repo <owner/repo> --add-label "completed" --remove-label "needs-triage"
+gh issue edit <N> --repo <owner/repo> --add-label "<existing-label>"
 ```
 
 Always use `--body-file` for multiline content, never inline `--body`.
@@ -85,27 +92,16 @@ Always use `--body-file` for multiline content, never inline `--body`.
 ### 6. Handle multi-issue batches
 
 When the user asks to check multiple issues (e.g., "check all issues by user X" or "scan backlog for closable issues"):
-1. List the issues: `gh issue list --repo <repo> --author <user> --state open --json number,title,updatedAt`
+1. List the full requested scope with pagination (for example `gh api --paginate 'repos/<repo>/issues?state=open&per_page=100'`, excluding entries with `pull_request`). Add an author filter only when the user requested one; the default page/limit is not evidence that all issues were checked.
 2. For each issue, run steps 1-5 above.
 3. Report a summary table of all triaged issues with verdicts.
 
-## Output format
+## Report
 
-### Issue Triage: #<N> — <title>
+Identify the issue and current state, verified implementation/PR evidence,
-
+remaining items, verdict, and action actually taken. Use a table for batches;
-**State**: OPEN / CLOSED
+a single issue does not require a heading for each field. Follow step 4's
-**Linked PRs**: <list with merge status>
+verdicts without repeating the assessment in another template.
-
-#### Assessment
-<what was requested vs what is implemented>
-
-#### Verdict
-- Close — all items resolved by <PR list>
-- Keep open — <remaining items>
-- Not started — <what needs to be done>
-
-#### Action taken
-- Closed with comment / Commented / No action
 
 ## Notes
 

.agents/skills/plugin-contract-guard/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: plugin-contract-guard
-description: Invariants and change procedure for the target-plugin / extension system — plugin manifests, admin plugin/extension catalog and instance APIs, secret redaction, external-plugin install policy. Use when editing crates/targets (manifest, plugin, control_plane, catalog, runtime), crates/extension-schema, or rustfs/src/admin plugin_contract.rs / plugins_*.rs / extensions.rs / target_descriptor.rs.
+description: Guard changes to target-plugin manifests, extension schemas, admin catalog/instance contracts, secret redaction, and external-plugin install policy. Use when a diff changes those contracts in crates/targets, crates/extension-schema, or admin plugin/extension handlers; path membership alone, comments, and unrelated runtime internals do not trigger it.
 ---
 
 # Plugin & Extension Contract Guard

.agents/skills/pr-creation-checker/SKILL.md

@@ -1,46 +0,0 @@
----
-name: pr-creation-checker
-description: Perform the final RustFS PR preflight and draft compliant English title/body metadata immediately before creating or updating a PR. Do not use during implementation or as a second general code review.
----
-
-# PR Creation Checker
-
-Use this skill only at the PR boundary. Reuse completed diff review and
-verification evidence; do not reread the repository or rerun equivalent checks.
-
-## Preflight
-
-1. Confirm the branch is based on current `origin/main` and contains only the
-   intended task diff.
-2. Inspect `git diff --stat`, `git diff --check`, and changed file names for
-   secrets, logs, generated artifacts, or unrelated edits.
-3. Confirm the checks selected by root `AGENTS.md` passed on the final diff.
-   Do not replace focused behavioral tests with a generic gate or rerun checks
-   already covered by an unchanged umbrella run.
-4. Read `.github/pull_request_template.md`. Consult `Makefile`, `.config/make/`,
-   or CI only when the required command/current gate is uncertain.
-5. Return `BLOCKED` for an unclean scope, missing required evidence, failed
-   required checks, or non-compliant metadata.
-
-## Metadata
-
-- Title: English Conventional Commit, at most 72 characters, with no tool
-  prefix.
-- Body: English, exact template headings, `N/A` where needed, concise rationale,
-  actual verification commands, and material risks/rollback notes.
-- Use repository-relative paths; never include local absolute paths.
-- Keep prose paragraphs on one logical line and never include the literal
-  sequence `\n`.
-- Use a temporary body file with `gh pr create --body-file` or
-  `gh pr edit --body-file`; never pass multiline Markdown inline.
-
-## Output
-
-- Status: `READY` or `BLOCKED`.
-- Title.
-- Complete PR body.
-- Verification commands and results.
-- Risks or `N/A`.
-
-Immediately before the GitHub write, repeat only the five preflight checks above
-against the final head.

.agents/skills/pr-creation-checker/agents/openai.yaml

@@ -1,4 +0,0 @@
-interface:
-  display_name: "PR Creation Checker"
-  short_description: "Draft RustFS-ready PRs with checks, template, and blockers."
-  default_prompt: "Use $pr-creation-checker for final PR preflight and compliant English title/body metadata."

.agents/skills/pr-review/SKILL.md

@@ -1,147 +1,93 @@
 ---
 name: pr-review
-description: Review a GitHub PR end-to-end from a URL or number — fetch metadata, inspect the diff, run multi-role adversarial review, check CI status, and post the review comment. Use when the user provides a PR link and asks to review it.
+description: Review a GitHub PR from a URL or number using its actual base/head and risk-appropriate code review. Use when the user asks for a PR review, not a status lookup or PR wording edit. Publish a review only when authorized; delegation and monitoring follow the requested scope and root AGENTS.md.
 ---
 
 # PR Review
 
-Use this skill when the user provides a GitHub PR URL or number and asks to review it. This covers the full review lifecycle: data gathering, code review, CI verification, and posting the result.
+Use this skill for PR context and review delivery. An ordinary review request is read-only unless the conversation also authorizes posting or fixes. Reuse that authorization without asking again; prepare the review before requesting any missing publication approval.
 
 ## Prerequisites
 
-- Read `AGENTS.md` for the repository's adversarial validation policy and change-style rules.
+- Follow root `AGENTS.md`; classify risk with the [review policy](../../references/adversarial-validation.md) and consult relevant [change-style and boundary rules](../../references/implementation.md).
-- The `adversarial-validation` skill handles the review role playbooks; this skill orchestrates the workflow around it.
+- Select `code-change-verification` for ordinary review or `adversarial-validation` for explicitly adversarial, substantial, or high-risk review; do not run both on the same diff.
 
 ## Workflow
 
 ### 1. Gather PR context
 
 ```bash
-gh pr view <N> --json title,author,state,body,additions,deletions,changedFiles,commits,baseRefName,headRefName
+gh pr view <N> --repo <owner/repo> --json title,author,state,body,additions,deletions,changedFiles,commits,baseRefName,headRefName,baseRefOid,headRefOid
-gh pr diff <N> --name-only
+gh pr diff <N> --repo <owner/repo> --name-only
 ```
 
 Read the PR body and linked issues to understand the change's purpose. If the PR references an issue, fetch that too:
 ```bash
-gh issue view <ISSUE> --json title,body,state
+gh issue view <ISSUE> --repo <issue-owner/repo> --json title,body,state
 ```
 
 ### 2. Fetch the diff and classify the change
 
 ```bash
-git fetch origin pull/<N>/head:pr-<N>
+git fetch <repo-remote> <baseRefName> refs/pull/<N>/head
-git diff main...pr-<N> --stat
+git diff <baseRefOid>...<headRefOid> --stat
 ```
 
-Classify the change by risk tier (per AGENTS.md):
+Resolve `<repo-remote>` to the PR repository; do not assume the current checkout's `origin` or `main` matches. Record the exact base/head used. If either moved during fetching, refresh the snapshot before reviewing. Classify using the repository review policy; instruction changes that affect agent execution are mechanical, not exempt.
-- **Exempt**: docs/comments/instruction-only, formatting, typos.
-- **Mechanical**: renames, file moves, test-only or tooling changes.
-- **Standard** (default): any behavior change.
-- **High risk**: locking, erasure coding, quorum/heal, replication, multipart, RPC, lifecycle/tiering, metadata formats, persistence/fsync, IAM/KMS/auth, on-disk/on-wire formats, S3 API-visible behavior.
 
-### 3. Cluster changed files and delegate review
+### 3. Review the changed behavior
 
-Group the changed files into logical clusters (by crate or functional area). For each cluster, spawn a subagent with a focused review prompt that includes:
+Group files by functional area to trace callers and invariants. Use the root risk tier's review shape and only matching lenses. File count does not authorize delegation. When delegation is explicitly authorized, high-risk/substantial reviews use exactly two independent reviewers with the applicable lenses split between them; otherwise use two fresh sequential passes. Reviewers do not spawn further agents.
-- The cluster's changed files and their diffs.
-- The applicable adversarial role probes (from the `adversarial-validation` skill).
-- The repository's AGENTS.md rules relevant to that domain.
 
-For standard-tier changes: correctness adversary + simplicity adversary + test-coverage skeptic, plus every role whose domain the diff touches.
+Findings need a concrete failure scenario with `file:line`; a null verdict briefly names the relevant probes. Reuse existing evidence and choose local checks from the final diff under the root verification policy.
-For high-risk changes: run all seven roles.
-
-Each subagent must produce findings (concrete failure scenario with file:line) or a null report ("attacked X, Y, Z — no break found").
 
 ### 4. Check CI status
 
 ```bash
-gh pr checks <N>
+gh pr checks <N> --repo <owner/repo>
 ```
 
-If any checks fail, investigate:
+Investigate a failed check when it bears on a finding or the user requested CI diagnosis/merge readiness:
 ```bash
-gh run view --log-failed --job=<JOB_ID>
+gh run view --repo <owner/repo> --log-failed --job=<JOB_ID>
 ```
 
-Determine whether failures are pre-existing (on main), flaky, or caused by the PR.
+Use current evidence to distinguish pre-existing, flaky, and PR-caused failures. Do not classify them by guesswork or turn a code-only review into unrelated CI repair.
 
 ### 5. Synthesize findings
 
-Combine all subagent findings into a structured review:
+Report the PR, reviewed base/head, and risk tier, then summarize the assessment.
-- **Summary**: one-paragraph overview of the change and overall assessment.
+Use the selected review's P0–P3 ratings and root finding standard: supported
-- **Findings**: each finding with severity (critical/major/minor/nit), file:line, concrete failure scenario, and suggested fix.
+findings with `file:line`, failure scenario, and fix, or `No findings`.
-- **CI status**: pass/fail with notes on any failures.
+State the observed check status, including pending or unavailable checks, and
-- **Verdict**: APPROVE, REQUEST_CHANGES, or COMMENT.
+the verdict (`APPROVE`, `REQUEST_CHANGES`, or `COMMENT`). Do not infer a pass
+from missing checks or add style nits to populate a clean review.
 
 ### 6. Post the review
 
-Write the review body to a temp file and post via CLI:
+Only when posting is authorized, write the review body to a temp file and post via CLI. Refresh the PR head first; if it changed, review the delta and update the verdict before posting:
 ```bash
 # Request changes
-gh pr review <N> --request-changes --body-file /tmp/pr_review.md
+gh pr review <N> --repo <owner/repo> --request-changes --body-file /tmp/pr_review.md
 
 # Approve
-gh pr review <N> --approve --body-file /tmp/pr_review.md
+gh pr review <N> --repo <owner/repo> --approve --body-file /tmp/pr_review.md
 
 # Comment only (no verdict)
-gh pr review <N> --comment --body-file /tmp/pr_review.md
+gh pr review <N> --repo <owner/repo> --comment --body-file /tmp/pr_review.md
 ```
 
-For inline comments on specific lines, use the GitHub API:
+For authorized inline comments, use [the submission example](references/posting.md).
-```bash
-cat > /tmp/pr_review.json <<'EOF'
-{
-  "body": "review body",
-  "event": "REQUEST_CHANGES",
-  "comments": [
-    {
-      "path": "crates/foo/src/bar.rs",
-      "line": 42,
-      "body": "finding description"
-    }
-  ]
-}
-EOF
-gh api --method POST /repos/{owner}/{repo}/pulls/<N>/reviews --input /tmp/pr_review.json
-```
 
 Always use `--body-file` or `--input`, never inline multiline `--body`.
 
 ### 7. Handle follow-up
 
-If the review requests changes:
+Follow the [PR lifecycle](../../references/pull-requests.md) and any explicit monitoring request. For follow-up, fetch the new head and compare the recorded reviewed SHA with the new SHA; revisit affected callers and findings. Never use an unfetched `origin/pull/<N>/head` ref as evidence. Update the posted review or resolve addressed threads only within existing authorization.
-- Monitor for new commits: `gh pr view <N> --json commits`
-- Re-review changed files only: `git diff pr-<N>..origin/pull/<N>/head`
-- Update the review when findings are addressed.
-
-If CI was failing due to pre-existing main breakage:
-- Comment on the PR noting the failure is pre-existing.
-- Suggest updating the branch: `gh pr update-branch <N>`
-
-## Output format
-
-### PR Review: #<N> — <title>
-
-**Author**: <author>
-**Risk tier**: exempt | mechanical | standard | high-risk
-**Changed files**: <count> across <cluster count> clusters
-
-#### Summary
-<one-paragraph overview>
-
-#### Findings
-| Severity | Location | Finding |
-|----------|----------|---------|
-| critical | file:line | concrete failure scenario |
-
-#### CI Status
-- All checks pass / Failing: <details>
-
-#### Verdict
-APPROVE / REQUEST_CHANGES / COMMENT
 
 ## Notes
 
 - The user may ask for review in Chinese; respond in the same language but keep the review body in English per AGENTS.md rules.
 - When the user asks for "多角色对抗 review", run the full adversarial validation protocol — this skill's step 3 covers that.
 - If the PR is from a fork, check `maintainerCanModify` before attempting to push fixes.
-- For very large PRs (>50 files), cluster aggressively and delegate in parallel to keep review time reasonable.
+- For very large PRs, batch the review by functional area while keeping the same bounded review shape.

.agents/skills/pr-review/references/posting.md

@@ -0,0 +1,22 @@
+# Inline PR Review Submission
+
+Read only when an inline review is authorized. Recheck the PR head before posting and bind the review to the reviewed commit.
+
+For inline comments on specific lines, use the GitHub API:
+```bash
+cat > /tmp/pr_review.json <<'EOF'
+{
+  "commit_id": "<reviewed-head-sha>",
+  "body": "review body",
+  "event": "REQUEST_CHANGES",
+  "comments": [
+    {
+      "path": "crates/foo/src/bar.rs",
+      "line": 42,
+      "body": "finding description"
+    }
+  ]
+}
+EOF
+gh api --method POST /repos/{owner}/{repo}/pulls/<N>/reviews --input /tmp/pr_review.json
+```

.agents/skills/rust-code-quality/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: rust-code-quality
-description: Run a focused Rust quality review when the user requests one, when reviewing a Rust PR/commit, or when another selected review workflow delegates Rust-specific checks. Do not auto-load for every implementation edit.
+description: Run a focused Rust quality review when the user requests one or a selected review workflow needs Rust-specific checks for changed behavior. Do not auto-load for every implementation edit, comment-only or formatting-only Rust diff, or repeat an already completed review.
 ---
 
 # Rust Code Quality Gate
@@ -8,12 +8,18 @@ description: Run a focused Rust quality review when the user requests one, when
 Use this skill for a dedicated Rust review to cover rules that `cargo clippy`
 does not catch.
 
+Search matches and checklist items are candidates, not findings. Apply the root
+finding standard; distinguish a demonstrated bug, an explicit rule violation,
+and an optional preference. P2/P3 suggestions do not need to be invented or
+included in an otherwise clean correctness review.
+
 ## Quick Start
 
 1. Identify changed `.rs` files.
-2. Run automated checks on changed files.
+2. Run the matching candidate searches on changed files.
-3. Run manual review checklist on the diff.
+3. Apply the manual checklist sections whose behavior the diff touches.
-4. Resolve or rebut every finding with evidence; P0/P1 findings cannot be deferred.
+4. Report or rebut every finding with evidence; P0/P1 findings block approval.
+   Fix them when implementation is authorized; a read-only review reports them.
 
 ## Automated Checks
 
@@ -35,7 +41,7 @@ rg -n 'Result<.*String>' <changed-files>
 rg -n 'Box<dyn.*Error' <changed-files>
 
 # 5. println/eprintln in production
-rg -n 'println!\|eprintln!' <changed-files>
+rg -n 'println!|eprintln!' <changed-files>
 
 # 6. Ordering::Relaxed usage (verify each is intentional)
 rg -n 'Ordering::Relaxed' <changed-files>
@@ -80,7 +86,7 @@ For the Rust diff under review, verify:
 - [ ] Test volume and line count are never treated as production-code growth
 
 ### Serde
-- [ ] Structs from untrusted input have `#[serde(deny_unknown_fields)]`
+- [ ] Structs from untrusted input reject unknown fields where the compatibility contract permits; otherwise validate security-critical fields explicitly and test the supported input shape
 - [ ] `#[serde(default)]` not used on security-critical fields without validation
 
 ### Code Hygiene
@@ -104,20 +110,7 @@ For the Rust diff under review, verify:
 
 ## Output Template
 
-```
+Use the calling review's output format. For a standalone review, report supported
-## Rust Code Quality Report
+findings with severity, location, impact, fix, and validation, or `No findings`.
-
+Include only material unverified checks. Candidate counts are not a quality
-### Automated Scan
+metric and do not need a separate scan report.
-- unwrap/expect candidates inspected: N
-- numeric-cast candidates inspected: N
-- error-type candidates inspected: N
-- output-macro candidates inspected: N
-
-### Findings
-- [P1] `path:line` — description
-  - Fix: ...
-  - Validation: ...
-
-### Verdict
-PASS / BLOCKED (list blocking findings)
-```

.agents/skills/rustfs-release-publish/SKILL.md

@@ -4,9 +4,14 @@ description: "Run the end-to-end RustFS console gate, version bump, preview vali
 ---
 # RustFS Release Publish (preview-validated pipeline)
 
-This skill orchestrates a full release. It wraps `rustfs-release-version-bump` (which only edits version files and opens the PR) with a mandatory preview-tag validation loop before the final tag is published.
+This skill orchestrates a full release. It wraps `rustfs-release-version-bump` (invoked here with the authorized commit/push/PR scope) with a mandatory preview-tag validation loop before the final tag is published.
 
-Core design: **version files never carry a `-preview.N` suffix**. The preview suffix exists only in tag names. A preview tag creates a visible GitHub Release marked Prerelease and uploads versioned assets, but it never becomes GitHub Latest and never updates `*-latest`, `latest.json`, R2, Docker, or Helm channels. That Release is temporary: `build.yml` deletes it automatically once the final tag's Release is published, so the Releases page ends up carrying deliverables only while the `-preview.N` tags stay behind as the traceability record. This works because the binary self-reports the git tag it was built from (`build::TAG` via shadow_rs, see `rustfs/src/config/cli.rs` `SHORT_VERSION`), and `build.yml` derives artifact names and preview classification from the tag name — Cargo.toml's version is only a no-tag fallback. Therefore the preview tag and the final tag can (and MUST) point at the exact same commit: what you validated is byte-for-byte the source that ships.
+The binary reports its build tag (`build::TAG` via shadow_rs; `SHORT_VERSION` in
+`rustfs/src/config/cli.rs`), and `build.yml` derives asset names and preview
+classification from that tag. Cargo.toml supplies only the no-tag fallback.
+Preview and final tags must therefore share the validated source commit;
+their tag-dependent version and asset names differ. The channel and cleanup
+constraints are defined once under Preview tag naming and Hard rules below.
 
 Pipeline shape:
 
@@ -29,9 +34,9 @@ On validation failure: fix lands on main via normal PR (version files are alread
 - Final target version, for example `1.0.0-beta.10`.
 - Preview iteration `N` (default: next unused preview tag for that target; check with `git tag -l '<target>-preview.*'` after `git fetch --tags`).
 
-If the target version is missing or ambiguous, stop and ask before doing anything (see the semver gate below).
+If the target version is missing or ambiguous, collect the current release/tag baseline and ask before version edits or publication. Continue independent read-only preflight while the answer is pending (see the semver gate below).
 
-## Semver gate — confirm the target version before touching anything
+## Semver gate — resolve the target before version edits or publication
 
 Versions follow [SemVer 2.0.0](https://semver.org/). Precedence reminder:
 
@@ -43,7 +48,7 @@ Numeric prerelease identifiers compare numerically (`beta.9 < beta.10`), not lex
 
 Rules:
 
-- A request like "发个版" / "release the next version" without an exact version string is ALWAYS ambiguous. Derive the current latest tag (`git tag --sort=-v:refname | head`), then ask the user to choose via AskUserQuestion with concrete candidates, e.g. from `1.0.0-beta.10`: next prerelease `1.0.0-beta.11`, promote to `1.0.0-rc.1`, promote to stable `1.0.0`. Never guess between these — they have very different meanings (channel promotion vs. iteration) and different CI classification consequences.
+- A request like "发个版" / "release the next version" without an exact version string is ALWAYS ambiguous. Derive the current latest tag (`git tag --sort=-v:refname | head`), then ask the user to choose with concrete candidates, e.g. from `1.0.0-beta.10`: next prerelease `1.0.0-beta.11`, promote to `1.0.0-rc.1`, promote to stable `1.0.0`. Never guess between these — they have very different meanings (channel promotion vs. iteration) and different CI classification consequences.
 - After a stable `X.Y.Z` exists, the next version must state which component bumps: patch `X.Y.(Z+1)` for fixes only, minor `X.(Y+1).0` for backward-compatible features, major `(X+1).0.0` for breaking changes. If the user names a bump type but not a number, compute it from the latest stable tag and echo the exact resulting version back for confirmation.
 - Echo the final confirmed version string verbatim in your first status report; every later phase must use exactly that string. If at any point the user's wording and the confirmed version diverge, stop and re-confirm.
 
@@ -77,58 +82,7 @@ Rules:
 
 ### Console release gate
 
-Complete this gate before changing any RustFS version file or creating any RustFS tag. RustFS `build.yml` downloads the asset returned by `repos/rustfs/console/releases/latest`, so a successful Console build alone is insufficient.
+Read and complete [the Console gate](references/console-gate.md) before Phase 1. Verify the latest published Console asset and exact commit; if Console main is ahead, complete its release and asset verification first. A successful build alone does not satisfy this gate.
-
-1. Read the latest published Console tag and compare it with Console `main`:
-
-```bash
-CONSOLE_REPO="rustfs/console"
-CONSOLE_LATEST=$(gh api "repos/${CONSOLE_REPO}/releases/latest" --jq .tag_name)
-gh api "repos/${CONSOLE_REPO}/compare/${CONSOLE_LATEST}...main" \
-  --jq '{status, ahead_by, behind_by, commits: [.commits[] | {sha, message: .commit.message}]}'
-```
-
-- `ahead_by == 0`: no merged Console change is waiting for release. Still verify the current latest asset using step 4, then continue to Phase 1.
-- `ahead_by > 0` and `behind_by == 0`: publish Console before continuing. Report the merged commits and select the next unused `vX.Y.Z` tag. Default to the next patch version when the changes are fixes or backward-compatible UI work; stop for confirmation if a minor/major bump is plausible.
-- Any diverged history or `behind_by > 0`: stop and resolve the Console release baseline explicitly. Do not guess a range or publish RustFS.
-
-2. Clone/fetch `rustfs/console` into a scratch directory and record its exact `main` commit. Before creating a tag, check for a `v*` tag or Release workflow already associated with that hash. If one is in progress, wait for it instead of creating another version:
-
-```bash
-CONSOLE_SCRATCH=$(mktemp -d)
-gh repo clone "$CONSOLE_REPO" "$CONSOLE_SCRATCH/console"
-git -C "$CONSOLE_SCRATCH/console" fetch origin main --tags
-CONSOLE_HASH=$(git -C "$CONSOLE_SCRATCH/console" rev-parse origin/main)
-git -C "$CONSOLE_SCRATCH/console" tag --points-at "$CONSOLE_HASH" 'v*'
-gh run list -R "$CONSOLE_REPO" --workflow release.yml --commit "$CONSOLE_HASH" --limit 5
-```
-
-If no release exists or is running for `CONSOLE_HASH`, create the selected annotated tag at that exact hash and push it:
-
-```bash
-git -C "$CONSOLE_SCRATCH/console" tag -a "<console-tag>" -m "Release <console-tag>" "$CONSOLE_HASH"
-git -C "$CONSOLE_SCRATCH/console" push origin "<console-tag>"
-```
-
-Console tags include the `v` prefix. Pushing the tag triggers `.github/workflows/release.yml` (`🚀 Release`). Remove `CONSOLE_SCRATCH` after the gate completes.
-
-3. Find the exact tag run and wait for completion:
-
-```bash
-gh run list -R "$CONSOLE_REPO" --workflow release.yml --branch "<console-tag>" --limit 1
-gh run watch -R "$CONSOLE_REPO" "<console-run-id>" --exit-status
-```
-
-4. Block until the published Release is non-draft, the latest endpoint returns the expected tag, and `rustfs-console-<console-tag>.zip` is uploaded, non-empty, and carries a `sha256:` digest:
-
-```bash
-gh release view -R "$CONSOLE_REPO" "<console-tag>" --json isDraft,isPrerelease,assets,url
-test "$(gh api "repos/${CONSOLE_REPO}/releases/latest" --jq .tag_name)" = "<console-tag>"
-test "$(gh api "repos/${CONSOLE_REPO}/releases/tags/<console-tag>" \
-  --jq '[.assets[] | select(.name == "rustfs-console-<console-tag>.zip" and .state == "uploaded" and .size > 0 and (.digest | startswith("sha256:")))] | length')" -eq 1
-```
-
-Treat a missing/mismatched asset, digest, latest tag, or failed/cancelled workflow as BLOCKED. Do not start Phase 1 until the Console gate passes. Record `CONSOLE_TAG`, `CONSOLE_HASH`, Console run URL, and Release URL for the final report.
 
 ## Phase 1 — Version bump to the final target (once)
 
@@ -164,54 +118,9 @@ On a restart (N+1), refresh `PREVIEW_HASH=$(git rev-parse origin/main)` first
 - Record `PREVIOUS_DELIVERABLE`, selected from published Releases by `publishedAt` after excluding the current tag and every `-preview.N` tag. Verify `gh release view "<preview-tag>" --json body --jq .body` contains `## What's Changed` and, when `PREVIOUS_DELIVERABLE` exists, `**Full Changelog**: https://github.com/rustfs/rustfs/compare/<PREVIOUS_DELIVERABLE>...<preview-tag>`. For a repository with no previous deliverable, verify a Full Changelog link exists and record the GitHub baseline fallback.
 - Confirm preview-triggered Docker and Helm jobs are skipped. Preview validation covers the built RustFS binaries, embedded console, and rc compatibility; Docker image construction and Helm publication are deferred to the final tag because the Dockerfiles consume GitHub Release assets.
 
-## Phase 4 — Run the artifact locally, verify the console
+## Phases 4–5 — Local artifact, Console, and rc acceptance
 
-Work inside the session scratchpad directory; never leave stray data dirs.
+Read and complete [preview acceptance](references/preview-acceptance.md): verify the downloaded binary's tag/SHA and readiness, exercise Console CRUD with byte-identical download, and pass the full latest-rc command matrix. Any failure blocks final publication. Retain the results for the confirmation gate below.
-
-```bash
-gh release download "<preview-tag>" -p "rustfs-macos-aarch64-v<preview-tag>.zip" -D "$SCRATCH"
-cd "$SCRATCH" && unzip -o rustfs-*.zip
-./rustfs --version        # must report the PREVIEW TAG (build::TAG), not the Cargo.toml version, plus expected short SHA
-mkdir -p data
-RUSTFS_ACCESS_KEY=rustfsadmin RUSTFS_SECRET_KEY=rustfsadmin ./rustfs ./data
-```
-
-Defaults: S3 endpoint `:9000`, embedded console `:9001`.
-
-Checks (all must pass):
-
-- `./rustfs --version` reports the preview tag name and the short SHA of `PREVIEW_HASH`. Reporting `<target>` without the `-preview.N` suffix means the build did not embed the tag — treat as FAIL and investigate before proceeding.
-- `curl -fsS http://localhost:9000/health/ready` returns ready.
-- Startup log shows the embedded console being served (this was the regression that `fix(release): require embedded console assets` guards).
-- Open `http://localhost:9001` in the browser: login with `rustfsadmin`/`rustfsadmin`; dashboard renders without JS console errors; create a bucket, upload a file, download it back (byte-identical), delete the object and bucket. Keep the server running for Phase 5.
-
-## Phase 5 — Validate with the latest rc client
-
-`rc` is the RustFS CLI client from <https://github.com/rustfs/cli>.
-
-- Ensure the latest release is installed: compare `rc --version` against `gh api repos/rustfs/cli/releases/latest --jq .tag_name`; update via `brew upgrade rustfs/tap/rc` (or download the release binary).
-- Point it at the preview server and run the command matrix, recording PASS/FAIL per command:
-
-```bash
-rc alias set preview http://localhost:9000 rustfsadmin rustfsadmin
-rc ls preview/
-rc mb preview/rel-check
-rc cp <local-file> preview/rel-check/
-rc stat preview/rel-check/<file>
-rc cat preview/rel-check/<file>          # matches source
-rc cp preview/rel-check/<file> ./out && cmp <local-file> ./out
-rc cp -r <local-dir>/ preview/rel-check/dir/
-rc find preview/rel-check --name "*"
-rc share download preview/rel-check/<file> --expire 1h   # presigned URL fetchable via curl
-rc rm preview/rel-check/<file> && rc rm -r --force preview/rel-check/dir
-rc rb preview/rel-check
-rc admin user list preview/
-rc admin user add preview/ relcheckuser relchecksecret12
-rc admin user remove preview/ relcheckuser
-rc alias remove preview
-```
-
-- Any FAIL blocks the release. Afterwards stop the server and delete the scratch data directory.
 
 ### Manual confirmation gate
 

.agents/skills/rustfs-release-publish/references/console-gate.md

@@ -0,0 +1,58 @@
+# Console Release Gate
+
+Read during Phase 0, before changing RustFS version files or tags. Follow the parent skill's release scope and authorization rules.
+
+### Console release gate
+
+Complete this gate before changing any RustFS version file or creating any RustFS tag. RustFS `build.yml` downloads the asset returned by `repos/rustfs/console/releases/latest`, so a successful Console build alone is insufficient.
+
+1. Read the latest published Console tag and compare it with Console `main`:
+
+```bash
+CONSOLE_REPO="rustfs/console"
+CONSOLE_LATEST=$(gh api "repos/${CONSOLE_REPO}/releases/latest" --jq .tag_name)
+gh api "repos/${CONSOLE_REPO}/compare/${CONSOLE_LATEST}...main" \
+  --jq '{status, ahead_by, behind_by, commits: [.commits[] | {sha, message: .commit.message}]}'
+```
+
+- `ahead_by == 0`: no merged Console change is waiting for release. Still verify the current latest asset using step 4, then continue to Phase 1.
+- `ahead_by > 0` and `behind_by == 0`: publish Console before continuing. Report the merged commits and select the next unused `vX.Y.Z` tag. Default to the next patch version when the changes are fixes or backward-compatible UI work; stop for confirmation if a minor/major bump is plausible.
+- Any diverged history or `behind_by > 0`: stop and resolve the Console release baseline explicitly. Do not guess a range or publish RustFS.
+
+2. Clone/fetch `rustfs/console` into a scratch directory and record its exact `main` commit. Before creating a tag, check for a `v*` tag or Release workflow already associated with that hash. If one is in progress, wait for it instead of creating another version:
+
+```bash
+CONSOLE_SCRATCH=$(mktemp -d)
+gh repo clone "$CONSOLE_REPO" "$CONSOLE_SCRATCH/console"
+git -C "$CONSOLE_SCRATCH/console" fetch origin main --tags
+CONSOLE_HASH=$(git -C "$CONSOLE_SCRATCH/console" rev-parse origin/main)
+git -C "$CONSOLE_SCRATCH/console" tag --points-at "$CONSOLE_HASH" 'v*'
+gh run list -R "$CONSOLE_REPO" --workflow release.yml --commit "$CONSOLE_HASH" --limit 5
+```
+
+If no release exists or is running for `CONSOLE_HASH`, create the selected annotated tag at that exact hash and push it:
+
+```bash
+git -C "$CONSOLE_SCRATCH/console" tag -a "<console-tag>" -m "Release <console-tag>" "$CONSOLE_HASH"
+git -C "$CONSOLE_SCRATCH/console" push origin "<console-tag>"
+```
+
+Console tags include the `v` prefix. Pushing the tag triggers `.github/workflows/release.yml` (`🚀 Release`). Remove `CONSOLE_SCRATCH` after the gate completes.
+
+3. Find the exact tag run and wait for completion:
+
+```bash
+gh run list -R "$CONSOLE_REPO" --workflow release.yml --branch "<console-tag>" --limit 1
+gh run watch -R "$CONSOLE_REPO" "<console-run-id>" --exit-status
+```
+
+4. Block until the published Release is non-draft, the latest endpoint returns the expected tag, and `rustfs-console-<console-tag>.zip` is uploaded, non-empty, and carries a `sha256:` digest:
+
+```bash
+gh release view -R "$CONSOLE_REPO" "<console-tag>" --json isDraft,isPrerelease,assets,url
+test "$(gh api "repos/${CONSOLE_REPO}/releases/latest" --jq .tag_name)" = "<console-tag>"
+test "$(gh api "repos/${CONSOLE_REPO}/releases/tags/<console-tag>" \
+  --jq '[.assets[] | select(.name == "rustfs-console-<console-tag>.zip" and .state == "uploaded" and .size > 0 and (.digest | startswith("sha256:")))] | length')" -eq 1
+```
+
+Treat a missing/mismatched asset, digest, latest tag, or failed/cancelled workflow as BLOCKED. Do not start Phase 1 until the Console gate passes. Record `CONSOLE_TAG`, `CONSOLE_HASH`, Console run URL, and Release URL for the final report.

.agents/skills/rustfs-release-publish/references/preview-acceptance.md

@@ -0,0 +1,52 @@
+# Preview Artifact Acceptance
+
+Read after Phase 3 succeeds. Complete every check below before the parent skill's manual confirmation gate. These checks cover the downloaded artifact, embedded Console, and latest rc client.
+
+## Phase 4 — Run the artifact locally, verify the console
+
+Work inside the session scratchpad directory; never leave stray data dirs.
+
+```bash
+gh release download "<preview-tag>" -p "rustfs-macos-aarch64-v<preview-tag>.zip" -D "$SCRATCH"
+cd "$SCRATCH" && unzip -o rustfs-*.zip
+./rustfs --version        # must report the PREVIEW TAG (build::TAG), not the Cargo.toml version, plus expected short SHA
+mkdir -p data
+RUSTFS_ACCESS_KEY=rustfsadmin RUSTFS_SECRET_KEY=rustfsadmin ./rustfs ./data
+```
+
+Defaults: S3 endpoint `:9000`, embedded console `:9001`.
+
+Checks (all must pass):
+
+- `./rustfs --version` reports the preview tag name and the short SHA of `PREVIEW_HASH`. Reporting `<target>` without the `-preview.N` suffix means the build did not embed the tag — treat as FAIL and investigate before proceeding.
+- `curl -fsS http://localhost:9000/health/ready` returns ready.
+- Startup log shows the embedded console being served (this was the regression that `fix(release): require embedded console assets` guards).
+- Open `http://localhost:9001` in the browser: login with `rustfsadmin`/`rustfsadmin`; dashboard renders without JS console errors; create a bucket, upload a file, download it back (byte-identical), delete the object and bucket. Keep the server running for Phase 5.
+
+## Phase 5 — Validate with the latest rc client
+
+`rc` is the RustFS CLI client from <https://github.com/rustfs/cli>.
+
+- Ensure the latest release is installed: compare `rc --version` against `gh api repos/rustfs/cli/releases/latest --jq .tag_name`; update via `brew upgrade rustfs/tap/rc` (or download the release binary).
+- Point it at the preview server and run the command matrix, recording PASS/FAIL per command:
+
+```bash
+rc alias set preview http://localhost:9000 rustfsadmin rustfsadmin
+rc ls preview/
+rc mb preview/rel-check
+rc cp <local-file> preview/rel-check/
+rc stat preview/rel-check/<file>
+rc cat preview/rel-check/<file>          # matches source
+rc cp preview/rel-check/<file> ./out && cmp <local-file> ./out
+rc cp -r <local-dir>/ preview/rel-check/dir/
+rc find preview/rel-check --name "*"
+rc share download preview/rel-check/<file> --expire 1h   # presigned URL fetchable via curl
+rc rm preview/rel-check/<file> && rc rm -r --force preview/rel-check/dir
+rc rb preview/rel-check
+rc admin user list preview/
+rc admin user add preview/ relcheckuser relchecksecret12
+rc admin user remove preview/ relcheckuser
+rc alias remove preview
+```
+
+- Any FAIL blocks the release. Afterwards stop the server and delete the scratch data directory.

.agents/skills/rustfs-release-version-bump/SKILL.md

@@ -4,17 +4,16 @@ description: "Prepare the version-file and release-asset bump for an exact RustF
 ---
 # RustFS Release Version Bump
 
-Use this skill to publish a RustFS release (alpha, beta, or stable) with a minimal, auditable diff and a complete ship flow (`edit -> verify -> commit -> push -> PR`).
+Use this skill to prepare and verify release version files. Commit, push, and PR steps apply only when included in the user's delivery scope; publishing release tags belongs to `rustfs-release-publish`.
 
 Validated baseline: release pattern used in PR `#2957`.
 
 ## Required inputs
 
 - Exact target version, for example `1.0.0-beta.4`.
-- Delivery scope:
+- Delivery scope: local (`edit/verify`), git (`commit/push`), or GitHub
-- Local only (`edit/verify`).
+  (`commit/push/PR`). Derive it from the conversation; when unspecified, prepare
-- Local + git (`commit/push`).
+  and verify locally without blocking on a delivery question.
-- Full GitHub flow (`commit/push/PR`).
 
 If target version is missing or ambiguous, stop and ask before editing.
 
@@ -23,7 +22,7 @@ Reject any target version containing `-preview`: preview identifiers are tag-onl
 ## Read before editing
 
 - `AGENTS.md` (root and nearest path-specific files).
-- `.github/pull_request_template.md`.
+- `.github/pull_request_template.md` only when preparing a PR.
 - Current branch status and diff against `origin/main`.
 
 ## Default release file scope
@@ -50,8 +49,7 @@ Only drop a file when the current repository release process clearly no longer r
 ## Step-by-step workflow
 
 1. Confirm intent and isolate scope
-- Confirm target version string exactly.
+- Use the exact target and delivery scope already supplied; ask only for a missing or ambiguous target or a material release-policy choice.
-- Confirm whether user requested local-only or full GitHub flow.
 - Inspect current branch and ensure only release-related files are touched for this task.
 
 2. Update workspace versions
@@ -82,18 +80,18 @@ Only drop a file when the current repository release process clearly no longer r
 4. Verify before shipping
 - Run:
 - `make pre-commit`
-- If `make pre-commit` fails, return `BLOCKED` with root cause and do not silently widen scope to fix unrelated issues unless user asks.
+- If `make pre-commit` fails, fix task-attributable failures and rerun affected checks. Report unresolved required checks as `BLOCKED`; do not silently widen scope to fix unrelated issues.
 
-5. Commit strategy
+5. Commit strategy (only when committing is authorized)
 - Preferred split when both parts changed:
 - `chore(release): prepare <version>` for `Cargo.toml` and `Cargo.lock`.
 - `chore(release): align release assets for <version>` for docs and packaging files.
 - If user asks for one commit, use one commit.
 - Stage only intended release files; do not include unrelated working tree changes.
 
-6. Push and PR
+6. Push and PR (only for the authorized delivery scope)
 - Push branch:
-- `git push -u origin <branch>` (first push), or `git push` (tracking already exists).
+- Use the user-requested or configured push remote: `git push -u <push-remote> <branch>` (first push), or `git push` when tracking is already configured.
 - Create PR with template headings unchanged:
 - `gh pr create --base main --head <branch> --title ... --body-file ...`
 - PR title/body must be English.

.agents/skills/security-advisory-lessons/SKILL.md

@@ -12,8 +12,9 @@ matched security surface, the concise security reference under
 ## Workflow
 
 1. Freeze the exact diff/head and identify the changed trust boundaries.
-2. Read [advisory-patterns.md](references/advisory-patterns.md), then apply only
+2. Inspect the headings in [advisory-patterns.md](references/advisory-patterns.md),
-   the matching sections. Useful headings are
+   then read the matching sections. Read the full map only for a broad security
+   audit. Useful headings are
    auth/admin, IAM/STS/OIDC, policy/plugins, S3/copy/multipart, protocols, paths,
    secrets/logging/RPC, browser/CORS/proxy, SSE, Object Lock, and serde.
 3. Trace unauthenticated, low-privilege, wrong-action/owner/bucket, malformed,

.agents/skills/security-advisory-lessons/references/advisory-patterns.md

@@ -108,9 +108,9 @@ Update this file only when an advisory adds or changes a reusable lesson, affect
 
 ### Serde deserialization and input validation
 
-- No `#[serde(deny_unknown_fields)]` found across the entire codebase. Lesson: all structs deserialized from untrusted input (S3 API XML/JSON, lifecycle rules, bucket policies, replication configs) should have `#[serde(deny_unknown_fields)]` to reject malformed or adversarial payloads.
+- Reject unknown fields in untrusted S3 API XML/JSON, lifecycle, policy, and replication input where compatibility permits. Check the current type and supported payload fixtures; do not infer the repository's current coverage from an older audit. Where extra fields are part of the compatibility contract, validate security-critical values explicitly.
 - `#[serde(default)]` on security-critical fields silently accepts missing values as zero/empty. Lesson: when a field has security implications (retention days, permissions, limits), validate the deserialized value explicitly rather than relying on defaults.
-- Integer fields deserialized from user input and cast with `as` (e.g., `i32 as u32`) can wrap negative values to large positives. Lesson: validate ranges before casting; use `try_into()` or clamp.
+- Integer fields deserialized from user input and cast with `as` (e.g., `i32 as u32`) can wrap negative values to large positives. Lesson: validate ranges before casting; use `try_into()` with a typed error, or clamp only when the domain explicitly requires saturation.
 - XML config typos (e.g., `"NoncurentDays"` instead of `"NoncurrentDays"`) are silently accepted when `deny_unknown_fields` is absent. Lesson: strict deserialization prevents silent misconfiguration that could cause data loss or unexpected retention behavior.
 
 ## Useful Search Seeds

.agents/skills/test-coverage-improver/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: test-coverage-improver
-description: Run project coverage checks, rank high-risk gaps, and propose high-impact tests to improve regression confidence for changed and critical code paths before release.
+description: Analyze a supplied coverage report or perform an explicitly requested RustFS coverage assessment, rank uncovered risks, and propose focused tests. Do not trigger for ordinary implementation verification, a single regression test, documentation wording, or release preparation without a coverage request.
 ---
 
 # Test Coverage Improver
@@ -9,58 +9,49 @@ Use this skill when you need a prioritized, risk-aware plan to improve tests fro
 
 ## Usage assumptions
 - Focus scope is either changed lines/files, a module, or the whole repository.
-- Coverage artifact must be generated or provided in a supported format.
+- Reuse a supplied coverage artifact when its revision, scope, and format match.
 - If required context is missing, call out assumptions explicitly before proposing work.
 
 ## Workflow
 
 1. Define scope and baseline
-   - Confirm target language, framework, and branch.
+   - Derive the revision and scope from the request, diff, or supplied report.
-   - Confirm whether the scope is changed files only or full-repo.
+   - Default to the affected files/module; whole-workspace coverage requires that
+     scope in the request. Ask only if a wrong scope would change the result.
 
-2. Produce coverage snapshot
+2. Obtain coverage evidence
-   - Rust: `cargo llvm-cov` (or `cargo tarpaulin`) with existing repo config.
+   - First inspect a matching existing artifact; do not regenerate it merely
-   - JavaScript/TypeScript: `npm test -- --coverage` and read `coverage/coverage-final.json`.
+     because this skill was selected.
-   - Python: `pytest --cov=<pkg> --cov-report=json` and read `coverage.json`.
+   - If measurement is needed, read the Coverage section of
-   - Collect total, per-file, and changed-line coverage.
+     [the testing guide](../../../docs/testing/README.md#coverage), check disk
+     space/tool availability, and select package/test-scoped `cargo llvm-cov`
+     using the repository's nextest configuration. `make coverage` measures the
+     whole workspace (excluding E2E) and is only for that requested scope.
+   - Collect only metrics the report supports. Missing branch/changed-line
+     coverage is unknown, not zero.
+   - If measurement cannot run, continue with code-based test proposals and
+     mark measured coverage unverified; do not invent a coverage percentage.
 
 3. Rank highest-risk gaps
    - Prioritize changed code, branch coverage gaps, and low-confidence boundaries.
    - Apply the risk rubric in [coverage-prioritization.md](references/coverage-prioritization.md).
-   - Keep shortlist to 5–8 gaps.
+   - Report up to 5–8 evidenced gaps; do not pad a small scope.
    - For each gap, capture: file, lines, uncovered branches, and estimated risk score.
 
 4. Propose high-impact tests
-   - For each shortlisted gap, output:
+   - For each gap, name the behavior and regression, distinguishing assertions,
-     - Intent and expected behavior.
+     relevant normal/edge/failure cases, necessary setup, and estimated effort.
-     - Normal, edge, and failure scenarios.
+   - Include only scenarios and setup that apply; reuse shared fixture details.
-     - Assertions and side effects to verify.
-     - Setup needs (fixtures, mocks, integration dependencies).
-     - Estimated effort (`S/M/L`).
 
 5. Close with validation plan
    - State which gaps remain after proposals.
-   - Provide concrete verification command and acceptance threshold.
+   - Give a scoped verification command and behavior-based acceptance criterion;
+     use a coverage threshold only when the task or repository requires one.
    - List assumptions or blockers (environment, fixtures, flaky dependencies).
 
-## Output template
+## Report
 
-### Coverage Snapshot
+Summarize the supported metrics, then combine each ranked gap with its proposed
-- total / branch coverage
+test and validation. Include source lines only when supplied or inspected;
-- changed-file coverage
+mark missing metrics or locations as unknown. Do not duplicate gaps and tests
-- top missing regions by size
+in separate templates or fill empty categories for an otherwise small report.
-
-### Top Gaps (ranked)
-- `path:line-range` | risk score | why critical
-
-### Test Proposals
-- `path:line-range`
-  - Test name
-  - scenarios
-  - assertions
-  - effort
-
-### Validation Plan
-- command
-- pass criteria
-- remaining risk

.agents/skills/test-coverage-improver/agents/openai.yaml

@@ -1,4 +1,4 @@
 interface:
   display_name: "Test Coverage Improver"
   short_description: "Find top uncovered risk areas and propose high-impact tests."
-  default_prompt: "Run coverage checks, identify largest gaps, and recommend highest-impact test cases to improve risk coverage."
+  default_prompt: "Use $test-coverage-improver to analyze coverage for the requested scope, reuse matching reports, and propose tests for evidenced risks."

.agents/skills/tier-debug/SKILL.md

@@ -5,8 +5,11 @@ description: Debug ILM tiering / lifecycle transition issues — NoSuchVersion o
 
 # Tier / ILM Debugging
 
-Full playbook: [docs/operations/tier-ilm-debugging.md](../../../docs/operations/tier-ilm-debugging.md)
+Playbook: [docs/operations/tier-ilm-debugging.md](../../../docs/operations/tier-ilm-debugging.md).
-— read it before changing tier code.
+Read the section matching the symptom: metadata/`xl.meta`, runtime versionId,
+manual jobs, or retained-record recovery. Read the local-first expiry invariant
+before changing cleanup ordering. Before a reconcile/disposition action, read
+its entire procedure and retain its exact-evidence and confirmation gates.
 
 Quick moves:
 

.codex/config.toml

@@ -0,0 +1,5 @@
+# Bound individual tool outputs retained in context; retrieve relevant ranges
+# from task-owned log files when more evidence is needed.
+# https://learn.chatgpt.com/docs/config-file/config-reference
+
+tool_output_token_limit = 4000

.config/e2e-full-selection.txt

@@ -1,2 +1,2 @@
-sha256-darwin=53b05ac745905809d3828c6994bdd8ecf9d20b2b61a8a9d80fe15eb62f932193
+sha256-darwin=f0c78fdb93471575d9a64c5c46eae6c806bdd0bc10a6e33d7fb574aabd8db5a3
-sha256-linux=7c892afa4b9d1591b46bd79c976b647109a277284fddb3b98edced4b0297eda2
+sha256-linux=03ed7016cab672de9320e31375a0358eceacb4408b0e79cf063614fa7c878b87

.config/e2e-nightly-selection.txt

@@ -1,2 +1,2 @@
-sha256-darwin=a5665318c9bdc0947514fb7008ba1b83b114b739fac775c3c446f207058b7c7a
+sha256-darwin=364f2329a7b72eb9f1608dbe1a3af37af4095354014f3cbe23ca448492d89961
-sha256-linux=45d80e1723de5d25bb5b81f3ef5c82f583efc3e4f036a8cd2bb99e4f1eca9e51
+sha256-linux=60983f1ebe7068cf660d473c5f76c76a650410ccc99d71934ddca7fd67607987

.config/e2e-repl-nightly-selection.txt

@@ -1 +1 @@
-sha256=95c8adc016bbc0df9fb2afa24a108bcdf6567ec4d0518725a6cae301593ab556
+sha256=0e338d305260229e17ccfb2adc48a6212dbdfea36a9ebfb5a4e0d38658e6cc45

.config/e2e-smoke-selection.txt

@@ -1 +1 @@
-sha256=5db88c6fec94d4f269c7d9cfc128bd2adc27b3d7021127e2fa0b1daccc5f900f
+sha256=6d18f9cce820c51d5589de944e8cc185f73eeca0ea9a9916651943e3759169d0

.config/make/tests.mak

@@ -47,6 +47,7 @@ script-tests: ## Run shell script tests
 	bash -n ./scripts/validate_object_data_cache_cold_stampede.sh
 	$(RUSTFS_PYTHON_BIN) ./scripts/check_object_data_cache_follower_samples.py --self-test
 	./scripts/validate_object_data_cache_cold_stampede.sh --self-test
+	./scripts/run_scanner_heal_evidence_case.sh --self-test
 
 .PHONY: test
 test: core-deps script-tests ## Run all tests (needs cargo-nextest; RUSTFS_ALLOW_CARGO_TEST_FALLBACK=1 to override)

.config/scanner-heal-required-tests.json

@@ -8,10 +8,26 @@
       "suite": "e2e_test",
       "name": "heal_erasure_disk_rebuild_test::tests::test_cluster_root_heal_recovers_remote_shards_after_background_target_restart",
       "oracle": "background-target-restart.json",
+      "evidence": "process-restart",
+      "unclean_shutdown_marker": false,
       "min_objects": 9,
       "max_objects": 65,
       "topology": {"nodes": 4, "drives_per_node": 1},
       "scope": "Target process restart, exact unversioned S3 bodies and replacement-disk shards; not power loss or EC8+4."
+    },
+    "background-target-crash": {
+      "gate": "G14",
+      "task": "W21",
+      "lane": "e2e-nightly",
+      "suite": "e2e_test",
+      "name": "heal_erasure_disk_rebuild_test::tests::test_cluster_root_heal_recovers_remote_shards_after_background_target_crash",
+      "oracle": "background-target-crash.json",
+      "evidence": "process-crash-restart",
+      "unclean_shutdown_marker": true,
+      "min_objects": 9,
+      "max_objects": 65,
+      "topology": {"nodes": 4, "drives_per_node": 1},
+      "scope": "Target process killed during partial background rebuild, real unclean-shutdown marker, exact unversioned S3 bodies and replacement-disk shards; not power loss or EC8+4."
     }
   },
   "release_pending": {

.config/security-smoke-floor.txt

@@ -9,4 +9,4 @@
 # if the selected count drops below this number, so a rename or removal that
 # thins the security smoke gate must update this file in the same PR.
 # Adding tests does not require a bump, but bumping keeps the guard tight.
-18
+26

.github/workflows/ci.yml

@@ -900,6 +900,12 @@ jobs:
         run: |
           sudo apt-get install -y iptables
           sudo -n iptables --version
+          # The endpoint-blackhole heal scenario needs CAP_NET_ADMIN. Containerised
+          # runners can run iptables but not touch the rule set; the test then logs
+          # a skip instead of failing, so surface that here where it is visible.
+          if ! sudo -n iptables -w 5 -S OUTPUT >/dev/null 2>&1; then
+            echo "::warning::iptables cannot read the OUTPUT chain on this runner (no CAP_NET_ADMIN); the endpoint-blackhole heal scenario will be skipped"
+          fi
 
       - name: Set up Python
         uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1 # v6.3.0

.github/workflows/docker.yml

@@ -121,6 +121,22 @@ jobs:
           create_latest=false
           source_ref="$GITHUB_SHA"
 
+          # Pre-GA policy: until the first stable (vX.Y.Z) tag exists, every
+          # prerelease (alpha/beta/rc) also moves `latest`, so users pulling
+          # `latest` get the newest test build. Once a stable tag is published
+          # this returns false and `latest` follows stable releases only.
+          prerelease_moves_latest() {
+            local stable_tags
+            stable_tags=$(git ls-remote --tags --refs origin 2>/dev/null \
+              | awk '{print $2}' \
+              | grep -E '^refs/tags/v?[0-9]+\.[0-9]+\.[0-9]+$' || true)
+            if [[ -z "$stable_tags" ]]; then
+              return 0
+            fi
+            echo "ℹ️  Stable release tag(s) already exist; prereleases no longer update latest"
+            return 1
+          }
+
           if [[ "${{ github.event_name }}" == "workflow_run" ]]; then
             # Triggered by build workflow completion
             echo "🔗 Triggered by build workflow completion"
@@ -184,8 +200,8 @@ jobs:
                 if [[ "$version" == *"alpha"* ]] || [[ "$version" == *"beta"* ]] || [[ "$version" == *"rc"* ]]; then
                   build_type="prerelease"
                   is_prerelease=true
-                  # Current policy: create latest tags for stable releases and selected prereleases (alpha/beta).
+                  # Pre-GA policy: prereleases update latest until the first stable tag exists.
-                  if [[ "$version" == *"alpha"* ]] || [[ "$version" == *"beta"* ]]; then
+                  if prerelease_moves_latest; then
                     create_latest=true
                     echo "🧪 Building Docker image for prerelease: $version (creating latest tag)"
                   else
@@ -243,8 +259,8 @@ jobs:
               v*alpha*|v*beta*|v*rc*|*alpha*|*beta*|*rc*)
                 build_type="prerelease"
                 is_prerelease=true
-                # Current policy: create latest tags for stable releases and selected prereleases (alpha/beta).
+                # Pre-GA policy: prereleases update latest until the first stable tag exists.
-                if [[ "$version" == *"alpha"* ]] || [[ "$version" == *"beta"* ]]; then
+                if prerelease_moves_latest; then
                   create_latest=true
                   echo "🧪 Building with prerelease version: $input_version (creating latest tag)"
                 else
@@ -394,11 +410,13 @@ jobs:
           TAG_BASE="${VERSION}${VARIANT_SUFFIX}"
           TAGS="${{ env.REGISTRY_DOCKERHUB }}:$TAG_BASE,${{ env.REGISTRY_GHCR }}:$TAG_BASE,${{ env.REGISTRY_QUAY }}:$TAG_BASE"
 
-          # Add channel tags for prereleases and latest for stable
+          # Add latest when requested (stable releases, and prereleases before GA)
           if [[ "$CREATE_LATEST" == "true" ]]; then
-            # Create latest tags for stable releases and selected prereleases when CREATE_LATEST=true.
             TAGS="$TAGS,${{ env.REGISTRY_DOCKERHUB }}:latest${VARIANT_SUFFIX},${{ env.REGISTRY_GHCR }}:latest${VARIANT_SUFFIX},${{ env.REGISTRY_QUAY }}:latest${VARIANT_SUFFIX}"
-          elif [[ "$BUILD_TYPE" == "prerelease" ]]; then
+          fi
+
+          # Always add the channel tag for prereleases, independent of latest
+          if [[ "$BUILD_TYPE" == "prerelease" ]]; then
             # Prerelease channel tags (alpha, beta, rc)
             if [[ "$VERSION" == *"alpha"* ]]; then
               CHANNEL="alpha"
@@ -555,7 +573,7 @@ jobs:
             "prerelease")
               echo "🧪 Prerelease Docker image has been built with ${VERSION} tags"
               echo "⚠️  This is a prerelease image - use with caution"
-              # Create latest tags for stable releases and selected prereleases when CREATE_LATEST=true.
+              # Prereleases move latest until the first stable tag exists (pre-GA policy).
               if [[ "$CREATE_LATEST" == "true" ]]; then
                 echo "🏷️  Latest tag has been created for prerelease: $VERSION"
               else

.github/workflows/e2e-upgrade.yml

@@ -82,6 +82,14 @@ jobs:
             cache_key: e2e-odm-config-rollback
             test: rc5_rollback_requires_restoring_odm_configuration
             artifact: odm-config-rollback
+          - name: Multipart layouts survive the rc.5 upgrade
+            cache_key: e2e-multipart-layout-upgrade
+            test: direct_upgrade_from_rc5_preserves_multipart_layouts
+            artifact: multipart-layout-upgrade
+          - name: rc.5 multipart replication baseline
+            cache_key: e2e-multipart-layout-baseline
+            test: rc5_baseline_replicates_multipart_layouts
+            artifact: multipart-layout-baseline
     runs-on: ubuntu-latest
     timeout-minutes: 60
     env:

.github/workflows/star-history.yml

@@ -17,7 +17,7 @@ jobs:
     runs-on: ubuntu-latest
     timeout-minutes: 10
     steps:
-      - uses: overtrue/repo-visuals-action@fd79cba437ecfac933d00a69add17eb95d3939c3 # v1.3.1
+      - uses: overtrue/repo-visuals-action@ee2c632f6ce617e851fb46ea935ee8af762ebb93 # v1.4.0
         with:
           github-token: ${{ github.token }}
           output-branch: star-history

AGENTS.md

@@ -7,8 +7,11 @@ This file contains repository-wide rules. Use the nearest subdirectory
 
 1. System/developer instructions.
 2. The current user request.
-3. The nearest `AGENTS.md`.
+3. Applicable `AGENTS.md` files, with the nearest file winning conflicts.
-4. This file.
+4. Selected skills and reference documents.
+
+Nested instructions add to ancestor rules; they do not discard non-conflicting
+rules. A skill cannot expand the user's requested scope or grant authorization.
 
 ## Operating Model
 
@@ -21,63 +24,29 @@ This file contains repository-wide rules. Use the nearest subdirectory
 - Do not load every skill or inspect unrelated modules preemptively. Select a
   skill only when its description directly matches the request or changed
   surface.
+- Resolve repository workflow skills under `.agents/skills/` when a global
+  skill has the same name, unless the user explicitly selects another path.
 - Avoid repeated reads and equivalent verification commands once enough
   evidence exists.
+- Search for relevant symbols/headings before reading long files; return only
+  matching ranges. If output is truncated, narrow the query instead of repeating
+  a full read. Keep reusable raw logs in task artifacts and report the evidence.
+- Reuse authorization already given in the conversation. Resolve routine choices
+  within that scope and continue independent work while a material question is
+  pending. Before requesting missing approval, prepare the concrete result that
+  is already authorized; retain explicit merge and release gates.
 
-## Worktree and Disk Hygiene
+## Task-Specific Guidance
 
-- Start implementation from the latest `origin/main` and confirm the requested
+Read only the reference needed for the current task, once per unchanged context:
-  change is not already present.
-- An existing clean, isolated task worktree is sufficient. Create another
-  worktree only when the current checkout is shared, dirty with unrelated work,
-  or belongs to another task.
-- Never commit from a shared checkout.
-- Use a task-specific branch named `<type>/<topic>`, such as `fix/...`,
-  `feat/...`, `test/...`, or `docs/...`, unless the user specifies a name.
-- Do not include agent, tool, contributor, account, or organization names in
-  branch names.
-- Push to the user-requested remote or the repository's configured push remote.
-  Do not hard-code or infer a remote from an account name.
-- Check free space before artifact-heavy builds, tests, coverage, or downloads.
-  Re-check before a broad gate when space is tight.
-- Remove only task-owned temporary/build artifacts. Never delete another task's
-  worktree or uncommitted data.
-- At handoff, mention disk or cleanup details only when they affected execution
-  or artifacts/worktrees remain intentionally.
 
-## Change Style
+- Before code changes or artifact-heavy work, read [implementation rules](.agents/references/implementation.md).
-
+  For a read-only code review, use its change-style and boundary sections as needed.
-- Preserve existing control flow unless changing it is required for correctness.
+- Before commits, pushes, PR creation/updates, or posting to PRs/issues/discussions,
-- Prefer a direct local edit over new files, wrappers, managers, or speculative
+  read [Git and PR rules](.agents/references/pull-requests.md).
-  abstractions.
+  Reuse existing authorization; a reference does not authorize posting, merging, or publishing.
-- Add a helper only when it removes current duplication, names a real domain
+- Preserve unrelated work. Never commit from a shared checkout or delete another task's artifacts.
-  boundary, or isolates a non-trivial invariant.
+- Source comments, commits, PR titles, and PR bodies are in English.
-- Remove an in-scope path superseded by the change. If compatibility requires it,
-  adapt at the boundary to one canonical core and use the repository's
-  `RUSTFS_COMPAT_TODO` policy.
-- Comments explain non-obvious invariants or reasons. Do not narrate code or
-  record change history.
-- Mention unrelated problems when useful; do not fix them in a narrow task.
-
-## Reuse and Boundary Rules
-
-- Before adding helpers, constants, fixtures, or wrappers, search the touched
-  crate, the domain-owning crate, `crates/utils`, `crates/common`, and relevant
-  direct dependencies.
-- Reuse requires matching semantics: normalization, error types, deadlines,
-  durability, and compatibility must fit the call site. A narrowly named local
-  helper is better than forced reuse with different semantics.
-- Validate untrusted input at its trust boundary, then trust the validated type.
-  Values crossing disk, RPC, persistence, or version boundaries remain
-  untrusted at every consumer.
-- Re-check boundary values immediately before destructive actions such as
-  delete, overwrite, or quorum decisions.
-- Every new branch needs a concrete triggering input/state. For decoded or peer
-  data, corruption and mixed-version input are valid triggers.
-- Required values must return a typed error when absent or corrupt; do not use a
-  default that converts corruption into a plausible result.
-- Attach error context once where it is actionable. Do not erase typed errors
-  below aggregation or quorum layers.
 
 ## Sources of Truth
 
@@ -147,73 +116,25 @@ requested adversarial/design reviews, and agent-instruction changes that alter
 execution. Ordinary questions, diagnoses, status reports, non-adversarial code
 reviews, and low-risk planning do not trigger it.
 
-Risk and review shape:
+For applicable work and substantial PR reviews, read the [risk tiers and review shape](.agents/references/adversarial-validation.md).
+Load only the matching domain probes; ordinary reviews do not become adversarial
+merely because this reference exists.
 
-- **Exempt:** documentation, comments, formatting, or typos with no runtime,
+A review has no finding quota; `No findings` is a complete outcome. A request to
-  build, test, or agent-execution effect.
+find problems is not evidence that a defect exists. Before reporting a candidate,
-- **Mechanical:** renames, moves, test/tooling-only changes, and agent-rule
+check callers, invariants, and existing tests for evidence that disproves it.
-  changes. Run correctness and simplicity lenses.
+Findings need `file:line` and a concrete failure or violation of an explicit
-- **Standard:** localized behavior changes. Run one integrated final-diff pass
+requirement. Missing required tests/checks are verification gaps, not proof of a
-  covering correctness, simplicity, and test coverage; add only domain lenses
+runtime bug; name the unprotected behavior or unmet gate. Keep optional style or
-  matched by the diff.
+refactoring preferences out of defect findings unless that review was requested.
-- **High risk / substantial PR review:** high risk includes locking,
-  erasure/quorum/heal, replication, multipart, RPC, lifecycle/tiering,
-  persistence/fsync, IAM/KMS/auth, cryptography, on-disk/on-wire formats, and
-  S3-visible semantics. Cover all applicable lenses using exactly two
-  independent reviewers when delegation is explicitly authorized. Split the
-  lenses between them. Otherwise perform two fresh sequential passes.
-- **Outbound client defaults:** what `TargetClient`, `PutObjectOptions`, or
-  the remote SDK configuration sends to every replication or migration target
-  is high risk for every target class even when the change fixes one. Follow
-  the SOP in `docs/postmortems/2026-09-03-replication-checksum-default-regression.md`:
-  run the outbound target matrix, document each new env knob in the same PR,
-  and list verified and unverified target classes in the PR Impact section.
 
-Available domain lenses are security, concurrency/durability, compatibility,
+Fix or rebut supported findings within the authorized scope. Once the required
-and performance. Select `.agents/skills/adversarial-validation/SKILL.md` for an
+passes are complete, stop. Reopen only for changed code, new evidence, an
-explicit adversarial request, a high-risk change, or a substantial PR review;
+unresolved finding, or an explicit re-review request; an unchanged diff does not
-then read only its matching role references. A routine standard pass does not
+need another pass at every conversation turn or workflow handoff.
-load the playbook unless the reviewer needs a RustFS-specific probe.
-
-A finding must name a concrete input/state/interleaving and wrong outcome, or a
-specific missing regression check, with `file:line`. Resolve it by fixing the
-diff or rebutting it with code-path/test/invariant evidence. After a non-trivial
-fix, rerun only affected lenses.
 
 For high-risk PRs, record one concise verdict per covered lens in the PR body.
 
-## Pull Request Lifecycle
-
-- Creating or updating a PR includes one immediate snapshot of checks,
-  mergeability, reviews, and unresolved threads.
-- Unless the user explicitly requests monitoring, a release workflow requires
-  it, or an automation already owns it, hand off after the PR is open with the
-  current state and next event to watch. Do not delay ordinary handoff with
-  fixed quiet-period sleeps.
-- For requested monitoring, use event-driven or bounded waits. Report only state
-  changes, actionable failures, or a meaningful prolonged delay.
-- Investigate failures/comments before changing code. Fix task-attributable
-  issues, rerun affected verification, push, reply or resolve the thread, then
-  resume the requested monitor.
-- Never merge without required reviewer approval or explicit authority.
-- After an observed merge, verify the commit reached the base, then clean the
-  task worktree/branch when safe. Preserve unmerged work for closed PRs unless
-  deletion was explicitly authorized.
-
-## Git and PR Baseline
-
-- Follow Conventional Commits; keep the subject at most 72 characters.
-- Source comments, commits, PR titles, and PR bodies are in English.
-- Keep every heading from `.github/pull_request_template.md`; use `N/A` where
-  needed and include commands actually run.
-- Use `--body-file` for multiline `gh pr create`/`gh pr edit` content.
-- PR/issue/discussion content must not contain the literal sequence `\n` or
-  hard-wrapped prose paragraphs.
-- Do not include local absolute paths or tool-specific labels/prefixes in GitHub
-  content.
-- Resolve review threads after the underlying issue is fixed. If declining a
-  suggestion, reply with a short evidence-based reason.
-
 ## Security Baseline
 
 - Never commit secrets, credentials, or key material.
@@ -250,12 +171,6 @@ Use `.agents/skills/rustfs-logging-governance/SKILL.md` for logging changes.
 - `DataUsageCacheInfo` and `DataUsageEntry` keep their hand-written map
   serialization and new fields remain `#[serde(default)]` for older readers.
 
-## Naming
-
-Use Rust API naming: `SCREAMING_SNAKE_CASE` constants/statics, `snake_case`
-functions/variables, and `PascalCase` types. Do not rename unrelated existing
-violations.
-
 ## Scoped Guidance
 
 Before editing, locate the nearest instructions with:
@@ -265,4 +180,4 @@ git ls-files '*AGENTS.md'
 ```
 
 The nearest file wins for domain invariants. Keep generic workflow and
-validation policy in this root file.
+validation policy in this root file and its task-specific references.

CHANGELOG.md

@@ -7,7 +7,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Security
+- **Presigned URLs honour only signed headers** (GHSA-g8w9-qw9q-fghr): a SigV4 presigned request that carries an `x-amz-*` request header not listed in `X-Amz-SignedHeaders` is now rejected with `403 AccessDenied` ("There were headers present in the request which were not signed"), matching AWS S3. Previously the holder of a presigned `PutObject` URL could add unsigned `x-amz-tagging`, `x-amz-storage-class`, `x-amz-website-redirect-location`, ACL, metadata, Object Lock or SSE headers and have them applied. Presigners that intend a property must set it before signing so the SDK lists the header in `SignedHeaders`; `x-amz-cf-id` (CloudFront) remains tolerated unsigned. Header-signed SigV4 and SigV2 requests are unchanged.
+
 ### Fixed
+- **Fresh multi-pool bootstrap with distinct format creators**: a new deployment whose pools have their first endpoint on different nodes (for example two single-node pools) could never publish its initial `pool.bin`: each node held fresh-bootstrap proof only for the pool it formatted, the deployment-wide proof collapsed to none, and every node died with `pool metadata recovery required: no durable bootstrap identity or pool.bin replica is available` after the startup retry budget. The first pool's creator now mints the pending cluster identity on its own pool, every other creator copies that nonce-bound identity onto the pool it formatted first-hand, and the elected writer publishes `pool.bin` once every pool replica carries the same pending identity. Corrupt or disagreeing replicas, pools that merely have a format, expansion pools joining an initialized deployment, and restarts without first-hand proof still fail closed. Non-elected nodes that start before `pool.bin` exists, and the elected writer while it waits for the other creators, no longer latch their pool-metadata write gate for the life of the process. Refs rustfs/backlog#2338, rustfs/backlog#2375.
+- **Lock RPC timeout storms** (#7363): the remote lock client no longer evicts and re-dials the shared internode HTTP/2 channel on every request deadline. A timeout evicts only when the peer has not completed any lock RPC for two deadlines, evictions and transport-failure re-dials are rate limited per peer (`RUSTFS_OBJECT_LOCK_RPC_EVICTION_COOLDOWN_MS`, default 5 s), and a timed-out request is left running instead of being reset (bounded per peer by `RUSTFS_OBJECT_LOCK_RPC_DETACHED_LIMIT`, default 256), so a slow lock endpoint can no longer drive the `RST_STREAM`/`GOAWAY too_many_resets`/reconnect loop. A lock granted after its caller timed out is released immediately, and unlocks that fail the quick retries continue on a deferred 1/2/4/8/16 s schedule before the server lease reclaims them. New `rustfs_remote_lock_*` metrics cover timeouts, evictions, suppressed evictions, detached streams, late completions and late releases per peer. Operator guide at `docs/operations/lock-rpc-storm-protection.md`.
+- **Multipart admission queue**: an `UploadPart` waiting for a foreground write permit now waits at most 10 s by default (`RUSTFS_PUT_MULTIPART_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS`, previously 30 s), so a queued part returns S3 `SlowDown` before the client's socket write timeout drops the connection. Separately, the API listener no longer forces a 4 MiB `SO_RCVBUF` on every accepted socket (kernel autotuning applies; `RUSTFS_HTTP_SOCKET_RECV_BUFFER_BYTES` restores a fixed size), so a queued part no longer lets up to 8 MiB of unread body accumulate in kernel memory per connection, which is what throttled whole nodes under SDK-default multipart concurrency. Fixes #7385.
 - **Helm Ingress**: `customAnnotations` are now merged with class-specific annotations (nginx/traefik) instead of being ignored when `ingress.className` is set.
 - **Per-pool erasure parity**: Erasure parity (STANDARD and reduced-redundancy) is now resolved independently for every pool instead of reusing the first pool's value. A heterogeneous topology — for example a 4-drive pool plus a 2-drive pool created during expansion — previously inherited the first pool's parity and could resolve to zero data shards in the smaller pool, panicking Reed-Solomon construction on write. Automatic parity now resolves per pool (for example `2+2` in the 4-drive pool and `1+1` in the 2-drive pool). Fixes #4801.
 

CLAUDE.md

@@ -13,7 +13,7 @@ what Claude Code needs on top: commands and pointers.
 cargo build --release --bin rustfs   # production binary
 cargo check -p <crate>               # fast type-check one crate
 cargo test -p <crate>                # test one crate
-cargo fmt --all                      # format (required before PR)
+cargo fmt --all --check              # for Rust changes; see AGENTS.md verification tiers
 make pre-commit                      # fast gate: fmt + arch checks + quick-check (NO clippy/tests)
 make pre-pr                          # optional full gate for broad cross-module changes
 make build-docker BUILD_OS=ubuntu22.04
@@ -42,5 +42,5 @@ make build-docker BUILD_OS=ubuntu22.04
 
 Repo-wide domain invariants (dual internal metadata keys, defensive UUID
 reads, unversioned tier buckets) live in [AGENTS.md](AGENTS.md) under
-"Cross-Cutting Domain Invariants" — read them before touching metadata or
+"Cross-Cutting Storage Invariants" — read them before touching metadata or
 tiering code.

Cargo.lock

@@ -2527,18 +2527,18 @@ checksum = "790eea4361631c5e7d22598ecd5723ff611904e3344ce8720784c93e3d83d40b"
 
 [[package]]
 name = "crossbeam-channel"
-version = "0.5.16"
+version = "0.5.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d85363c37faeca707aef026efa9f3b34d077bce547e48f770770625c6013679e"
+checksum = "98b0cc327b5bc766e7fda9c9260cc0fa81b43a8e240440422dff70788e3f9ef1"
 dependencies = [
  "crossbeam-utils",
 ]
 
 [[package]]
 name = "crossbeam-deque"
-version = "0.8.7"
+version = "0.8.8"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5181e0de7b61eb03a81e347d6dd8797bae9da5146707b51077e2d71a54ec0ceb"
+checksum = "622f3fc73690be383c7214310406f28a90e6edeadc3cea882f9d71e495b9711a"
 dependencies = [
  "crossbeam-epoch",
  "crossbeam-utils",
@@ -2546,27 +2546,27 @@ dependencies = [
 
 [[package]]
 name = "crossbeam-epoch"
-version = "0.9.20"
+version = "0.9.21"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2d6914041f254d6e9176c01941b21115dcfb7089e55135a35411081bd106ef3f"
+checksum = "dc74980687109a3b14c72fd458107bf0baa1da1a1a805e178d15501ba9b86d9d"
 dependencies = [
  "crossbeam-utils",
 ]
 
 [[package]]
 name = "crossbeam-queue"
-version = "0.3.13"
+version = "0.3.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "803d13fb3b09d88be9f4dbc29062c66b19bf7170867ceb746d2a8689bf6c7a26"
+checksum = "03e8bd762f7479489c70ed6c768ddca99d7296857de437a68dcb2a94365b3fae"
 dependencies = [
  "crossbeam-utils",
 ]
 
 [[package]]
 name = "crossbeam-utils"
-version = "0.8.22"
+version = "0.8.23"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "61803da095bee82a81bb1a452ecc25d3b2f1416d1897eb86430c6159ef717c17"
+checksum = "a31eee39dddec8330830986fcd7625edb5a24ec90ea038215273bbc3adb08ac6"
 
 [[package]]
 name = "crunchy"
@@ -3673,9 +3673,9 @@ dependencies = [
 
 [[package]]
 name = "der"
-version = "0.8.1"
+version = "0.8.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a69dedd701da44b0536442edf09c81a64b0ab97a7a4a5e3d1971f00027cbc63d"
+checksum = "a878c850e9e421b20262e9b41f9c860e4785fa07541c266b62ff9d1ef998a80a"
 dependencies = [
  "const-oid 0.10.2",
  "pem-rfc7468 1.0.0",
@@ -4091,7 +4091,7 @@ version = "0.17.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c0681a4fc24c767085329728d8dfba959af91228aa4610cca4f8ce317ba46ae0"
 dependencies = [
- "der 0.8.1",
+ "der 0.8.2",
  "digest 0.11.3",
  "elliptic-curve 0.14.1",
  "rfc6979 0.6.0",
@@ -5735,9 +5735,9 @@ dependencies = [
 
 [[package]]
 name = "ipnet"
-version = "2.12.1"
+version = "2.12.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6a756c3fac73139e83f14c2d742155dd2b78d3ee56597b419a0579b7bdd6dd78"
+checksum = "791930b43c0d5973160d90a8f3894509f2b273430f5c5c73b668636d0287c5c0"
 dependencies = [
  "serde",
 ]
@@ -6140,9 +6140,9 @@ checksum = "3eaf3ede3fee6db1a4c2ee091bf8a8b4dccdc6d17f656fb07896ee72867612f2"
 
 [[package]]
 name = "libflate"
-version = "2.3.1"
+version = "2.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a4da9b700e758e57152a1fd1c52cbdc5727c1aa6d8743dc1acda917398f1d76c"
+checksum = "561a8da1a50e1428d3c51321dafeca849df992a5bb67720c386131234caba82e"
 dependencies = [
  "adler32",
  "crc32fast",
@@ -7934,7 +7934,7 @@ version = "0.8.0-rc.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "986d2e952779af96ea048f160fd9194e1751b4faea78bcf3ceb456efe008088e"
 dependencies = [
- "der 0.8.1",
+ "der 0.8.2",
  "spki 0.8.0",
 ]
 
@@ -7977,7 +7977,7 @@ dependencies = [
  "aes 0.9.3",
  "aes-gcm",
  "cbc 0.2.1",
- "der 0.8.1",
+ "der 0.8.2",
  "pbkdf2 0.13.0",
  "rand_core 0.10.1",
  "scrypt 0.12.0",
@@ -8001,7 +8001,7 @@ version = "0.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "451913da69c775a56034ea8d9003d27ee8948e12443eae7c038ba100a4f21cb7"
 dependencies = [
- "der 0.8.1",
+ "der 0.8.2",
  "pkcs5 0.8.1",
  "rand_core 0.10.1",
  "spki 0.8.0",
@@ -8943,9 +8943,9 @@ dependencies = [
 
 [[package]]
 name = "redis"
-version = "1.6.0"
+version = "1.7.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e37a4ca5c6ca42aa3e6df2fd32b987a65d32a4c2159a6f3fe0fd1df306a2658f"
+checksum = "2acbc41a996f7652b2ddd9dfd98cc4ff602cfd742ae35382f07f608405ab50ed"
 dependencies = [
  "arc-swap",
  "arcstr",
@@ -9327,7 +9327,7 @@ dependencies = [
  "curve25519-dalek 5.0.0",
  "data-encoding",
  "delegate",
- "der 0.8.1",
+ "der 0.8.2",
  "digest 0.11.3",
  "ecdsa 0.17.0",
  "ed25519-dalek 3.0.0",
@@ -9907,6 +9907,7 @@ dependencies = [
  "regex",
  "rmp",
  "rmp-serde",
+ "rustfs-config",
  "rustfs-utils",
  "s3s",
  "serde",
@@ -10000,6 +10001,7 @@ dependencies = [
  "rustls-pki-types",
  "serde",
  "serde_json",
+ "serde_with",
  "serial_test",
  "sha1 0.11.0",
  "sha2 0.11.0",
@@ -10746,6 +10748,7 @@ dependencies = [
  "rustfs-data-usage",
  "rustfs-ecstore",
  "rustfs-filemeta",
+ "rustfs-heal",
  "rustfs-heal-contracts",
  "rustfs-lifecycle",
  "rustfs-lock",
@@ -10943,9 +10946,9 @@ dependencies = [
 
 [[package]]
 name = "rustfs-uring"
-version = "0.2.1"
+version = "0.2.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0486e62d0efe25db95c00aeacb2da84368adcba299216cda99fcb11328061c84"
+checksum = "b29bc57b4bd62a73f4fae408b536adf578332e50e464797d09dc2382c7cb68c2"
 dependencies = [
  "io-uring",
  "libc",
@@ -11420,7 +11423,7 @@ checksum = "d56d437c2f19203ce5f7122e507831de96f3d2d4d3be5af44a0b0a09d8a80e4d"
 dependencies = [
  "base16ct 1.0.0",
  "ctutils",
- "der 0.8.1",
+ "der 0.8.2",
  "hybrid-array",
  "subtle",
  "zeroize",
@@ -11994,7 +11997,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1d9efca8738c78ee9484207732f728b1ef517bbb1833d6fc0879ca898a522f6f"
 dependencies = [
  "base64ct",
- "der 0.8.1",
+ "der 0.8.2",
 ]
 
 [[package]]

Cargo.toml

@@ -191,6 +191,7 @@ rmp = { version = "0.8.15" }
 rmp-serde = { version = "1.3.1" }
 serde = { version = "1.0.229" }
 serde_ignored = { version = "0.1" }
+serde_with = { version = "3", default-features = false, features = ["macros", "std"] }
 serde_json = { version = "1.0.151" }
 serde_urlencoded = "0.7.1"
 
@@ -256,10 +257,10 @@ clap = { version = "4.6.6" }
 const-str = { version = "1.1.0" }
 convert_case = "0.12.0"
 criterion = { version = "0.8" }
-crossbeam-queue = "0.3.13"
+crossbeam-queue = "0.3.14"
-crossbeam-channel = "0.5.16"
+crossbeam-channel = "0.5.17"
-crossbeam-deque = "0.8.7"
+crossbeam-deque = "0.8.8"
-crossbeam-utils = "0.8.22"
+crossbeam-utils = "0.8.23"
 datafusion = { default-features = false, version = "55.0.0" }
 derive_builder = "0.20.2"
 enumset = "1.1.14"
@@ -306,7 +307,7 @@ rustfs-erasure-codec = { version = "8.0.2" }
 reed-solomon-simd = "3.1.0"
 regex = { version = "1.13.1" }
 rumqttc = { package = "rumqttc-next", version = "0.34.0" }
-redis = { version = "1.6.0" }
+redis = { version = "1.7.0" }
 rustify = { version = "0.7", default-features = false }
 rustix = { version = "1.1.4" }
 rust-embed = { version = "8.12.0" }

README.md

@@ -109,6 +109,15 @@ Star RustFS on GitHub and be instantly notified of new releases.
 
 ## Quickstart
 
+> [!IMPORTANT]
+> **Pool expansion notice:**
+>
+> - A single-node single-drive (SNSD) deployment is supported only as a standalone local path. It cannot expand in place or be added as a Pool. To move to a multi-drive topology, create a new deployment and migrate data through S3.
+> - Keep an existing multi-drive Pool's endpoints and Erasure Set width unchanged; expand by appending a new Pool. With ellipsis-based expansion, every Pool argument must contain an ellipsis expression and expand to at least two drive endpoints.
+> - Single-node multi-drive Pools and multi-node Pools with one drive per node are allowed, subject to valid Erasure Set geometry and EC settings; acceptance does not guarantee host-failure tolerance.
+>
+> These topology rules follow MinIO, but automatic parity selection differs between the projects. See the [Pool layout compatibility and regression tests](docs/testing/pool-layout-compatibility.md) before expanding a deployment.
+
 To get started with RustFS, follow these steps:
 
 ### 1. One-click Installation (Option 1)

README_ZH.md

@@ -89,6 +89,15 @@ RustFS 是一个基于 Rust 构建的高性能分布式对象存储系统。Rust
 
 ## 快速开始
 
+> [!IMPORTANT]
+> **Pool 扩容 Notice：**
+>
+> - 单节点单盘（SNSD）部署仅支持使用本地路径独立运行，不支持原地扩容，也不能作为 Pool 加入集群。如需改为多盘拓扑，请创建新部署并通过 S3 迁移数据。
+> - 已有多盘 Pool 的端点和 Erasure Set 宽度应保持不变，扩容应追加新的 Pool。使用省略号表达式扩容时，每个 Pool 参数都必须包含省略号表达式，并展开为至少两个磁盘端点。
+> - 允许单节点多盘 Pool，也允许多节点、每节点一盘的 Pool，但必须满足 Erasure Set 布局和 EC 配置要求；配置合法不代表能够容忍整台主机故障。
+>
+> 这些拓扑规则与 MinIO 一致，但两者的默认 parity 选择方式存在差异。扩容前请阅读 [Pool 布局兼容性与回归测试说明](docs/testing/pool-layout-compatibility.md)。
+
 请按照以下步骤快速上手 RustFS：
 
 ### 1. 一键安装脚本 (选项 1)

crates/AGENTS.md

@@ -33,8 +33,11 @@ Applies to all paths under `crates/`.
 
 ## Type Casting
 
-- Never use `as` for numeric conversions that may truncate or overflow. Use `try_into()` with explicit error handling, or clamp with `value.max(0) as usize` when the domain is bounded.
+- Never use `as` for numeric conversions that may truncate or overflow. Use
-- `f64 as usize` saturates but is fragile; clamp to `[0, usize::MAX as f64]` first.
+  `try_into()` with typed error handling; clamp or saturate only when the domain
+  explicitly requires it.
+- Before converting floating-point input to an integer, validate finiteness,
+  sign, and the destination range. A lower-bound clamp alone is insufficient.
 - Treat every `as` cast in a PR review as a potential bug; require justification.
 
 ## Testing

crates/common/src/mrf_channel.rs

@@ -90,6 +90,61 @@ pub struct MrfIntent {
     pub attempts: u8,
 }
 
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub struct MrfDurableRepairAnchor {
+    pub kind: MrfKind,
+    pub bucket: Arc<str>,
+    pub object: Arc<str>,
+    pub version_id: Option<[u8; 16]>,
+    pub scope: Option<MrfScope>,
+    pub lease: MrfIngressLease,
+    pub bucket_incarnation_id: Uuid,
+}
+
+impl MrfDurableRepairAnchor {
+    /// Build a dischargeable anchor only when the caller supplies the storage
+    /// incarnation and the original ingress lease. Legacy replay records lack
+    /// both pieces and therefore remain fail-closed.
+    pub fn from_intent(intent: &MrfIntent, bucket_incarnation_id: Uuid) -> Option<Self> {
+        if bucket_incarnation_id.is_nil() {
+            return None;
+        }
+        let lease = intent.lease?;
+        let (version_id, scope) = canonical_identity(intent.kind, intent.version_id, intent.scope);
+        Some(Self {
+            kind: intent.kind,
+            bucket: intent.bucket.clone(),
+            object: intent.object.clone(),
+            version_id,
+            scope,
+            lease,
+            bucket_incarnation_id,
+        })
+    }
+
+    pub fn is_proven_by(&self, event: &MrfVerifiedRepairEvent) -> bool {
+        let Some(lease) = event.lease else {
+            return false;
+        };
+        self.kind == event.kind
+            && self.bucket == event.bucket
+            && self.object == event.object
+            && self.version_id == event.version_id
+            && self.scope == event.scope
+            && self.lease == lease
+            && self.bucket_incarnation_id == event.bucket_incarnation_id
+    }
+}
+
+/// Consume only anchors proven by a complete verified-repair identity. The
+/// caller remains responsible for persisting the resulting anchor set before
+/// deleting older replay files.
+pub fn consume_verified_mrf_repair_events(anchors: &mut Vec<MrfDurableRepairAnchor>, events: &[MrfVerifiedRepairEvent]) -> usize {
+    let before = anchors.len();
+    anchors.retain(|anchor| !events.iter().any(|event| anchor.is_proven_by(event)));
+    before.saturating_sub(anchors.len())
+}
+
 #[derive(Clone, Copy, Debug, PartialEq, Eq, Hash)]
 pub struct MrfScope {
     pub pool_index: u32,
@@ -386,6 +441,27 @@ pub fn try_send_mrf_intent_typed(
     }
 }
 
+/// Acquire a fresh process-local lease for one durable replay record.
+///
+/// Journal records deliberately do not persist leases. A replay consumer must
+/// call this before submitting the record so a later verified repair event can
+/// identify the exact replay admission. Replay does not reserve the live
+/// producer coalescer key: the replay queue first deduplicates legacy records,
+/// then manager admission owns task-level deduplication with live producers.
+pub fn try_rearm_mrf_replay_intent(intent: &mut MrfIntent) -> MrfIngressResult {
+    if intent.lease.is_some() {
+        return MrfIngressResult::Enqueued;
+    }
+    if intent.bucket.len() > MRF_MAX_IDENTITY_COMPONENT || intent.object.len() > MRF_MAX_IDENTITY_COMPONENT {
+        return MrfIngressResult::Dropped(MrfDropReason::OversizedIdentity);
+    }
+    let (version_id, scope) = canonical_identity(intent.kind, intent.version_id, intent.scope);
+    intent.version_id = version_id;
+    intent.scope = scope;
+    intent.lease = Some(MrfIngressLease::new(NEXT_MRF_LEASE.fetch_add(1, Ordering::Relaxed)));
+    MrfIngressResult::Enqueued
+}
+
 /// Release the ingress key once the consumer owns the intent.
 pub fn release_mrf_intent(intent: &MrfIntent) {
     release_mrf_identity(intent.kind, &intent.bucket, &intent.object, intent.version_id, intent.scope, intent.lease);
@@ -422,9 +498,9 @@ fn unix_now_ms() -> u64 {
         .unwrap_or(0)
 }
 
-/// A repair the MRF consumer landed, fanned out so retry ledgers can drop
+/// Legacy, unverified repair notice. Its identity lacks kind, set scope,
-/// entries the journal no longer tracks (backlog#1894 axis B). The payload
+/// bucket incarnation and responsibility generation. Consumers must not use
-/// mirrors the intent identity so consumers match without re-parsing.
+/// it to discharge persisted repair responsibility.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub struct MrfRepairedEvent {
     pub bucket: Arc<str>,
@@ -432,15 +508,36 @@ pub struct MrfRepairedEvent {
     pub version_id: Option<[u8; 16]>,
 }
 
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum MrfVerifiedRepairDisposition {
+    Repaired,
+    VerifiedHealthy,
+    AuthoritativelyAbsent,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct MrfVerifiedRepairEvent {
+    pub kind: MrfKind,
+    pub bucket: Arc<str>,
+    pub object: Arc<str>,
+    pub version_id: Option<[u8; 16]>,
+    pub scope: Option<MrfScope>,
+    pub lease: Option<MrfIngressLease>,
+    pub bucket_incarnation_id: Uuid,
+    pub disposition: MrfVerifiedRepairDisposition,
+}
+
 /// Bound on the repaired-event backlog. Notices are best-effort hints; when
 /// the ring is full the oldest are dropped and the affected ledger entries
 /// simply expire through their own attempts/age limits.
 const MRF_REPAIRED_EVENT_CAP: usize = 4096;
 
 static MRF_REPAIRED_EVENTS: OnceLock<std::sync::Mutex<std::collections::VecDeque<MrfRepairedEvent>>> = OnceLock::new();
+static MRF_VERIFIED_REPAIR_EVENTS: OnceLock<std::sync::Mutex<std::collections::VecDeque<MrfVerifiedRepairEvent>>> =
+    OnceLock::new();
 
-/// Record that the MRF consumer landed a repair. Never blocks: the critical
+/// Record a legacy notification for compatibility. This is not an
-/// section is a deque push under a std mutex.
+/// acknowledgement of storage verification or durable repair completion.
 pub fn note_mrf_repaired(bucket: &str, object: &str, version_id: Option<[u8; 16]>) {
     let registry = MRF_REPAIRED_EVENTS.get_or_init(|| std::sync::Mutex::new(std::collections::VecDeque::new()));
     let Ok(mut events) = registry.lock() else {
@@ -478,6 +575,43 @@ pub fn take_mrf_repaired_events_for(bucket: &str) -> Vec<MrfRepairedEvent> {
     taken
 }
 
+/// Record a storage-owned MRF completion proof. Unlike the legacy repaired
+/// event, this identity is complete enough for future durable ledgers to make
+/// an exact responsibility decision.
+pub fn note_mrf_verified_repair(event: MrfVerifiedRepairEvent) {
+    let registry = MRF_VERIFIED_REPAIR_EVENTS.get_or_init(|| std::sync::Mutex::new(std::collections::VecDeque::new()));
+    let Ok(mut events) = registry.lock() else {
+        return;
+    };
+    if events.len() >= MRF_REPAIRED_EVENT_CAP {
+        events.pop_front();
+    }
+    events.push_back(event);
+}
+
+/// Take verified repair events recorded for `bucket`, leaving other buckets'
+/// proofs in place. Consumers still have to match kind, object, version, scope
+/// lease and incarnation before discharging durable responsibility.
+pub fn take_mrf_verified_repair_events_for(bucket: &str) -> Vec<MrfVerifiedRepairEvent> {
+    let Some(registry) = MRF_VERIFIED_REPAIR_EVENTS.get() else {
+        return Vec::new();
+    };
+    let Ok(mut events) = registry.lock() else {
+        return Vec::new();
+    };
+    let mut taken = Vec::new();
+    let mut retained = std::collections::VecDeque::with_capacity(events.len());
+    while let Some(event) = events.pop_front() {
+        if event.bucket.as_ref() == bucket {
+            taken.push(event);
+        } else {
+            retained.push_back(event);
+        }
+    }
+    *events = retained;
+    taken
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -515,6 +649,9 @@ mod tests {
         }
         coalescer_release(&key, Some(lease));
         let retry_lease = coalescer_admit(key.clone()).expect("released identity must admit a retry");
+        assert_ne!(lease, retry_lease);
+        coalescer_release(&key, Some(lease));
+        assert_eq!(coalescer_admit(key.clone()), Err(MrfIngressResult::Coalesced));
         coalescer_release(&key, Some(retry_lease));
     }
 
@@ -543,6 +680,147 @@ mod tests {
         assert_eq!(metadata_scope, None);
     }
 
+    #[test]
+    fn durable_repair_anchor_requires_lease_and_bucket_incarnation() {
+        let mut intent = MrfIntent {
+            bucket: Arc::from("durable-anchor-bucket"),
+            object: Arc::from("object"),
+            version_id: Some([0; 16]),
+            kind: MrfKind::PartialWrite,
+            scope: Some(MrfScope {
+                pool_index: 1,
+                set_index: 2,
+            }),
+            lease: None,
+            enqueued_at_ms: 0,
+            attempts: 0,
+        };
+        assert!(
+            MrfDurableRepairAnchor::from_intent(&intent, Uuid::new_v4()).is_none(),
+            "legacy replay records without the ingress lease must remain anchored"
+        );
+
+        intent.lease = Some(MrfIngressLease::new(7));
+        assert!(
+            MrfDurableRepairAnchor::from_intent(&intent, Uuid::nil()).is_none(),
+            "nil bucket incarnation cannot prove durable successor ownership"
+        );
+
+        let anchor = MrfDurableRepairAnchor::from_intent(&intent, Uuid::new_v4())
+            .expect("complete identity should create a durable repair anchor");
+        assert_eq!(anchor.version_id, None, "nil UUID is canonicalized before matching");
+        assert_eq!(
+            anchor.scope,
+            Some(MrfScope {
+                pool_index: 1,
+                set_index: 2
+            })
+        );
+    }
+
+    #[test]
+    fn durable_replay_rearm_assigns_a_fresh_dischargeable_lease() {
+        let unique = Uuid::new_v4();
+        let mut intent = MrfIntent {
+            bucket: Arc::from(format!("replay-{unique}")),
+            object: Arc::from("object"),
+            version_id: Some([0; 16]),
+            kind: MrfKind::PartialWrite,
+            scope: Some(MrfScope {
+                pool_index: 2,
+                set_index: 3,
+            }),
+            lease: None,
+            enqueued_at_ms: 1,
+            attempts: 0,
+        };
+
+        assert_eq!(try_rearm_mrf_replay_intent(&mut intent), MrfIngressResult::Enqueued);
+        assert_eq!(intent.version_id, None, "nil versions remain canonical during replay");
+        assert!(intent.lease.is_some(), "replay admission must carry a fresh lease");
+        assert!(
+            MrfDurableRepairAnchor::from_intent(&intent, Uuid::new_v4()).is_some(),
+            "a rearmed replay record can participate in exact durable proof matching"
+        );
+        release_mrf_intent(&intent);
+    }
+
+    #[test]
+    fn verified_repair_events_consume_only_exact_durable_anchors() {
+        let bucket = Arc::<str>::from("proof-bucket");
+        let object = Arc::<str>::from("object");
+        let incarnation = Uuid::new_v4();
+        let lease = MrfIngressLease::new(11);
+        let anchor = MrfDurableRepairAnchor {
+            kind: MrfKind::PartialWrite,
+            bucket: bucket.clone(),
+            object: object.clone(),
+            version_id: Some([3; 16]),
+            scope: Some(MrfScope {
+                pool_index: 4,
+                set_index: 5,
+            }),
+            lease,
+            bucket_incarnation_id: incarnation,
+        };
+        let event = MrfVerifiedRepairEvent {
+            kind: anchor.kind,
+            bucket,
+            object,
+            version_id: anchor.version_id,
+            scope: anchor.scope,
+            lease: Some(lease),
+            bucket_incarnation_id: incarnation,
+            disposition: MrfVerifiedRepairDisposition::Repaired,
+        };
+
+        for rejected in [
+            MrfVerifiedRepairEvent {
+                lease: None,
+                ..event.clone()
+            },
+            MrfVerifiedRepairEvent {
+                lease: Some(MrfIngressLease::new(12)),
+                ..event.clone()
+            },
+            MrfVerifiedRepairEvent {
+                bucket_incarnation_id: Uuid::new_v4(),
+                ..event.clone()
+            },
+            MrfVerifiedRepairEvent {
+                version_id: Some([4; 16]),
+                ..event.clone()
+            },
+            MrfVerifiedRepairEvent {
+                scope: Some(MrfScope {
+                    pool_index: 4,
+                    set_index: 6,
+                }),
+                ..event.clone()
+            },
+            MrfVerifiedRepairEvent {
+                kind: MrfKind::DecodeFailure,
+                ..event.clone()
+            },
+            MrfVerifiedRepairEvent {
+                bucket: Arc::from("other-bucket"),
+                ..event.clone()
+            },
+            MrfVerifiedRepairEvent {
+                object: Arc::from("other"),
+                ..event.clone()
+            },
+        ] {
+            let mut retained = vec![anchor.clone()];
+            assert_eq!(consume_verified_mrf_repair_events(&mut retained, &[rejected]), 0);
+            assert_eq!(retained, vec![anchor.clone()]);
+        }
+
+        let mut retained = vec![anchor];
+        assert_eq!(consume_verified_mrf_repair_events(&mut retained, &[event]), 1);
+        assert!(retained.is_empty());
+    }
+
     #[tokio::test]
     async fn try_send_delivers_and_respects_capacity() {
         let mut receiver = init_mrf_channel().expect("first initialization should succeed");
@@ -607,4 +885,32 @@ mod tests {
         assert_eq!(flooded.len(), MRF_REPAIRED_EVENT_CAP);
         assert_eq!(flooded[0].object.as_ref(), "object-9", "the oldest notices past the cap are dropped");
     }
+
+    #[test]
+    fn verified_repair_events_preserve_full_identity_and_bucket_scope() {
+        let bucket_incarnation_id = Uuid::new_v4();
+        let event = MrfVerifiedRepairEvent {
+            kind: MrfKind::PartialWrite,
+            bucket: Arc::from("verified-bucket-a"),
+            object: Arc::from("object-a"),
+            version_id: Some([4u8; 16]),
+            scope: Some(MrfScope {
+                pool_index: 2,
+                set_index: 3,
+            }),
+            lease: Some(MrfIngressLease::new(42)),
+            bucket_incarnation_id,
+            disposition: MrfVerifiedRepairDisposition::Repaired,
+        };
+        note_mrf_verified_repair(event.clone());
+        note_mrf_verified_repair(MrfVerifiedRepairEvent {
+            bucket: Arc::from("verified-bucket-b"),
+            ..event.clone()
+        });
+
+        let taken = take_mrf_verified_repair_events_for("verified-bucket-a");
+        assert_eq!(taken, vec![event]);
+        assert!(take_mrf_verified_repair_events_for("verified-bucket-a").is_empty());
+        assert_eq!(take_mrf_verified_repair_events_for("verified-bucket-b").len(), 1);
+    }
 }

crates/config/README.md

@@ -66,6 +66,11 @@ Current guidance:
 
 - `RUSTFS_BROWSER_REDIRECT_URL` sets the externally reachable browser origin used for OIDC callback, console success redirect, and logout fallback URLs. Configure it to the public scheme and authority without a path, for example `https://console.example.com`. In load-balancer deployments, keep OIDC authorize and callback requests on the same backend node because the in-flight OIDC `state` is local to the RustFS node.
 
+## S3 API environment variables
+
+- `RUSTFS_API_OBJECT_MAX_VERSIONS` caps the number of retained versions for a single object. It defaults to `9223372036854775807`, matching MinIO's practical-unlimited default. Set a positive integer to enforce a lower per-object metadata bound.
+- `MINIO_API_OBJECT_MAX_VERSIONS` is accepted as a compatibility alias when the canonical RustFS variable is not set.
+
 ## Distributed endpoint locality
 
 - `RUSTFS_LOCAL_ENDPOINT_HOST` identifies this server's host in a distributed `RUSTFS_VOLUMES` topology without resolving every peer during startup. Set it to exactly one host, without a scheme, port, or path. It is accepted only for orchestrated URL topologies and must match at least one endpoint on the RustFS server port; invalid or unmatched values fail startup. Leave it unset to retain DNS-based locality discovery.
@@ -130,6 +135,47 @@ Scanner cycle budget controls:
   - timeout returns S3 `SlowDown`, so clients should use normal SDK retry handling.
   - this is not a fdatasync or group-commit switch. Track fdatasync batching separately with `rustfs_s3_put_object_rename_fdatasync_batch_files`.
 
+## Foreground write admission environment variables
+
+Large direct `PutObject` requests and multipart `UploadPart` requests share one
+per-process permit pool that bounds how many bodies are ingested and written
+concurrently. Small direct PUTs stay on the legacy path.
+
+- `RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_ENABLE`
+  - enables the default-on pool; `false` keeps only the soft request counter.
+  - default is `true`.
+- `RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_LIMIT`
+  - permits in the pool; `0` derives half of `RUSTFS_OBJECT_MAX_CONCURRENT_DISK_READS`, clamped to `32`.
+  - default is `0` (32 permits at stock settings).
+- `RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_MIN_SIZE_BYTES`
+  - smallest direct `PutObject` that takes a permit; unknown-size requests always do.
+  - default is `33554432` (32 MiB).
+- `RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS`
+  - how long a direct `PutObject` waits for a permit before returning S3 `SlowDown`.
+  - default is `250`.
+- `RUSTFS_PUT_MULTIPART_FOREGROUND_ADMISSION_MIN_SIZE_BYTES`
+  - smallest `UploadPart` that takes a permit; `0` gates every part.
+  - default is `0`.
+- `RUSTFS_PUT_MULTIPART_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS`
+  - how long an `UploadPart` waits in the bounded queue for a permit before returning S3 `SlowDown`; `0` rejects immediately when the pool is full.
+  - default is `10000`. Parts wait before body ingest, so SDK-default clients that send every part of an upload concurrently drain through the pool instead of failing on a full pool.
+  - RustFS does not read the request body while a part is queued, so the client's socket write stalls for the whole wait and whatever timeout the client or an intermediary has configured competes with this value. Keep it with margin below the shortest such timeout in use (botocore applies its 60 s `connect_timeout` to the body write; the AWS SDK for Java v2 has a 30 s socket write timeout; reverse proxies add their own body timeouts); a wait that outlives the client timeout surfaces as a dropped connection instead of `SlowDown`.
+- `RUSTFS_PUT_MULTIPART_FOREGROUND_ADMISSION_MAX_PENDING`
+  - maximum `UploadPart` requests waiting for a permit at once; parts beyond it return `SlowDown` without waiting.
+  - default is `0`, which derives 16 times the permit limit (512 at stock settings).
+  - each queued HTTP/1 part holds whatever unread body the client already pushed into the connection's kernel receive buffer (an HTTP/2 part holds up to its flow-control window in process memory), so this depth also bounds that memory. RustFS leaves the receive buffer to kernel autotuning (see `RUSTFS_HTTP_SOCKET_RECV_BUFFER_BYTES` below), which keeps an unread connection at the kernel's initial size (128 KiB on current Linux).
+- `RUSTFS_PUT_FOREGROUND_ADMISSION_ENABLE`, `RUSTFS_PUT_FOREGROUND_ADMISSION_LIMIT`, `RUSTFS_PUT_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS`
+  - experimental strict gate that applies to every foreground write regardless of size and replaces the pool above when enabled.
+  - default is disabled; enabling it with limit `0` disables foreground write admission entirely.
+
+## HTTP listener socket environment variables
+
+- `RUSTFS_HTTP_SOCKET_RECV_BUFFER_BYTES`
+  - fixed `SO_RCVBUF` for the API listener, inherited by every accepted socket; `0` leaves the receive buffer to kernel autotuning.
+  - default is `0`. Earlier releases hard-coded 4 MiB, which Linux doubles to 8 MiB and which disables autotuning, so every connection whose body was not being read yet (a multipart part queued for a foreground write permit) could accumulate up to 8 MiB of unread body in kernel memory; at SDK-default multipart concurrency that was enough to push a node into TCP memory pressure.
+  - with autotuning the per-connection receive ceiling is the kernel's (`net.ipv4.tcp_rmem` max, 6 MiB on stock Linux) instead of the former fixed 8 MiB, so a single very high-bandwidth-delay connection may see a somewhat lower ceiling; raise `net.ipv4.tcp_rmem` first, and set this variable only on kernels without receive-buffer autotuning (illumos/Solaris) or where the sysctl cannot be changed.
+  - the send buffer stays fixed at 4 MiB because the stock Linux send autotuning ceiling (`net.ipv4.tcp_wmem` max, 4 MiB) is lower than a GB-level response stream needs.
+
 ## Remote tier timeout environment variables
 
 - `RUSTFS_TIER_REMOTE_CONNECT_TIMEOUT_SECS`

crates/config/src/constants/api.rs

@@ -90,3 +90,20 @@ pub const ENV_API_MAX_CONNECTIONS: &str = "RUSTFS_API_MAX_CONNECTIONS";
 
 /// Default for `RUSTFS_API_MAX_CONNECTIONS` (`0` = unlimited).
 pub const DEFAULT_API_MAX_CONNECTIONS: usize = 0;
+
+/// Maximum retained versions per object.
+///
+/// The default follows MinIO and is effectively unlimited for practical
+/// deployments. Operators can lower it to bound per-object metadata growth.
+/// Environment variable: RUSTFS_API_OBJECT_MAX_VERSIONS
+/// MinIO-compatible alias: MINIO_API_OBJECT_MAX_VERSIONS
+/// Example: RUSTFS_API_OBJECT_MAX_VERSIONS=50000
+pub const ENV_API_OBJECT_MAX_VERSIONS: &str = "RUSTFS_API_OBJECT_MAX_VERSIONS";
+
+/// Default and maximum accepted value for `RUSTFS_API_OBJECT_MAX_VERSIONS`.
+#[cfg(target_pointer_width = "64")]
+pub const DEFAULT_API_OBJECT_MAX_VERSIONS: usize = 9_223_372_036_854_775_807;
+
+/// Default and maximum accepted value for `RUSTFS_API_OBJECT_MAX_VERSIONS`.
+#[cfg(not(target_pointer_width = "64"))]
+pub const DEFAULT_API_OBJECT_MAX_VERSIONS: usize = usize::MAX;

crates/config/src/constants/object.rs

@@ -365,13 +365,54 @@ pub const ENV_PUT_MULTIPART_FOREGROUND_ADMISSION_MIN_SIZE_BYTES: &str =
     "RUSTFS_PUT_MULTIPART_FOREGROUND_ADMISSION_MIN_SIZE_BYTES";
 pub const DEFAULT_PUT_MULTIPART_FOREGROUND_ADMISSION_MIN_SIZE_BYTES: usize = 0;
 
-/// Time in milliseconds an automatic foreground write waits for a permit.
+/// Time in milliseconds an automatic foreground direct PutObject waits for a permit.
 ///
 /// A short wait smooths transient bursts while still returning S3
 /// `SlowDown`/503 before body ingest when the node is already saturated.
 pub const ENV_PUT_LARGE_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS: &str = "RUSTFS_PUT_LARGE_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS";
 pub const DEFAULT_PUT_LARGE_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS: u64 = 250;
 
+/// Time in milliseconds a multipart UploadPart waits for a foreground write permit.
+///
+/// SDK-default multipart clients send every part of an upload concurrently, so
+/// a single node routinely sees several times more parts in flight than the
+/// permit pool allows. A queued part waits before body ingest, so the pool
+/// still bounds the number of parts being written, but the wait is not free:
+/// RustFS does not read the request body while the part is queued (hyper only
+/// sends `100 Continue` once the body is first polled, and the AWS SDKs send
+/// the body after a 1-3 s `Expect: 100-continue` grace anyway), so the
+/// client's socket write stalls once the kernel buffers fill, and whatever
+/// timeout the client or an intermediary has configured decides the outcome.
+/// botocore applies its `connect_timeout` (60 s) to the body write, the AWS
+/// SDK for Java v2 has a 30 s socket write timeout, and MinIO bounds the same
+/// wait with a 10 s request deadline. The wait must leave margin under the
+/// shortest of those, not merely fall below an SDK default, so the part
+/// receives S3 `SlowDown`/503 for the client to retry instead of losing its
+/// connection (issue #7385). `0` rejects immediately when the pool is full.
+pub const ENV_PUT_MULTIPART_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS: &str =
+    "RUSTFS_PUT_MULTIPART_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS";
+pub const DEFAULT_PUT_MULTIPART_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS: u64 = 10_000;
+
+// A queued part holds the client's body write open for the whole wait. The
+// shortest write timeout among mainstream S3 SDKs is the AWS SDK for Java v2's
+// 30 s socket write timeout; keep the compiled default at no more than a third
+// of it. This locks only the default; the environment variable may still raise
+// the wait past any client timeout.
+const _: () = assert!(DEFAULT_PUT_MULTIPART_FOREGROUND_ADMISSION_WAIT_TIMEOUT_MS * 3 <= 30_000);
+
+/// Maximum multipart UploadPart requests waiting for a foreground write permit per process.
+///
+/// Parts beyond this queue depth are rejected with S3 `SlowDown`/503 without
+/// waiting, so a genuinely saturated node still fails fast instead of holding
+/// an unbounded set of connections open for the whole wait timeout. Each
+/// queued HTTP/1 part also holds whatever unread body the client already
+/// pushed into that connection's kernel receive buffer, and a queued HTTP/2
+/// part holds up to its flow-control window in process memory, so the depth
+/// bounds socket and window memory as well as connections.
+/// `0` derives the depth from the permit limit.
+pub const ENV_PUT_MULTIPART_FOREGROUND_ADMISSION_MAX_PENDING: &str = "RUSTFS_PUT_MULTIPART_FOREGROUND_ADMISSION_MAX_PENDING";
+pub const DEFAULT_PUT_MULTIPART_FOREGROUND_ADMISSION_MAX_PENDING: usize = 0;
+
 const _: () = assert!(DEFAULT_PUT_LARGE_FOREGROUND_ADMISSION_ENABLE);
 
 /// Environment variable for minimum GetObject timeout in seconds.
@@ -568,6 +609,36 @@ pub const ENV_OBJECT_LOCK_RPC_TIMEOUT_MS: &str = "RUSTFS_OBJECT_LOCK_RPC_TIMEOUT
 /// Default remote lock RPC transport timeout: 3000 milliseconds.
 pub const DEFAULT_OBJECT_LOCK_RPC_TIMEOUT_MS: u64 = 3000;
 
+/// Environment variable for the minimum interval between evictions of the
+/// cached lock RPC channel to one peer, in milliseconds.
+///
+/// A lock RPC that fails on transport, or that times out while the peer has
+/// not completed any lock RPC for two deadlines, evicts the shared HTTP/2
+/// channel so the next request re-dials. Evictions are rate limited per peer
+/// so one slow lock endpoint cannot drive a reset/GOAWAY/reconnect loop
+/// (issue #7363). `0` disables the cooldown.
+///
+/// Default: 5000 milliseconds.
+pub const ENV_OBJECT_LOCK_RPC_EVICTION_COOLDOWN_MS: &str = "RUSTFS_OBJECT_LOCK_RPC_EVICTION_COOLDOWN_MS";
+
+/// Default minimum interval between lock RPC channel evictions per peer: 5000 milliseconds.
+pub const DEFAULT_OBJECT_LOCK_RPC_EVICTION_COOLDOWN_MS: u64 = 5000;
+
+/// Environment variable for how many timed-out lock RPCs per peer may keep
+/// running in the background instead of being cancelled.
+///
+/// Cancelling a timed-out stream sends `RST_STREAM`; enough of them make the
+/// peer answer `GOAWAY too_many_resets` and drop every stream on the
+/// connection. A detached RPC ends on its own within the internode RPC
+/// timeout, and a lock it acquires after its caller gave up is released
+/// immediately. Beyond this budget timed-out RPCs are cancelled as before.
+///
+/// Default: 256.
+pub const ENV_OBJECT_LOCK_RPC_DETACHED_LIMIT: &str = "RUSTFS_OBJECT_LOCK_RPC_DETACHED_LIMIT";
+
+/// Default per-peer budget of detached (timed-out but still running) lock RPCs: 256.
+pub const DEFAULT_OBJECT_LOCK_RPC_DETACHED_LIMIT: usize = 256;
+
 /// Environment variable to enable object namespace lock diagnostics.
 ///
 /// When enabled, RustFS emits slow lock acquisition and long lock hold

crates/config/src/constants/tls.rs

@@ -159,6 +159,24 @@ pub const DEFAULT_HTTP1_HEADER_READ_TIMEOUT: u64 = 75;
 pub const ENV_HTTP1_MAX_BUF_SIZE: &str = "RUSTFS_HTTP1_MAX_BUF_SIZE";
 pub const DEFAULT_HTTP1_MAX_BUF_SIZE: usize = 64 * 1024; // 64 KB
 
+/// Environment variable for a fixed kernel receive buffer (`SO_RCVBUF`, bytes)
+/// on the API listener. Default: 0, which leaves the buffer to kernel
+/// autotuning.
+///
+/// A fixed `SO_RCVBUF` is inherited by every accepted socket and disables
+/// receive-buffer autotuning, so a connection whose request body is not being
+/// read yet (a multipart part queued for a foreground write permit) lets up to
+/// the fixed size of unread body accumulate in kernel memory — Linux doubles
+/// the requested value, so the former hard-coded 4 MiB held up to 8 MiB per
+/// queued connection (issue #7385). Autotuning keeps an unread connection at
+/// the kernel's initial size and grows only connections that are being
+/// drained. Set this only on kernels without receive-buffer autotuning
+/// (illumos/Solaris) or on very high-bandwidth-delay links where the kernel's
+/// autotuning ceiling (`net.ipv4.tcp_rmem` on Linux) is too low and cannot be
+/// raised.
+pub const ENV_HTTP_SOCKET_RECV_BUFFER_BYTES: &str = "RUSTFS_HTTP_SOCKET_RECV_BUFFER_BYTES";
+pub const DEFAULT_HTTP_SOCKET_RECV_BUFFER_BYTES: usize = 0;
+
 /// Environment variable for the S3 request-body inter-chunk read timeout
 /// (seconds). Default: 300. Set to 0 to disable.
 ///

crates/e2e_test/README.md

@@ -184,6 +184,8 @@ the wiring source of truth. Committed test-ID digests under
 
 ## Troubleshooting
 
+**Endpoint blackhole scenario skipped** — `heal_erasure_disk_rebuild_test::tests::test_cluster_root_heal_recovers_after_target_endpoint_blackhole` installs a loopback `iptables` DROP rule and therefore needs `CAP_NET_ADMIN` (root or passwordless `sudo -n iptables`). A host where `iptables` is missing or cannot read the OUTPUT chain (typical inside an unprivileged container, where the nf_tables backend reports "Permission denied" even under `sudo`) logs a `heal_interruption_skipped` warning and returns without exercising heal. Set `RUSTFS_E2E_REQUIRE_NET_FAULT_INJECTION=1` on lanes that do provision the capability so a broken runner fails instead of skipping.
+
 **Reproduce a CI failure locally** — run the exact profile/lane:
 
 ```bash

crates/e2e_test/src/admin_timeout_regression_test.rs

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use crate::common::{RustFSTestClusterEnvironment, init_logging, local_http_client};
+use crate::common::{RustFSTestClusterEnvironment, init_logging, local_http_client, signal_process};
 use aws_sdk_s3::primitives::ByteStream;
 use http::header::HOST;
 use reqwest::StatusCode;
@@ -22,7 +22,6 @@ use rustfs_signer::sign_v4;
 use s3s::Body;
 use serde::Deserialize;
 use std::error::Error;
-use std::process::Command;
 use tokio::time::{Duration, sleep, timeout};
 use uuid::Uuid;
 
@@ -82,15 +81,6 @@ async fn parse_json_response<T: serde::de::DeserializeOwned>(
     Ok(serde_json::from_slice(&body)?)
 }
 
-fn signal_process(pid: u32, signal: &str) -> Result<(), Box<dyn Error + Send + Sync>> {
-    let output = Command::new("kill").arg(format!("-{signal}")).arg(pid.to_string()).output()?;
-    if output.status.success() {
-        return Ok(());
-    }
-
-    Err(format!("kill -{signal} {pid} failed: {}", String::from_utf8_lossy(&output.stderr)).into())
-}
-
 fn offline_server_count(info: &InfoMessage) -> usize {
     info.servers
         .as_ref()

crates/e2e_test/src/common.rs

@@ -57,6 +57,8 @@ const RUSTFS_FULL_FEATURE: &str = "full";
 const TEST_PORT_MIN: u16 = 20_000;
 // Keep allocator ports below the ephemeral range used by bind(..., 0) test helpers.
 const TEST_PORT_RANGE: u16 = 10_000;
+const TEST_PORT_MIN_ENV: &str = "RUSTFS_E2E_TEST_PORT_MIN";
+const TEST_PORT_RANGE_ENV: &str = "RUSTFS_E2E_TEST_PORT_RANGE";
 const TEST_PORT_COUNTER_PATH: &str = "/tmp/rustfs_e2e_next_port";
 const TEST_PORT_LOCK_DIR: &str = "/tmp/rustfs_e2e_port_allocator.lock";
 const TEST_PORT_LOCK_STALE_AFTER: Duration = Duration::from_secs(30);
@@ -99,22 +101,74 @@ impl Drop for PortAllocatorGuard {
     }
 }
 
-fn advance_test_port(port: u16) -> u16 {
+#[derive(Clone, Copy, Debug, Eq, PartialEq)]
-    let offset = (port - TEST_PORT_MIN + 1) % TEST_PORT_RANGE;
+struct TestPortAllocatorConfig {
-    TEST_PORT_MIN + offset
+    min: u16,
+    range: u16,
 }
 
-fn seeded_test_port() -> u16 {
+impl TestPortAllocatorConfig {
-    let offset = (Uuid::new_v4().as_u128() % u128::from(TEST_PORT_RANGE)) as u16;
+    fn max_exclusive(self) -> u32 {
-    TEST_PORT_MIN + offset
+        u32::from(self.min) + u32::from(self.range)
+    }
+
+    fn contains(self, port: &u16) -> bool {
+        (u32::from(self.min)..self.max_exclusive()).contains(&u32::from(*port))
+    }
 }
 
-fn read_next_test_port() -> u16 {
+fn parse_test_port_allocator_config(
+    min_override: Option<&str>,
+    range_override: Option<&str>,
+) -> Result<TestPortAllocatorConfig, Box<dyn std::error::Error + Send + Sync>> {
+    let min = match min_override {
+        Some(value) => value
+            .parse::<u16>()
+            .map_err(|err| format!("{TEST_PORT_MIN_ENV} must be a valid u16: {err}"))?,
+        None => TEST_PORT_MIN,
+    };
+    let range = match range_override {
+        Some(value) => value
+            .parse::<u16>()
+            .map_err(|err| format!("{TEST_PORT_RANGE_ENV} must be a valid u16: {err}"))?,
+        None => TEST_PORT_RANGE,
+    };
+    if range == 0 {
+        return Err(format!("{TEST_PORT_RANGE_ENV} must be greater than zero").into());
+    }
+    if min < 1024 {
+        return Err(format!("{TEST_PORT_MIN_ENV} must be at least 1024").into());
+    }
+    let max_exclusive = u32::from(min) + u32::from(range);
+    if max_exclusive > u32::from(u16::MAX) + 1 {
+        return Err(format!("{TEST_PORT_MIN_ENV} + {TEST_PORT_RANGE_ENV} exceeds u16 port space").into());
+    }
+    Ok(TestPortAllocatorConfig { min, range })
+}
+
+fn test_port_allocator_config() -> Result<TestPortAllocatorConfig, Box<dyn std::error::Error + Send + Sync>> {
+    parse_test_port_allocator_config(
+        std::env::var(TEST_PORT_MIN_ENV).ok().as_deref(),
+        std::env::var(TEST_PORT_RANGE_ENV).ok().as_deref(),
+    )
+}
+
+fn advance_test_port(port: u16, config: TestPortAllocatorConfig) -> u16 {
+    let offset = (port - config.min + 1) % config.range;
+    config.min + offset
+}
+
+fn seeded_test_port(config: TestPortAllocatorConfig) -> u16 {
+    let offset = (Uuid::new_v4().as_u128() % u128::from(config.range)) as u16;
+    config.min + offset
+}
+
+fn read_next_test_port(config: TestPortAllocatorConfig) -> u16 {
     stdfs::read_to_string(TEST_PORT_COUNTER_PATH)
         .ok()
         .and_then(|value| value.trim().parse::<u16>().ok())
-        .filter(|port| (TEST_PORT_MIN..TEST_PORT_MIN + TEST_PORT_RANGE).contains(port))
+        .filter(|port| config.contains(port))
-        .unwrap_or_else(seeded_test_port)
+        .unwrap_or_else(|| seeded_test_port(config))
 }
 
 fn remove_stale_port_allocator_lock() {
@@ -210,6 +264,15 @@ pub fn local_http_client() -> HttpClient {
         .expect("failed to build local reqwest client")
 }
 
+pub(crate) fn signal_process(pid: u32, signal: &str) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let output = Command::new("kill").arg(format!("-{signal}")).arg(pid.to_string()).output()?;
+    if output.status.success() {
+        return Ok(());
+    }
+
+    Err(format!("kill -{signal} {pid} failed: {}", String::from_utf8_lossy(&output.stderr)).into())
+}
+
 pub(crate) async fn signed_s3_request(
     method: http::Method,
     url: &str,
@@ -629,11 +692,12 @@ impl RustFSTestEnvironment {
     pub async fn find_available_port() -> Result<u16, Box<dyn std::error::Error + Send + Sync>> {
         use std::net::TcpListener;
         let _guard = PortAllocatorGuard::acquire().await?;
-        let mut next_port = read_next_test_port();
+        let config = test_port_allocator_config()?;
+        let mut next_port = read_next_test_port(config);
 
-        for _ in 0..TEST_PORT_RANGE {
+        for _ in 0..config.range {
             let port = next_port;
-            next_port = advance_test_port(next_port);
+            next_port = advance_test_port(next_port, config);
             write_next_test_port(next_port)?;
 
             if let Ok(listener) = TcpListener::bind(("127.0.0.1", port)) {
@@ -2108,6 +2172,35 @@ mod tests {
         );
     }
 
+    #[test]
+    fn e2e_port_allocator_uses_default_range() {
+        assert_eq!(
+            parse_test_port_allocator_config(None, None).expect("default port allocator config"),
+            TestPortAllocatorConfig {
+                min: TEST_PORT_MIN,
+                range: TEST_PORT_RANGE
+            }
+        );
+    }
+
+    #[test]
+    fn e2e_port_allocator_accepts_explicit_test_range() {
+        let config = parse_test_port_allocator_config(Some("31000"), Some("128")).expect("explicit port range");
+
+        assert_eq!(advance_test_port(31127, config), 31000);
+        assert!(config.contains(&31000));
+        assert!(config.contains(&31127));
+        assert!(!config.contains(&31128));
+    }
+
+    #[test]
+    fn e2e_port_allocator_rejects_invalid_override() {
+        assert!(parse_test_port_allocator_config(Some("1023"), Some("1")).is_err());
+        assert!(parse_test_port_allocator_config(Some("65000"), Some("1000")).is_err());
+        assert!(parse_test_port_allocator_config(Some("31000"), Some("0")).is_err());
+        assert!(parse_test_port_allocator_config(Some("not-a-port"), Some("128")).is_err());
+    }
+
     #[test]
     fn resolves_rustfs_binary_in_configured_cargo_target_directory() {
         let workspace = Path::new("workspace");

crates/e2e_test/src/distributed/observability_test.rs

@@ -12,8 +12,10 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use super::harness::{DistCluster, DistLayout, TestResult, cluster_admin_ok, unique_bucket, wait_for_ready};
+use super::harness::{
-use crate::common::{admin_request, init_logging, local_http_client};
+    DistCluster, DistLayout, TestResult, assert_object_bytes, cluster_admin_ok, unique_bucket, wait_for_ready, wait_until,
+};
+use crate::common::{admin_request, init_logging, local_http_client, signal_process, signed_request};
 use aws_sdk_s3::operation::RequestId;
 use aws_sdk_s3::primitives::ByteStream;
 use bytes::Bytes;
@@ -24,7 +26,7 @@ use hyper::service::service_fn;
 use hyper::{Request, Response};
 use hyper_util::rt::TokioIo;
 use local_ip_address::local_ip;
-use rustfs_madmin::metrics::RealtimeMetrics;
+use rustfs_madmin::metrics::{HttpMetrics, RealtimeMetrics};
 use rustfs_utils::egress::ENV_OUTBOUND_ALLOW_ORIGINS;
 use serde_json::Value;
 use std::convert::Infallible;
@@ -126,6 +128,7 @@ async fn four_node_health_inventory_metrics_and_audit_delivery_are_consistent()
     let (audit_endpoint, mut audit_entries, collector) = spawn_audit_collector().await?;
     let audit_origin = reqwest::Url::parse(&audit_endpoint)?.origin().ascii_serialization();
     let audit_env = [
+        ("RUST_LOG", "warn"),
         ("RUSTFS_AUDIT_ENABLE", "true"),
         ("RUSTFS_AUDIT_WEBHOOK_ENABLE_DISTRIBUTED", "on"),
         ("RUSTFS_AUDIT_WEBHOOK_ENDPOINT_DISTRIBUTED", audit_endpoint.as_str()),
@@ -231,6 +234,186 @@ async fn four_node_health_inventory_metrics_and_audit_delivery_are_consistent()
         "audit entry leaked the root secret key"
     );
 
+    let result = verify_write_observations_during_peer_failure(&dist, &bucket).await;
     collector.abort();
+    result
+}
+
+async fn node_admin_body(dist: &DistCluster, node: usize, path: &str) -> TestResult<String> {
+    let (status, body) = timeout(
+        Duration::from_secs(30),
+        admin_request(
+            &dist.cluster.nodes[node].url,
+            Method::GET,
+            path,
+            None,
+            &dist.cluster.access_key,
+            &dist.cluster.secret_key,
+        ),
+    )
+    .await??;
+    assert!(status.is_success(), "node {node} admin request {path}: {status} {body}");
+    Ok(body)
+}
+
+async fn http_put_counts(dist: &DistCluster, node: usize) -> TestResult<[u64; 2]> {
+    let body = node_admin_body(dist, node, "/rustfs/admin/v3/metrics?types=512&by-host=true&n=1").await?;
+    let sample: RealtimeMetrics = serde_json::from_str(body.lines().next().ok_or("empty HTTP metrics stream")?)?;
+    assert!(sample.errors.is_empty(), "HTTP metrics returned errors: {:?}", sample.errors);
+    let http = sample.aggregated.http.ok_or("HTTP metrics missing at WARN log level")?;
+    let count = |http: &HttpMetrics, outcome: &str| {
+        http.requests
+            .iter()
+            .filter(|row| row.method == "PUT" && row.outcome == outcome)
+            .map(|row| row.total)
+            .sum::<u64>()
+    };
+    assert_eq!(sample.by_host.len(), 1, "HTTP admin metrics must remain node-local");
+    let host = sample.by_host.values().next().expect("one reporting host");
+    let host = host.http.as_ref().ok_or("by-host HTTP metrics missing")?;
+    let totals = [count(&http, "2xx"), count(&http, "5xx")];
+    assert_eq!(totals, [count(host, "2xx"), count(host, "5xx")]);
+    Ok(totals)
+}
+
+async fn observed_put(dist: &DistCluster, node: usize, bucket: &str, key: &str) -> TestResult<http::StatusCode> {
+    // One signed HTTP attempt: SDK retries must not change the expected denominator.
+    timeout(Duration::from_secs(90), async {
+        let response = signed_request(
+            Method::PUT,
+            &format!("{}/{bucket}/{key}", dist.cluster.nodes[node].url),
+            &dist.cluster.access_key,
+            &dist.cluster.secret_key,
+            Some(b"write-observation".to_vec()),
+            Some("application/octet-stream"),
+        )
+        .await?;
+        assert!(response.headers().contains_key("x-amz-request-id"), "PUT omitted correlation ID");
+        let status = response.status();
+        let body = response.text().await?;
+        assert!(!body.contains(&dist.cluster.secret_key), "PUT response leaked credentials");
+        Ok::<_, Box<dyn std::error::Error + Send + Sync>>(status)
+    })
+    .await?
+}
+
+struct SuspendedPeer<'a> {
+    // Borrowing the owned child keeps its PID from being reaped/reused before cleanup.
+    child: &'a std::process::Child,
+    suspended: bool,
+}
+
+impl<'a> SuspendedPeer<'a> {
+    fn suspend(dist: &'a DistCluster, node: usize) -> TestResult<Self> {
+        let child = dist.cluster.nodes[node].process.as_ref().ok_or("peer process missing")?;
+        signal_process(child.id(), "STOP")?;
+        Ok(Self { child, suspended: true })
+    }
+
+    fn resume(&mut self) -> TestResult {
+        signal_process(self.child.id(), "CONT")?;
+        self.suspended = false;
+        Ok(())
+    }
+}
+
+impl Drop for SuspendedPeer<'_> {
+    fn drop(&mut self) {
+        if self.suspended {
+            let _ = signal_process(self.child.id(), "CONT");
+        }
+    }
+}
+
+async fn verify_write_observations_during_peer_failure(dist: &DistCluster, bucket: &str) -> TestResult {
+    let mut baseline = Vec::new();
+    for node in 0..dist.cluster.nodes.len() {
+        let before = http_put_counts(dist, node).await?;
+        assert!(
+            observed_put(dist, node, bucket, &format!("healthy-{node}"))
+                .await?
+                .is_success()
+        );
+        let after = http_put_counts(dist, node).await?;
+        assert_eq!(after, [before[0] + 1, before[1]], "node {node} lost its successful PUT denominator");
+        baseline.push(after);
+    }
+
+    // Refresh provenance immediately before the first failed probe, within the cache age budget.
+    node_admin_body(dist, 0, "/rustfs/admin/v3/storageinfo").await?;
+    let mut suspended = [SuspendedPeer::suspend(dist, 2)?, SuspendedPeer::suspend(dist, 3)?];
+    let storage: Value = serde_json::from_str(&node_admin_body(dist, 0, "/rustfs/admin/v3/storageinfo").await?)?;
+    let observations = storage["info"]["observations"]
+        .as_array()
+        .ok_or("storageinfo omitted observations")?;
+    let disks = storage["info"]["disks"]
+        .as_array()
+        .ok_or("storageinfo omitted disks during peer failure")?;
+    assert_eq!(disks.len(), 16, "failed peers must not vanish from inventory");
+    for node in [2, 3] {
+        let endpoint = &dist.cluster.nodes[node].address;
+        let observation = observations
+            .iter()
+            .find(|item| item["endpoint"].as_str().is_some_and(|value| value.contains(endpoint)))
+            .ok_or_else(|| format!("missing failed peer observation {endpoint}: {storage}"))?;
+        assert_eq!(observation["status"], "failed", "suspension did not affect peer RPC: {observation}");
+        assert_eq!(observation["cached"], true, "first failure must identify the warm cache: {observation}");
+        assert!(observation["last_success_unix_millis"].as_u64().is_some());
+        assert!(observation["snapshot_age_seconds"].as_u64().is_some_and(|age| age < 60));
+        let peer_disks: Vec<_> = disks
+            .iter()
+            .filter(|disk| disk["endpoint"].as_str().is_some_and(|value| value.contains(endpoint)))
+            .collect();
+        assert_eq!(peer_disks.len(), 4, "failed peer lost its four drive identities: {storage}");
+        for disk in peer_disks {
+            assert_eq!(disk["state"], "unknown");
+            assert_eq!(disk["runtimeState"], "unknown");
+        }
+    }
+
+    let snapshot: Value = serde_json::from_str(&node_admin_body(dist, 0, "/rustfs/admin/v4/cluster/snapshot").await?)?;
+    let metadata = &snapshot["snapshot"]["pool_meta_write_gate"];
+    assert_eq!(
+        metadata["state"], "writable",
+        "peer probe failure must not invent a metadata latch: {snapshot}"
+    );
+    assert!(metadata.get("sinceUnixSecs").is_none());
+
+    for attempt in 0..2 {
+        let status = observed_put(dist, 0, bucket, &format!("unavailable-{attempt}")).await?;
+        assert!(status.is_server_error(), "sub-quorum write unexpectedly returned {status}");
+    }
+    assert_eq!(http_put_counts(dist, 0).await?, [baseline[0][0], baseline[0][1] + 2]);
+    assert_eq!(
+        http_put_counts(dist, 1).await?,
+        baseline[1],
+        "internal RPCs must not count as external PUTs"
+    );
+
+    for peer in &mut suspended {
+        peer.resume()?;
+    }
+    wait_until(
+        Duration::from_secs(90),
+        || async {
+            let storage: Value = serde_json::from_str(&node_admin_body(dist, 0, "/rustfs/admin/v3/storageinfo").await?)?;
+            let observations = storage["info"]["observations"]
+                .as_array()
+                .ok_or("recovery omitted observations")?;
+            Ok(observations.len() == 4
+                && observations
+                    .iter()
+                    .all(|item| item["status"] == "succeeded" && item["cached"] == false))
+        },
+        "peer probes recover to fresh successful observations",
+    )
+    .await?;
+    wait_for_ready(&dist.cluster).await?;
+    assert!(observed_put(dist, 0, bucket, "recovered").await?.is_success());
+    assert_eq!(http_put_counts(dist, 0).await?, [baseline[0][0] + 1, baseline[0][1] + 2]);
+    for node in 0..dist.cluster.nodes.len() {
+        assert_object_bytes(&dist.client(node)?, bucket, "healthy-0", b"write-observation").await?;
+        assert_object_bytes(&dist.client(node)?, bucket, "recovered", b"write-observation").await?;
+    }
     Ok(())
 }

crates/e2e_test/src/distributed_startup_regression_test.rs

@@ -424,10 +424,10 @@ mod tests {
                 std::io::Error::other("PRECONDITION: provide the existing hooks binary; this fixture never invokes Cargo")
             })?;
         let binary = std::fs::canonicalize(explicit)?;
-        if let Some(other) = std::env::var_os("CARGO_BIN_EXE_rustfs") {
+        if let Some(other) = std::env::var_os("CARGO_BIN_EXE_rustfs")
-            if binary != std::fs::canonicalize(other)? {
+            && binary != std::fs::canonicalize(other)?
-                return Err(std::io::Error::other("PRECONDITION: conflicting startup binary paths"));
+        {
-            }
+            return Err(std::io::Error::other("PRECONDITION: conflicting startup binary paths"));
         }
         let manifest_path = std::env::var_os("RUSTFS_E2E_STARTUP_CAS_BUILD_MANIFEST")
             .ok_or_else(|| std::io::Error::other("PRECONDITION: missing hooks binary build manifest"))?;
@@ -509,11 +509,11 @@ mod tests {
         .await
         .map_err(|_| std::io::Error::other("PRECONDITION: binary capability probe timed out"))??;
         let records = startup_cas_log(&path)?;
-        let matching: Vec<_> = records
+        let matching_count = records
             .iter()
             .filter(|r| r["nonce"] == nonce && r["kind"] == "capability" && r["schema"] == "fresh-startup-cas/v1")
-            .collect();
+            .count();
-        if !status.success() || matching.len() != 1 {
+        if !status.success() || matching_count != 1 {
             return Err(std::io::Error::other(
                 "PRECONDITION: binary lacks the startup CAS hooks; no cluster was started",
             ));
@@ -1257,7 +1257,7 @@ mod tests {
                 assert_eq!(commits[0]["payload_sha256"], commits[1]["payload_sha256"]);
             }
             let mut replicas = Vec::new();
-            for pool in 0..pool_count {
+            for (pool, commit) in commits.iter().enumerate() {
                 let matching: Vec<_> = source
                     .iter()
                     .copied()
@@ -1266,7 +1266,7 @@ mod tests {
                             && event["startup_phase"] == "persist"
                             && event["attempt"] == *attempt
                             && event["pool"] == pool
-                            && event["payload_sha256"] == commits[pool]["payload_sha256"]
+                            && event["payload_sha256"] == commit["payload_sha256"]
                     })
                     .collect();
                 assert_eq!(matching.len(), 1, "actual final complete decoded read for pool {pool}");
@@ -1275,7 +1275,7 @@ mod tests {
                 assert_eq!(replica["committed"], true);
                 assert_eq!(replica["pool_count"], pool_count);
                 assert_eq!(replica["raw_sha256"], replica["payload_sha256"]);
-                assert_eq!(replica["etag"], commits[pool]["etag"]);
+                assert_eq!(replica["etag"], commit["etag"]);
                 assert_eq!(replica["cas"], "existing");
                 assert!(
                     replica["cluster_id"]

crates/e2e_test/src/fake_s3_target/mod.rs

@@ -34,12 +34,14 @@ use s3s::dto::{
     AbortMultipartUploadInput, AbortMultipartUploadOutput, CommonPrefix, CompleteMultipartUploadInput,
     CompleteMultipartUploadOutput, CreateMultipartUploadInput, CreateMultipartUploadOutput, DeleteMarkerEntry, DeleteObjectInput,
     DeleteObjectOutput, DeleteObjectTaggingInput, DeleteObjectTaggingOutput, ETag, GetBucketVersioningInput,
-    GetBucketVersioningOutput, GetObjectInput, GetObjectLockConfigurationInput, GetObjectLockConfigurationOutput,
+    GetBucketVersioningOutput, GetObjectInput, GetObjectLegalHoldInput, GetObjectLegalHoldOutput,
-    GetObjectOutput, GetObjectTaggingInput, GetObjectTaggingOutput, HeadBucketInput, HeadBucketOutput, HeadObjectInput,
+    GetObjectLockConfigurationInput, GetObjectLockConfigurationOutput, GetObjectOutput, GetObjectRetentionInput,
+    GetObjectRetentionOutput, GetObjectTaggingInput, GetObjectTaggingOutput, HeadBucketInput, HeadBucketOutput, HeadObjectInput,
     HeadObjectOutput, ListObjectVersionsInput, ListObjectVersionsOutput, ListObjectsV2Input, ListObjectsV2Output, Object,
-    ObjectLockConfiguration, ObjectLockEnabled, ObjectStorageClass, ObjectVersionId, PutObjectInput, PutObjectOutput,
+    ObjectLockConfiguration, ObjectLockEnabled, ObjectLockLegalHold, ObjectLockLegalHoldStatus, ObjectLockMode,
-    PutObjectTaggingInput, PutObjectTaggingOutput, Range, StreamingBlob, Tag, TagSet, Timestamp, TimestampFormat,
+    ObjectLockRetention, ObjectLockRetentionMode, ObjectStorageClass, ObjectVersionId, PutObjectInput, PutObjectLegalHoldInput,
-    UploadPartInput, UploadPartOutput,
+    PutObjectLegalHoldOutput, PutObjectOutput, PutObjectRetentionInput, PutObjectRetentionOutput, PutObjectTaggingInput,
+    PutObjectTaggingOutput, Range, StreamingBlob, Tag, TagSet, Timestamp, TimestampFormat, UploadPartInput, UploadPartOutput,
 };
 use s3s::service::{S3Service, S3ServiceBuilder};
 use s3s::validation::{AwsNameValidation, NameValidation};
@@ -127,6 +129,10 @@ pub enum Operation {
     GetObjectTagging,
     PutObjectTagging,
     DeleteObjectTagging,
+    GetObjectRetention,
+    PutObjectRetention,
+    GetObjectLegalHold,
+    PutObjectLegalHold,
     ListObjectVersions,
     ListObjectsV2,
     CreateMultipartUpload,
@@ -501,6 +507,10 @@ struct StoreState {
     /// PutObject carrying any `x-amz-object-lock-*` header must also carry
     /// `Content-MD5` or an `x-amz-checksum-*` header.
     require_checksum_for_object_lock: bool,
+    /// Models Wasabi (rustfs/backlog#2340): a version-addressed DELETE of a
+    /// version id the target never had answers 404 `NoSuchVersion` instead of
+    /// the idempotent 204 RustFS/MinIO give.
+    reject_unknown_version_deletes: bool,
     limits: StoreLimits,
     buckets: HashMap<String, BucketState>,
     uploads: HashMap<String, MultipartState>,
@@ -565,6 +575,41 @@ struct ObjectVersion {
     /// SSE-C passthrough transport headers stored with the version (RustFS
     /// target behavior); empty when the drop mode discarded them.
     replication_sse_headers: Vec<(String, String)>,
+    /// Object Lock state of the version: retention (mode, retain-until) from
+    /// the PUT / CreateMultipartUpload headers or PutObjectRetention, and the
+    /// legal hold flag; replayed on HEAD.
+    lock: VersionLock,
+}
+
+#[derive(Clone, Default)]
+struct VersionLock {
+    retention: Option<(String, Timestamp)>,
+    /// `None` until a legal hold status was ever set; like S3, HEAD then
+    /// reports nothing, while an explicit OFF is reported as `OFF`.
+    legal_hold: Option<bool>,
+}
+
+impl VersionLock {
+    fn from_headers(
+        mode: Option<ObjectLockMode>,
+        retain_until: Option<Timestamp>,
+        legal_hold: Option<ObjectLockLegalHoldStatus>,
+    ) -> Self {
+        Self {
+            retention: mode.zip(retain_until).map(|(mode, until)| (mode.as_str().to_string(), until)),
+            legal_hold: legal_hold.map(|status| status.as_str().eq_ignore_ascii_case("ON")),
+        }
+    }
+
+    fn legal_hold_status(&self) -> Option<ObjectLockLegalHoldStatus> {
+        self.legal_hold.map(|on| {
+            ObjectLockLegalHoldStatus::from_static(if on {
+                ObjectLockLegalHoldStatus::ON
+            } else {
+                ObjectLockLegalHoldStatus::OFF
+            })
+        })
+    }
 }
 
 #[derive(Clone)]
@@ -576,6 +621,7 @@ struct MultipartState {
     metadata: Option<HashMap<String, String>>,
     standard_headers: StandardHeaders,
     replication_sse_headers: Vec<(String, String)>,
+    lock: VersionLock,
     parts: BTreeMap<i32, MultipartPart>,
 }
 
@@ -584,6 +630,10 @@ struct MultipartPart {
     body: Bytes,
     e_tag: String,
     digest: [u8; 16],
+    /// Plaintext length declared by an SSE-C passthrough sender
+    /// (`x-rustfs-replication-part-actual-size`); RustFS validates the 5 MiB
+    /// minimum against it rather than against the stored bytes.
+    actual_size: Option<usize>,
 }
 
 #[derive(Clone)]
@@ -845,6 +895,7 @@ impl FakeS3Target {
             standard_headers: seed.standard_headers.clone(),
             tags: Vec::new(),
             replication_sse_headers: Vec::new(),
+            lock: VersionLock::default(),
         };
         upsert_version(&mut state, bucket, key.into(), version).expect("seed object must fit the storage budget");
         e_tag
@@ -918,6 +969,12 @@ impl FakeS3Target {
     /// PutObject that carries Object Lock parameters (AWS S3 / MinIO rule,
     /// rustfs#7082). `Content-MD5`, when present, is always verified against
     /// the body regardless of this mode.
+    /// Wasabi-like mode: DELETE of an unknown version id answers 404
+    /// `NoSuchVersion` (the default 204 models RustFS/MinIO).
+    pub fn reject_unknown_version_deletes(&self, enabled: bool) {
+        lock(&self.backend.store).reject_unknown_version_deletes = enabled;
+    }
+
     pub fn require_checksum_for_object_lock(&self, enabled: bool) {
         lock(&self.backend.store).require_checksum_for_object_lock = enabled;
     }
@@ -1152,6 +1209,10 @@ fn operation_from_s3_name(name: &str) -> Operation {
         "GetObjectTagging" => Operation::GetObjectTagging,
         "PutObjectTagging" => Operation::PutObjectTagging,
         "DeleteObjectTagging" => Operation::DeleteObjectTagging,
+        "GetObjectRetention" => Operation::GetObjectRetention,
+        "PutObjectRetention" => Operation::PutObjectRetention,
+        "GetObjectLegalHold" => Operation::GetObjectLegalHold,
+        "PutObjectLegalHold" => Operation::PutObjectLegalHold,
         "ListObjectsV2" => Operation::ListObjectsV2,
         "CreateMultipartUpload" => Operation::CreateMultipartUpload,
         "UploadPart" => Operation::UploadPart,
@@ -1289,6 +1350,18 @@ fn parse_request(method: &Method, uri: &Uri) -> ParsedRequest {
         (&Method::DELETE, true) if query.contains_key("tagging") && only_query_keys(&["tagging", "versionId"]) => {
             Operation::DeleteObjectTagging
         }
+        (&Method::GET, true) if query.contains_key("retention") && only_query_keys(&["retention", "versionId"]) => {
+            Operation::GetObjectRetention
+        }
+        (&Method::PUT, true) if query.contains_key("retention") && only_query_keys(&["retention", "versionId"]) => {
+            Operation::PutObjectRetention
+        }
+        (&Method::GET, true) if query.contains_key("legal-hold") && only_query_keys(&["legal-hold", "versionId"]) => {
+            Operation::GetObjectLegalHold
+        }
+        (&Method::PUT, true) if query.contains_key("legal-hold") && only_query_keys(&["legal-hold", "versionId"]) => {
+            Operation::PutObjectLegalHold
+        }
         // A replication PUT addresses the source version via `?versionId=`.
         (&Method::PUT, true) if only_query_keys(&["versionId"]) => Operation::PutObject,
         (&Method::GET, true) if only_query_keys(&["versionId"]) => Operation::GetObject,
@@ -1842,6 +1915,28 @@ fn set_version_tags(
     Ok(resolved)
 }
 
+fn update_version_lock(
+    state: &mut StoreState,
+    bucket: &str,
+    key: &str,
+    version_id: Option<&str>,
+    update: impl FnOnce(&mut VersionLock),
+) -> S3Result<String> {
+    let resolved = find_version(state, bucket, key, version_id)?.version_id;
+    let version = state
+        .buckets
+        .get_mut(bucket)
+        .expect("bucket existence checked by find_version")
+        .objects
+        .get_mut(key)
+        .expect("key existence checked by find_version")
+        .iter_mut()
+        .find(|version| version.version_id == resolved)
+        .expect("version existence checked by find_version");
+    update(&mut version.lock);
+    Ok(resolved)
+}
+
 /// Whether version ids are surfaced for this bucket. Unknown buckets report
 /// `true`; the caller's lookup raises `NoSuchBucket` first.
 fn bucket_versioned(state: &StoreState, bucket: &str) -> bool {
@@ -2281,6 +2376,11 @@ impl S3 for FakeBackend {
             standard_headers,
             tags: Vec::new(),
             replication_sse_headers: captured_replication_sse_headers(&headers, drop_unlisted),
+            lock: VersionLock::from_headers(
+                input.object_lock_mode,
+                input.object_lock_retain_until_date,
+                input.object_lock_legal_hold_status,
+            ),
         };
         upsert_version(&mut lock(&self.store), &input.bucket, input.key, version)?;
         Ok(apply_response_fault(
@@ -2339,6 +2439,13 @@ impl S3 for FakeBackend {
             last_modified: Some(version.last_modified.clone()),
             version_id: versioned.then_some(version.version_id),
             sse_customer_algorithm,
+            object_lock_mode: version
+                .lock
+                .retention
+                .as_ref()
+                .map(|(mode, _)| ObjectLockMode::from(mode.clone())),
+            object_lock_retain_until_date: version.lock.retention.as_ref().map(|(_, until)| until.clone()),
+            object_lock_legal_hold_status: version.lock.legal_hold_status(),
             ..Default::default()
         });
         response.status = served.status;
@@ -2373,6 +2480,13 @@ impl S3 for FakeBackend {
             last_modified: Some(version.last_modified.clone()),
             version_id: versioned.then_some(version.version_id),
             sse_customer_algorithm,
+            object_lock_mode: version
+                .lock
+                .retention
+                .as_ref()
+                .map(|(mode, _)| ObjectLockMode::from(mode.clone())),
+            object_lock_retain_until_date: version.lock.retention.as_ref().map(|(_, until)| until.clone()),
+            object_lock_legal_hold_status: version.lock.legal_hold_status(),
             ..Default::default()
         });
         response.status = served.status;
@@ -2432,6 +2546,82 @@ impl S3 for FakeBackend {
         ))
     }
 
+    async fn get_object_retention(
+        &self,
+        req: S3Request<GetObjectRetentionInput>,
+    ) -> S3Result<S3Response<GetObjectRetentionOutput>> {
+        let fault = request_fault(&req);
+        apply_non_body_fault(fault.as_ref(), &self.control).await?;
+        let input = req.input;
+        let version = find_version(&lock(&self.store), &input.bucket, &input.key, input.version_id.as_deref())?;
+        Ok(apply_response_fault(
+            S3Response::new(GetObjectRetentionOutput {
+                retention: version.lock.retention.map(|(mode, until)| ObjectLockRetention {
+                    mode: Some(ObjectLockRetentionMode::from(mode)),
+                    retain_until_date: Some(until),
+                }),
+            }),
+            fault.as_ref(),
+        ))
+    }
+
+    async fn put_object_retention(
+        &self,
+        req: S3Request<PutObjectRetentionInput>,
+    ) -> S3Result<S3Response<PutObjectRetentionOutput>> {
+        let fault = request_fault(&req);
+        apply_non_body_fault(fault.as_ref(), &self.control).await?;
+        let input = req.input;
+        let retention = input
+            .retention
+            .and_then(|retention| retention.mode.zip(retention.retain_until_date))
+            .map(|(mode, until)| (mode.as_str().to_string(), until));
+        update_version_lock(&mut lock(&self.store), &input.bucket, &input.key, input.version_id.as_deref(), |lock| {
+            lock.retention = retention;
+        })?;
+        Ok(apply_response_fault(S3Response::new(PutObjectRetentionOutput::default()), fault.as_ref()))
+    }
+
+    async fn get_object_legal_hold(
+        &self,
+        req: S3Request<GetObjectLegalHoldInput>,
+    ) -> S3Result<S3Response<GetObjectLegalHoldOutput>> {
+        let fault = request_fault(&req);
+        apply_non_body_fault(fault.as_ref(), &self.control).await?;
+        let input = req.input;
+        let version = find_version(&lock(&self.store), &input.bucket, &input.key, input.version_id.as_deref())?;
+        Ok(apply_response_fault(
+            S3Response::new(GetObjectLegalHoldOutput {
+                legal_hold: Some(ObjectLockLegalHold {
+                    status: Some(
+                        version
+                            .lock
+                            .legal_hold_status()
+                            .unwrap_or_else(|| ObjectLockLegalHoldStatus::from_static(ObjectLockLegalHoldStatus::OFF)),
+                    ),
+                }),
+            }),
+            fault.as_ref(),
+        ))
+    }
+
+    async fn put_object_legal_hold(
+        &self,
+        req: S3Request<PutObjectLegalHoldInput>,
+    ) -> S3Result<S3Response<PutObjectLegalHoldOutput>> {
+        let fault = request_fault(&req);
+        apply_non_body_fault(fault.as_ref(), &self.control).await?;
+        let input = req.input;
+        let legal_hold_on = input
+            .legal_hold
+            .and_then(|hold| hold.status)
+            .is_some_and(|status| status.as_str().eq_ignore_ascii_case("ON"));
+        update_version_lock(&mut lock(&self.store), &input.bucket, &input.key, input.version_id.as_deref(), |lock| {
+            lock.legal_hold = Some(legal_hold_on);
+        })?;
+        Ok(apply_response_fault(S3Response::new(PutObjectLegalHoldOutput::default()), fault.as_ref()))
+    }
+
     async fn delete_object_tagging(
         &self,
         req: S3Request<DeleteObjectTaggingInput>,
@@ -2485,6 +2675,7 @@ impl S3 for FakeBackend {
             return Ok(apply_response_fault(S3Response::new(DeleteObjectOutput::default()), fault.as_ref()));
         }
         if let Some(version_id) = input.version_id {
+            let reject_unknown = state.reject_unknown_version_deletes;
             let (removed_bytes, removed_versions, delete_marker, remove_key) = {
                 let Some(versions) = state
                     .buckets
@@ -2493,6 +2684,9 @@ impl S3 for FakeBackend {
                     .objects
                     .get_mut(&input.key)
                 else {
+                    if reject_unknown {
+                        return Err(s3s::s3_error!(NoSuchVersion, "The specified version does not exist."));
+                    }
                     return Ok(apply_response_fault(
                         S3Response::new(DeleteObjectOutput {
                             version_id: Some(version_id),
@@ -2501,6 +2695,9 @@ impl S3 for FakeBackend {
                         fault.as_ref(),
                     ));
                 };
+                if reject_unknown && !versions.iter().any(|version| version.version_id == version_id) {
+                    return Err(s3s::s3_error!(NoSuchVersion, "The specified version does not exist."));
+                }
                 let mut removed_bytes = 0usize;
                 let mut removed_versions = 0usize;
                 let mut delete_marker = None;
@@ -2554,6 +2751,7 @@ impl S3 for FakeBackend {
                 standard_headers: StandardHeaders::default(),
                 tags: Vec::new(),
                 replication_sse_headers: Vec::new(),
+                lock: VersionLock::default(),
             },
         )?;
         Ok(apply_response_fault(
@@ -2608,6 +2806,11 @@ impl S3 for FakeBackend {
                 metadata: input.metadata,
                 standard_headers,
                 replication_sse_headers: captured_replication_sse_headers(&headers, drop_unlisted),
+                lock: VersionLock::from_headers(
+                    input.object_lock_mode,
+                    input.object_lock_retain_until_date,
+                    input.object_lock_legal_hold_status,
+                ),
                 parts: BTreeMap::new(),
             },
         );
@@ -2624,6 +2827,11 @@ impl S3 for FakeBackend {
 
     async fn upload_part(&self, req: S3Request<UploadPartInput>) -> S3Result<S3Response<UploadPartOutput>> {
         let fault = request_fault(&req);
+        let declared_actual_size = req
+            .headers
+            .get("x-rustfs-replication-part-actual-size")
+            .and_then(|value| value.to_str().ok())
+            .and_then(|value| value.parse::<usize>().ok());
         let _body_permit = timeout(MAX_FAULT_DURATION, Arc::clone(&self.body_limit).acquire_owned())
             .await
             .map_err(|_| s3s::s3_error!(RequestTimeout, "fake target body limiter wait exceeded 30 seconds"))?
@@ -2665,6 +2873,7 @@ impl S3 for FakeBackend {
                 body,
                 e_tag: e_tag.clone(),
                 digest,
+                actual_size: declared_actual_size,
             },
         );
         Ok(apply_response_fault(
@@ -2734,7 +2943,9 @@ impl S3 for FakeBackend {
                 if requested_etag != &stored.e_tag {
                     return Err(s3s::s3_error!(InvalidPart, "part ETag does not match"));
                 }
-                if index + 1 != requested_parts.len() && stored.body.len() < MIN_MULTIPART_PART_BYTES {
+                if index + 1 != requested_parts.len()
+                    && stored.actual_size.unwrap_or(stored.body.len()) < MIN_MULTIPART_PART_BYTES
+                {
                     return Err(s3s::s3_error!(EntityTooSmall, "non-final multipart part is smaller than 5 MiB"));
                 }
                 selected.push((*number, stored.clone()));
@@ -2748,6 +2959,7 @@ impl S3 for FakeBackend {
                     metadata: upload.metadata.clone(),
                     standard_headers: upload.standard_headers.clone(),
                     replication_sse_headers: upload.replication_sse_headers.clone(),
+                    lock: upload.lock.clone(),
                     parts: BTreeMap::new(),
                 },
                 selected,
@@ -2778,6 +2990,7 @@ impl S3 for FakeBackend {
             standard_headers: upload.standard_headers,
             tags: Vec::new(),
             replication_sse_headers: upload.replication_sse_headers,
+            lock: upload.lock,
         };
         let mut state = lock(&self.store);
         let versioned = bucket_versioned(&state, &input.bucket);
@@ -4609,6 +4822,7 @@ mod tests {
                         metadata: None,
                         standard_headers: StandardHeaders::default(),
                         replication_sse_headers: Vec::new(),
+                        lock: VersionLock::default(),
                         parts: BTreeMap::new(),
                     },
                 );

crates/e2e_test/src/heal_erasure_disk_rebuild_test.rs

@@ -34,6 +34,8 @@ mod tests {
     use tokio::net::TcpStream;
     use tokio::time::{Duration, Instant, sleep, timeout};
     use tracing::info;
+    #[cfg(target_os = "linux")]
+    use tracing::warn;
 
     const POOL_METADATA_OBJECT: &str = "pool.bin";
 
@@ -52,6 +54,34 @@ mod tests {
         test_binary: EvidenceBuild,
     }
 
+    #[derive(Clone, Copy)]
+    struct ScannerHealEvidenceCase {
+        id: &'static str,
+        oracle: &'static str,
+        evidence: &'static str,
+        unclean_shutdown_marker: bool,
+    }
+
+    const BACKGROUND_TARGET_RESTART_EVIDENCE: ScannerHealEvidenceCase = ScannerHealEvidenceCase {
+        id: "background-target-restart",
+        oracle: "background-target-restart.json",
+        evidence: "process-restart",
+        unclean_shutdown_marker: false,
+    };
+
+    const BACKGROUND_TARGET_CRASH_EVIDENCE: ScannerHealEvidenceCase = ScannerHealEvidenceCase {
+        id: "background-target-crash",
+        oracle: "background-target-crash.json",
+        evidence: "process-crash-restart",
+        unclean_shutdown_marker: true,
+    };
+
+    struct RestartEvidenceContext {
+        directory: PathBuf,
+        run: RestartEvidenceRun,
+        case: ScannerHealEvidenceCase,
+    }
+
     fn file_sha256(path: &Path) -> Result<String, Box<dyn Error + Send + Sync>> {
         let mut file = std::fs::File::open(path)?;
         let mut digest = Sha256::new();
@@ -66,10 +96,24 @@ mod tests {
         Ok(digest.finalize().iter().map(|byte| format!("{byte:02x}")).collect())
     }
 
-    fn restart_evidence_run(binary: &Path) -> Result<Option<(PathBuf, RestartEvidenceRun)>, Box<dyn Error + Send + Sync>> {
+    fn restart_evidence_run(
+        binary: &Path,
+        case: ScannerHealEvidenceCase,
+    ) -> Result<Option<RestartEvidenceContext>, Box<dyn Error + Send + Sync>> {
         let Some(directory) = std::env::var_os("RUSTFS_SCANNER_HEAL_RUN_DIR") else {
             return Ok(None);
         };
+        if case.id.is_empty()
+            || case.oracle.is_empty()
+            || !case.oracle.ends_with(".json")
+            || case.oracle.contains('/')
+            || case.oracle.contains('\\')
+            || case.oracle.contains("..")
+            || !matches!(case.evidence, "process-restart" | "process-crash-restart")
+            || (case.evidence == "process-crash-restart") != case.unclean_shutdown_marker
+        {
+            return Err("invalid scanner/heal evidence case".into());
+        }
         let directory = PathBuf::from(directory);
         let receipt = directory.join("run.json");
         if receipt.metadata()?.len() > 1024 * 1024 {
@@ -89,10 +133,10 @@ mod tests {
             run.test_binary.sha256,
             "test executable must match the run receipt"
         );
-        if directory.join("background-target-restart.json").exists() {
+        if directory.join(case.oracle).exists() {
             return Err("scanner/heal oracle already exists; create a new execution receipt".into());
         }
-        Ok(Some((directory, run)))
+        Ok(Some(RestartEvidenceContext { directory, run, case }))
     }
 
     fn compiled_test_identity() -> serde_json::Value {
@@ -115,6 +159,49 @@ mod tests {
     }
 
     impl TcpPortBlackhole {
+        /// Environment flag that turns an unusable fault-injection host into a
+        /// hard failure instead of a logged skip. Lanes that provision
+        /// `CAP_NET_ADMIN` set it so a broken runner cannot pass silently.
+        #[cfg(target_os = "linux")]
+        const REQUIRE_ENV: &str = "RUSTFS_E2E_REQUIRE_NET_FAULT_INJECTION";
+
+        /// Probe whether this host can manipulate the OUTPUT chain at all.
+        ///
+        /// Returns `Ok(Some(reason))` when `iptables` is missing or lacks
+        /// `CAP_NET_ADMIN` (the nf_tables backend reports "Permission denied"
+        /// even under `sudo` inside an unprivileged container) and the lane did
+        /// not demand fault injection; returns an error when the lane demands
+        /// it; returns `Ok(None)` when the blackhole can be installed.
+        #[cfg(target_os = "linux")]
+        fn unavailable_reason() -> Result<Option<String>, Box<dyn Error + Send + Sync>> {
+            let id = Command::new("id").arg("-u").output()?;
+            if !id.status.success() {
+                return Err(format!("failed to determine the test process uid: {}", String::from_utf8_lossy(&id.stderr)).into());
+            }
+            let use_sudo = String::from_utf8_lossy(&id.stdout).trim() != "0";
+            let mut command = if use_sudo {
+                let mut command = Command::new("sudo");
+                command.args(["-n", "iptables"]);
+                command
+            } else {
+                Command::new("iptables")
+            };
+            let probe = command.args(["-w", "5", "-S", "OUTPUT"]).output();
+            let reason = match probe {
+                Ok(output) if output.status.success() => return Ok(None),
+                Ok(output) => format!(
+                    "iptables cannot read the OUTPUT chain (status {}): {}",
+                    output.status,
+                    String::from_utf8_lossy(&output.stderr).trim()
+                ),
+                Err(err) => format!("iptables is not runnable: {err}"),
+            };
+            if std::env::var_os(Self::REQUIRE_ENV).is_some() {
+                return Err(format!("{} is set but network fault injection is unavailable: {reason}", Self::REQUIRE_ENV).into());
+            }
+            Ok(Some(reason))
+        }
+
         fn install(address: &str) -> Result<Self, Box<dyn Error + Send + Sync>> {
             let address = address.parse::<SocketAddr>()?;
             if !address.ip().is_loopback() {
@@ -199,6 +286,27 @@ mod tests {
         }
     }
 
+    /// Remove a disk directory underneath a running server. Background writers
+    /// (scanner, usage cache, heal markers) can recreate entries between the
+    /// recursive listing and the final `rmdir`, which surfaces as
+    /// `DirectoryNotEmpty` on macOS; retry briefly so the wipe reflects the
+    /// operator action rather than a listing race.
+    fn wipe_directory_while_server_runs(disk: &Path) -> std::io::Result<()> {
+        let mut last_err = None;
+        for _ in 0..20 {
+            match std::fs::remove_dir_all(disk) {
+                Ok(()) => return Ok(()),
+                Err(err) if err.kind() == std::io::ErrorKind::NotFound => return Ok(()),
+                Err(err) if err.kind() == std::io::ErrorKind::DirectoryNotEmpty => {
+                    last_err = Some(err);
+                    std::thread::sleep(std::time::Duration::from_millis(100));
+                }
+                Err(err) => return Err(err),
+            }
+        }
+        Err(last_err.expect("retry loop only exits without success after recording an error"))
+    }
+
     fn has_file_under(path: &Path) -> bool {
         let Ok(entries) = std::fs::read_dir(path) else {
             return false;
@@ -481,7 +589,7 @@ mod tests {
             );
         }
 
-        std::fs::remove_dir_all(&disk0).expect("disk0 wipe should succeed while server is running");
+        wipe_directory_while_server_runs(&disk0).expect("disk0 wipe should succeed while server is running");
         std::fs::create_dir_all(&disk0).expect("disk0 should be recreated empty while server is running");
         assert!(!has_file_under(&disk0), "disk0 must be empty immediately after runtime wipe");
 
@@ -855,6 +963,16 @@ mod tests {
         .await?
     }
 
+    #[tokio::test(flavor = "multi_thread")]
+    async fn test_cluster_root_heal_recovers_remote_shards_after_background_target_crash()
+    -> Result<(), Box<dyn Error + Send + Sync>> {
+        timeout(
+            Duration::from_secs(420),
+            run_cluster_root_heal_interruption(InterruptionScenario::BackgroundTargetCrash),
+        )
+        .await?
+    }
+
     #[tokio::test(flavor = "multi_thread")]
     async fn test_cluster_root_heal_recovers_remote_shards_after_coordinator_restart() -> Result<(), Box<dyn Error + Send + Sync>>
     {
@@ -868,6 +986,18 @@ mod tests {
     #[cfg(target_os = "linux")]
     #[tokio::test(flavor = "multi_thread")]
     async fn test_cluster_root_heal_recovers_after_target_endpoint_blackhole() -> Result<(), Box<dyn Error + Send + Sync>> {
+        if let Some(reason) = TcpPortBlackhole::unavailable_reason()? {
+            init_logging();
+            warn!(
+                event = "heal_interruption_skipped",
+                component = "e2e_test",
+                subsystem = "heal",
+                interruption_kind = "target_endpoint_blackhole",
+                reason,
+                "Skipping endpoint blackhole scenario: network fault injection is unavailable on this host"
+            );
+            return Ok(());
+        }
         timeout(
             Duration::from_secs(420),
             run_cluster_root_heal_interruption(InterruptionScenario::TargetEndpointBlackhole),
@@ -879,21 +1009,27 @@ mod tests {
     enum InterruptionScenario {
         IsolatedTargetRestart,
         BackgroundTargetRestart,
+        BackgroundTargetCrash,
         BackgroundCoordinatorRestart,
         TargetEndpointBlackhole,
     }
 
     async fn run_cluster_root_heal_interruption(scenario: InterruptionScenario) -> Result<(), Box<dyn Error + Send + Sync>> {
         let server_binary = rustfs_binary_path();
-        let evidence_run = if scenario == InterruptionScenario::BackgroundTargetRestart {
+        let evidence_run = match scenario {
-            restart_evidence_run(&server_binary)?
+            InterruptionScenario::BackgroundTargetRestart => {
-        } else {
+                restart_evidence_run(&server_binary, BACKGROUND_TARGET_RESTART_EVIDENCE)?
-            None
+            }
+            InterruptionScenario::BackgroundTargetCrash => {
+                restart_evidence_run(&server_binary, BACKGROUND_TARGET_CRASH_EVIDENCE)?
+            }
+            _ => None,
         };
         let mut evidence_objects = Vec::new();
         let (background_enabled, interruption_node, interruption_kind) = match scenario {
             InterruptionScenario::IsolatedTargetRestart => (false, 1, "target_restart"),
             InterruptionScenario::BackgroundTargetRestart => (true, 1, "background_target_restart"),
+            InterruptionScenario::BackgroundTargetCrash => (true, 1, "background_target_crash"),
             InterruptionScenario::BackgroundCoordinatorRestart => (true, 0, "coordinator_restart"),
             InterruptionScenario::TargetEndpointBlackhole => (false, 1, "target_endpoint_blackhole"),
         };
@@ -960,6 +1096,7 @@ mod tests {
             .unwrap_or(4 * 1024 * 1024)
             .clamp(1024 * 1024, 16 * 1024 * 1024);
         let mut expected_manifests = Vec::with_capacity(online_object_count);
+        let mut unclean_shutdown_marker_observed = None;
         for index in 0..online_object_count {
             let key = format!("cluster/online/object-{index:04}.bin");
             let payload_seed = u8::try_from(index + 1).expect("clamped object count must fit in u8");
@@ -1326,7 +1463,11 @@ mod tests {
                 "Restored target endpoint forwarding"
             );
         } else {
-            cluster.stop_node(interruption_node)?;
+            if scenario == InterruptionScenario::BackgroundTargetRestart {
+                cluster.stop_node_gracefully(interruption_node).await?;
+            } else {
+                cluster.stop_node(interruption_node)?;
+            }
             let stopped_count = metadata_count(&replaced_disk, bucket, &expected_manifests);
             assert!(
                 stopped_count > 0 && stopped_count < expected_manifests.len(),
@@ -1342,9 +1483,12 @@ mod tests {
                 .join(".rustfs.sys")
                 .join("unclean-shutdown");
             if background_enabled {
+                let marker_exists = unclean_shutdown_marker.is_file();
+                unclean_shutdown_marker_observed = Some(marker_exists);
+                let expected_marker = !matches!(scenario, InterruptionScenario::BackgroundTargetRestart);
                 assert!(
-                    unclean_shutdown_marker.is_file(),
+                    marker_exists == expected_marker,
-                    "background restart must retain the real unclean-shutdown marker"
+                    "background restart/crash lane observed unexpected unclean-shutdown marker state"
                 );
             } else {
                 match std::fs::remove_file(&unclean_shutdown_marker) {
@@ -1535,17 +1679,23 @@ mod tests {
             return Err(format!("heal data rebuilt but task did not finish successfully: {task_status}").into());
         }
 
-        if let Some((directory, run)) = evidence_run {
+        if let Some(evidence_context) = evidence_run {
             let restarted_pid = cluster.nodes[1].process.as_ref().ok_or("restarted target is absent")?.id();
             assert_ne!(target_pid, restarted_pid, "target must be a new process");
-            assert_eq!(file_sha256(&server_binary)?, run.binary.sha256, "server build changed during restart");
+            assert_eq!(
+                file_sha256(&server_binary)?,
+                evidence_context.run.binary.sha256,
+                "server build changed during restart"
+            );
             let evidence = serde_json::json!({
-                "schema": 1, "case": "background-target-restart", "evidence": "process-restart",
+                "schema": 1, "case": evidence_context.case.id, "evidence": evidence_context.case.evidence,
-                "run_id": run.run_id, "source_revision": run.source_revision,
+                "run_id": evidence_context.run.run_id, "source_revision": evidence_context.run.source_revision,
                 "test_build": compiled_test_identity(),
-                "binary_sha256": run.binary.sha256, "test_binary_sha256": run.test_binary.sha256,
+                "binary_sha256": evidence_context.run.binary.sha256,
+                "test_binary_sha256": evidence_context.run.test_binary.sha256,
                 "topology": {"nodes": cluster.nodes.len(), "drives_per_node": cluster.nodes[0].data_dirs.len()},
                 "pid_before": target_pid, "pid_after": restarted_pid,
+                "unclean_shutdown_marker": unclean_shutdown_marker_observed.unwrap_or(false),
                 "objects": evidence_objects, "node_listings": node_listings,
             });
             let data = serde_json::to_vec(&evidence)?;
@@ -1555,7 +1705,7 @@ mod tests {
             let mut output = std::fs::OpenOptions::new()
                 .write(true)
                 .create_new(true)
-                .open(directory.join("background-target-restart.json"))?;
+                .open(evidence_context.directory.join(evidence_context.case.oracle))?;
             output.write_all(&data)?;
             output.sync_all()?;
         }

crates/e2e_test/src/inline_fast_path_cluster_test.rs

@@ -1676,6 +1676,8 @@ async fn four_node_inline_storage_and_get_boundaries() -> TestResult {
     let collector = OtlpMetricCollector::start().await?;
     let mut cluster = RustFSTestClusterEnvironment::new(4).await?;
     configure_reader_metric_cluster(&mut cluster, &collector);
+    // Inspect every disk only after the PUT rename fanout has drained.
+    cluster.set_env("RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE", "false");
     cluster.start().await?;
 
     for (state_index, state) in [VersionState::Unversioned, VersionState::Enabled, VersionState::Suspended]
@@ -2232,7 +2234,6 @@ async fn four_node_manual_transition_distributed_admission_conflict_reports_stat
     let bucket = format!("distributed-admission-{}", Uuid::new_v4().simple());
     let prefix = "transition/distributed-admission/";
     hot_client.create_bucket().bucket(&bucket).send().await?;
-    put_lifecycle_with_transition_retry(&hot_client, &bucket, &tier_name).await?;
     for index in 0u8..64 {
         let key = format!("{prefix}object-{index:02}.bin");
         hot_client
@@ -2243,6 +2244,7 @@ async fn four_node_manual_transition_distributed_admission_conflict_reports_stat
             .send()
             .await?;
     }
+    put_lifecycle_with_transition_retry(&hot_client, &bucket, &tier_name).await?;
 
     let (node0, node1) = tokio::join!(
         start_manual_transition_job_on_node(&hot, 0, &bucket, prefix, &tier_name, false, 64),

crates/e2e_test/src/kms/kms_fault_recovery_test.rs

@@ -59,6 +59,36 @@ async fn test_kms_key_directory_unavailable() -> Result<(), Box<dyn std::error::
 
     assert_eq!(put_response.server_side_encryption(), Some(&ServerSideEncryption::Aes256));
 
+    // A missing key in the healthy store is a client error, unlike a store outage.
+    let missing_key_object = "test-missing-kms-key";
+    let missing_key_error = s3_client
+        .put_object()
+        .bucket(TEST_BUCKET)
+        .key(missing_key_object)
+        .body(aws_sdk_s3::primitives::ByteStream::from_static(b"must not be published"))
+        .server_side_encryption(ServerSideEncryption::AwsKms)
+        .ssekms_key_id("rustfs-e2e-test-missing-key")
+        .send()
+        .await
+        .expect_err("an unknown key in a healthy Local KMS store must reject the write");
+    assert_eq!(missing_key_error.raw_response().map(|response| response.status().as_u16()), Some(400));
+    assert_eq!(
+        missing_key_error.as_service_error().and_then(ProvideErrorMetadata::code),
+        Some("KMS.NotFoundException")
+    );
+    let missing_key_absence = s3_client
+        .get_object()
+        .bucket(TEST_BUCKET)
+        .key(missing_key_object)
+        .send()
+        .await
+        .expect_err("a write rejected by a missing KMS key must not publish an object");
+    assert_eq!(missing_key_absence.raw_response().map(|response| response.status().as_u16()), Some(404));
+    assert_eq!(
+        missing_key_absence.as_service_error().and_then(ProvideErrorMetadata::code),
+        Some("NoSuchKey")
+    );
+
     // Temporarily rename the key directory to simulate unavailability
     info!("🔧 Simulating key directory unavailability");
     let backup_dir = format!("{}.backup", kms_env.kms_keys_dir);

crates/e2e_test/src/on_demand_migration/get_basic_test.rs

@@ -21,7 +21,7 @@
 use super::common::{BoxError, OdmSourceSpec, OdmTestEnv, SeedObject};
 use crate::fake_s3_target::{BucketMode, Operation};
 use aws_sdk_s3::error::ProvideErrorMetadata;
-use aws_sdk_s3::types::{BucketVersioningStatus, VersioningConfiguration};
+use aws_sdk_s3::types::{BucketVersioningStatus, ObjectAttributes, VersioningConfiguration};
 use bytes::Bytes;
 use std::time::Duration;
 
@@ -103,11 +103,19 @@ async fn get_miss_pulls_inline_and_serves_locally_afterwards() -> TestResult {
 
 #[tokio::test]
 async fn get_large_object_streams_through_and_backfills_in_background() -> TestResult {
+    const PART_SIZE: usize = 5 * 1024 * 1024;
     let bucket = "odm-get-large";
-    let env = configured_env(bucket, |spec| spec.policy.inline_max_bytes = 4096).await?;
+    let env = configured_env(bucket, |spec| {
+        spec.policy.inline_max_bytes = 4096;
+        spec.policy.multipart_part_size_bytes = u64::try_from(PART_SIZE).expect("part size fits in u64");
+    })
+    .await?;
     let key = "large/archive.bin";
-    let body = payload(512 * 1024);
+    let body = payload(PART_SIZE + 4096);
-    env.seed_source(SOURCE_BUCKET, &[SeedObject::new(key, body.clone())]);
+    let etag = env
+        .seed_source(SOURCE_BUCKET, &[SeedObject::new(key, body.clone())])
+        .remove(0);
+    assert_eq!(etag.len(), 32, "the source fixture has a plain MD5 ETag");
 
     let response = env.raw_get(bucket, key).await?;
     assert_eq!(response.status, 200, "{}", String::from_utf8_lossy(&response.body));
@@ -125,6 +133,68 @@ async fn get_large_object_streams_through_and_backfills_in_background() -> TestR
         vec![None, None],
         "one passthrough GET plus one background pull, both unranged"
     );
+
+    let source_requests = env.source.requests().len();
+    let second_part = env.client.get_object().bucket(bucket).key(key).part_number(2).send().await?;
+    assert_eq!(second_part.content_length(), Some(4096), "the completed second part is the tail");
+    assert_eq!(
+        second_part.content_range(),
+        Some(format!("bytes {PART_SIZE}-{}/{}", body.len() - 1, body.len()).as_str()),
+        "partNumber reads the stored multipart boundary"
+    );
+    assert_eq!(
+        second_part.body.collect().await?.into_bytes(),
+        body.slice(PART_SIZE..),
+        "the local second part contains the exact source tail"
+    );
+    let third_part = env
+        .client
+        .get_object()
+        .bucket(bucket)
+        .key(key)
+        .part_number(3)
+        .send()
+        .await
+        .expect_err("the completed object has exactly two parts");
+    assert_eq!(third_part.code(), Some("InvalidPart"));
+
+    let mut part_marker = None;
+    for (part_number, part_size) in [(1, PART_SIZE), (2, 4096)] {
+        let attributes = env
+            .client
+            .get_object_attributes()
+            .bucket(bucket)
+            .key(key)
+            .object_attributes(ObjectAttributes::ObjectParts)
+            .object_attributes(ObjectAttributes::Etag)
+            .max_parts(1)
+            .set_part_number_marker(part_marker.clone())
+            .send()
+            .await?;
+        assert_eq!(
+            attributes.e_tag().map(|value| value.trim_matches('"')),
+            Some(etag.as_str()),
+            "multipart write-back preserves the source MD5 ETag"
+        );
+        let parts = attributes
+            .object_parts()
+            .expect("RustFS must expose the stored multipart layout");
+        assert_eq!(parts.total_parts_count(), Some(2));
+        assert_eq!(parts.max_parts(), Some(1));
+        assert_eq!(parts.is_truncated(), Some(part_number == 1));
+        assert_eq!(parts.parts().len(), 1, "RustFS returns one stored part per requested page");
+        assert_eq!(parts.parts()[0].part_number(), Some(part_number));
+        assert_eq!(parts.parts()[0].size(), Some(i64::try_from(part_size).expect("part size fits in i64")));
+        part_marker = parts.next_part_number_marker().map(str::to_owned);
+        if part_number == 1 {
+            assert_eq!(part_marker.as_deref(), Some("1"), "the next request continues after the first part");
+        }
+    }
+    assert_eq!(
+        env.source.requests().len(),
+        source_requests,
+        "local part reads must not consult the source"
+    );
     Ok(())
 }
 

crates/e2e_test/src/on_demand_migration/interaction_test.rs

@@ -23,16 +23,20 @@
 
 use super::common::{
     ALLOW_LOOPBACK_SOURCE_ENV, AdminResponse, BackfillOp, BackfillRequest, BoxError, ODM_MODULE_SWITCH_ENV, ODM_SERVER_ENV,
-    OdmEnvOptions, OdmSourceSpec, OdmTestEnv, SeedObject, start_configured_env, start_configured_env_with,
+    OdmEnvOptions, OdmSourceSpec, OdmTestEnv, SeedObject, start_configured_env, start_configured_env_with, start_source_rustfs,
 };
 use crate::common::{RustFSTestEnvironment, replication_fast_env, signed_request};
 use crate::fake_s3_target::{BucketMode, FAKE_ACCESS_KEY, FAKE_SECRET_KEY, FakeS3Target, Operation};
 use crate::object_lock::common::put_object_lock_configuration;
+use crate::replication_extension_test::{
+    ReplicationTargetOptions, enable_bucket_versioning, set_replication_target_with_options,
+};
 use aws_sdk_s3::error::ProvideErrorMetadata;
 use aws_sdk_s3::types::{
     BucketVersioningStatus, Event, FilterRule, FilterRuleName, NotificationConfiguration, NotificationConfigurationFilter,
-    ObjectLockRetentionMode, QueueConfiguration, S3KeyFilter, ServerSideEncryption, ServerSideEncryptionByDefault,
+    ObjectAttributes, ObjectLockRetentionMode, QueueConfiguration, S3KeyFilter, ServerSideEncryption,
-    ServerSideEncryptionConfiguration, ServerSideEncryptionRule, Tag, Tagging, VersioningConfiguration,
+    ServerSideEncryptionByDefault, ServerSideEncryptionConfiguration, ServerSideEncryptionRule, Tag, Tagging,
+    VersioningConfiguration,
 };
 use bytes::Bytes;
 use local_ip_address::local_ip;
@@ -580,6 +584,130 @@ async fn test_odm_pulled_object_replicates_and_target_as_source_is_rejected() ->
         "a bucket may not migrate from its own replication target: {}",
         rejected.body
     );
+    Box::pin(assert_odm_multipart_replicates_to_rustfs(&env, bucket)).await?;
+    Ok(())
+}
+
+async fn assert_odm_multipart_replicates_to_rustfs(env: &OdmTestEnv, bucket: &str) -> TestResult {
+    const PART_SIZE: usize = 5 * 1024 * 1024;
+    let replica = start_source_rustfs().await?;
+    let replica_bucket = "odm-real-replica";
+    replica.create_test_bucket(replica_bucket).await?;
+    enable_bucket_versioning(&replica, replica_bucket).await?;
+    let arn = set_replication_target_with_options(
+        &env.rustfs,
+        bucket,
+        ReplicationTargetOptions {
+            endpoint: &replica.address,
+            access_key: &replica.access_key,
+            secret_key: &replica.secret_key,
+            target_bucket: replica_bucket,
+            secure: false,
+            skip_tls_verify: false,
+            ca_cert_pem: None,
+        },
+    )
+    .await?;
+    put_bucket_replication(&env.rustfs, bucket, &arn).await?;
+    let mut spec = env.fake_source_spec(SOURCE_BUCKET);
+    // Below the 16 MiB inline default the pull is one tee'd PUT with a single
+    // part; force the passthrough + background multipart write-back instead.
+    spec.policy.inline_max_bytes = 4096;
+    spec.policy.multipart_part_size_bytes = PART_SIZE as u64;
+    spec.policy.preserve_etag = true;
+    env.configure_and_wait(bucket, &spec).await?;
+
+    let key = "replicated/preserved-md5-multipart.bin";
+    let body = payload(PART_SIZE + 4096);
+    let source_put = env
+        .source_client()
+        .put_object()
+        .bucket(SOURCE_BUCKET)
+        .key(key)
+        .body(aws_sdk_s3::primitives::ByteStream::from(body.clone()))
+        .send()
+        .await?;
+    let etag = source_put.e_tag().ok_or("source PUT omitted its ETag")?.trim_matches('"');
+    assert_eq!(etag.len(), 32, "the source must retain a single-PUT MD5 ETag");
+    assert!(etag.bytes().all(|byte| byte.is_ascii_hexdigit()));
+    let pulled = env.raw_get(bucket, key).await?;
+    assert_eq!(pulled.status, 200, "{}", String::from_utf8_lossy(&pulled.body));
+    assert_eq!(pulled.body, body);
+    assert!(env.wait_local_listed(bucket, key, SETTLE).await?, "the multipart pull must persist");
+
+    let deadline = Instant::now() + SETTLE;
+    let source_head = loop {
+        let head = env.client.head_object().bucket(bucket).key(key).send().await?;
+        match head.replication_status().map(|status| status.as_str()) {
+            Some("COMPLETED") => break head,
+            Some("FAILED") => return Err("the ODM multipart copy failed replication to RustFS".into()),
+            _ => {
+                assert!(Instant::now() < deadline, "the ODM multipart copy never completed replication to RustFS");
+                tokio::time::sleep(Duration::from_millis(200)).await;
+            }
+        }
+    };
+    let version = source_head
+        .version_id()
+        .ok_or("the versioned ODM copy omitted its version id")?;
+    assert_ne!(version, "null");
+    let replica_client = replica.create_s3_client();
+    for (client, object_bucket) in [(&env.client, bucket), (&replica_client, replica_bucket)] {
+        let attributes = client
+            .get_object_attributes()
+            .bucket(object_bucket)
+            .key(key)
+            .version_id(version)
+            .object_attributes(ObjectAttributes::Etag)
+            .object_attributes(ObjectAttributes::ObjectParts)
+            .send()
+            .await?;
+        assert_eq!(attributes.e_tag().map(|value| value.trim_matches('"')), Some(etag));
+        let parts = attributes
+            .object_parts()
+            .ok_or("the local copy and replica must both expose two parts")?;
+        assert_eq!(parts.total_parts_count(), Some(2));
+        assert_eq!(
+            parts
+                .parts()
+                .iter()
+                .map(|part| (part.part_number(), part.size()))
+                .collect::<Vec<_>>(),
+            [(Some(1), Some(PART_SIZE as i64)), (Some(2), Some(4096))]
+        );
+    }
+    // REPLICA status surfaces on HEAD, like the other inbound-replica checks.
+    let replica_head = replica_client
+        .head_object()
+        .bucket(replica_bucket)
+        .key(key)
+        .version_id(version)
+        .send()
+        .await?;
+    assert_eq!(replica_head.replication_status().map(|status| status.as_str()), Some("REPLICA"));
+    let replica_get = replica_client
+        .get_object()
+        .bucket(replica_bucket)
+        .key(key)
+        .version_id(version)
+        .send()
+        .await?;
+    assert_eq!(replica_get.version_id(), Some(version));
+    assert_eq!(replica_get.body.collect().await?.into_bytes(), body);
+    let boundary = replica_client
+        .get_object()
+        .bucket(replica_bucket)
+        .key(key)
+        .version_id(version)
+        .range(format!("bytes={}-{}", PART_SIZE - 32, PART_SIZE + 31))
+        .send()
+        .await?;
+    assert_eq!(boundary.body.collect().await?.into_bytes(), body.slice(PART_SIZE - 32..PART_SIZE + 32));
+    assert_eq!(
+        env.source.count_requests(Operation::GetObject, key),
+        2,
+        "one passthrough GET plus one background pull; replication and local reads must not fetch the migration source again"
+    );
     Ok(())
 }
 

crates/e2e_test/src/presigned_negative_test.rs

@@ -356,3 +356,238 @@ async fn tampered_presigned_put_returns_signature_does_not_match() -> Result<(),
     );
     Ok(())
 }
+
+/// GHSA-g8w9-qw9q-fghr: a presigned PUT signed with `SignedHeaders=host` must
+/// not honour `x-amz-*` headers the uploader adds afterwards. The presign
+/// authorised one plain upload; the extra headers would set tags, storage
+/// class and a website redirect the presigner never covered. AWS S3 rejects
+/// this with 403 `AccessDenied`, and so must RustFS — and the object must not
+/// be stored at all, not merely stored without the properties.
+#[tokio::test]
+async fn ghsa_g8w9_presigned_put_rejects_unsigned_x_amz_headers() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    init_logging();
+    let mut env = RustFSTestEnvironment::new().await?;
+    setup(&mut env).await?;
+
+    let key = "presigned-put-unsigned-amz-headers.txt";
+    let pr = env
+        .create_s3_client()
+        .put_object()
+        .bucket(BUCKET)
+        .key(key)
+        .presigned(valid_config())
+        .await?;
+    assert!(
+        !pr.headers().any(|(name, _)| name.eq_ignore_ascii_case("x-amz-tagging")),
+        "fixture must presign a plain PutObject without tagging so the header below is unsigned"
+    );
+
+    let unsigned: Vec<(&str, &str)> = vec![
+        ("x-amz-tagging", "owner=attacker&classification=public"),
+        ("x-amz-website-redirect-location", "https://attacker.example/phish"),
+        ("x-amz-storage-class", "REDUCED_REDUNDANCY"),
+    ];
+    let headers = pr.headers().chain(unsigned.iter().copied());
+    let resp = send_raw(pr.method(), pr.uri(), headers, Some(b"should-not-be-stored".to_vec())).await?;
+    let status = resp.status();
+    let body = resp.text().await?;
+    assert_eq!(
+        status.as_u16(),
+        403,
+        "presigned PUT with unsigned x-amz-* headers must be 403, body:\n{body}"
+    );
+    assert_error_code(&body, "AccessDenied");
+    assert!(
+        body.contains("were not signed"),
+        "rejection must name unsigned headers as the cause, got:\n{body}"
+    );
+
+    let error = env
+        .create_s3_client()
+        .head_object()
+        .bucket(BUCKET)
+        .key(key)
+        .send()
+        .await
+        .expect_err("presigned PUT with unsigned x-amz-* headers must not store the object");
+    assert_eq!(
+        error.raw_response().map(|response| response.status().as_u16()),
+        Some(404),
+        "absence probe after the rejected upload must return HTTP 404, got {error:?}"
+    );
+    Ok(())
+}
+
+/// GHSA-g8w9-qw9q-fghr positive control: when the presigner itself sets the
+/// property, the SDK lists `x-amz-tagging` in `SignedHeaders`, the uploader
+/// replays it, and the upload succeeds with the tags applied. Without this the
+/// negative test above could pass because the server rejects every tagged
+/// presigned upload.
+#[tokio::test]
+async fn ghsa_g8w9_presigned_put_accepts_signed_x_amz_headers() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    init_logging();
+    let mut env = RustFSTestEnvironment::new().await?;
+    setup(&mut env).await?;
+
+    let key = "presigned-put-signed-tagging.txt";
+    let pr = env
+        .create_s3_client()
+        .put_object()
+        .bucket(BUCKET)
+        .key(key)
+        .tagging("owner=app")
+        .presigned(valid_config())
+        .await?;
+    assert!(
+        pr.headers().any(|(name, _)| name.eq_ignore_ascii_case("x-amz-tagging")),
+        "fixture must carry x-amz-tagging as a signed header"
+    );
+    assert!(
+        pr.uri().contains("x-amz-tagging"),
+        "X-Amz-SignedHeaders must list x-amz-tagging, uri: {}",
+        pr.uri()
+    );
+
+    let resp = send_presigned(&pr, Some(b"stored-with-signed-tagging".to_vec())).await?;
+    let status = resp.status();
+    let body = resp.text().await?;
+    assert!(
+        status.is_success(),
+        "presigned PUT with signed x-amz-tagging must succeed, got {status}, body:\n{body}"
+    );
+
+    let tags = env
+        .create_s3_client()
+        .get_object_tagging()
+        .bucket(BUCKET)
+        .key(key)
+        .send()
+        .await?;
+    let tag_set: Vec<(String, String)> = tags
+        .tag_set()
+        .iter()
+        .map(|tag| (tag.key().to_string(), tag.value().to_string()))
+        .collect();
+    assert_eq!(tag_set, vec![("owner".to_string(), "app".to_string())], "signed tagging must be applied");
+    info!("signed presigned tagging control passed");
+    Ok(())
+}
+
+/// GHSA-g8w9-qw9q-fghr on the read side: a presigned GET signed with
+/// `SignedHeaders=host` must not accept an unsigned SSE-C header. The header
+/// would otherwise select a decryption path the presigner never authorised.
+#[tokio::test]
+async fn ghsa_g8w9_presigned_get_rejects_unsigned_x_amz_headers() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    init_logging();
+    let mut env = RustFSTestEnvironment::new().await?;
+    setup(&mut env).await?;
+
+    let pr = env
+        .create_s3_client()
+        .get_object()
+        .bucket(BUCKET)
+        .key(CANONICAL_KEY)
+        .presigned(valid_config())
+        .await?;
+
+    let unsigned: Vec<(&str, &str)> = vec![("x-amz-server-side-encryption-customer-algorithm", "AES256")];
+    let headers = pr.headers().chain(unsigned.iter().copied());
+    let resp = send_raw(pr.method(), pr.uri(), headers, None).await?;
+    let status = resp.status();
+    let body = resp.text().await?;
+    assert_eq!(
+        status.as_u16(),
+        403,
+        "presigned GET with an unsigned x-amz-* header must be 403, body:\n{body}"
+    );
+    assert_error_code(&body, "AccessDenied");
+    assert!(
+        !body.contains(std::str::from_utf8(CANONICAL_BODY)?),
+        "rejected GET must not leak the object body"
+    );
+    Ok(())
+}
+
+/// GHSA-g8w9-qw9q-fghr: an unsigned `x-amz-copy-source` would turn a presigned
+/// PutObject into a CopyObject of an arbitrary readable key, since operation
+/// routing happens before authorization. The presigned upload must fail and
+/// leave nothing behind.
+#[tokio::test]
+async fn ghsa_g8w9_presigned_put_rejects_unsigned_copy_source() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    init_logging();
+    let mut env = RustFSTestEnvironment::new().await?;
+    setup(&mut env).await?;
+
+    let key = "presigned-put-unsigned-copy-source.txt";
+    let pr = env
+        .create_s3_client()
+        .put_object()
+        .bucket(BUCKET)
+        .key(key)
+        .presigned(valid_config())
+        .await?;
+
+    let copy_source = format!("/{BUCKET}/{CANONICAL_KEY}");
+    let unsigned: Vec<(&str, &str)> = vec![("x-amz-copy-source", copy_source.as_str())];
+    let headers = pr.headers().chain(unsigned.iter().copied());
+    let resp = send_raw(pr.method(), pr.uri(), headers, None).await?;
+    let status = resp.status();
+    let body = resp.text().await?;
+    assert_eq!(
+        status.as_u16(),
+        403,
+        "presigned PUT with an unsigned copy source must be 403, body:\n{body}"
+    );
+    assert_error_code(&body, "AccessDenied");
+
+    let error = env
+        .create_s3_client()
+        .head_object()
+        .bucket(BUCKET)
+        .key(key)
+        .send()
+        .await
+        .expect_err("rejected copy must not create the destination object");
+    assert_eq!(error.raw_response().map(|response| response.status().as_u16()), Some(404));
+    Ok(())
+}
+
+/// GHSA-g8w9-qw9q-fghr boundary control: the rule covers `x-amz-*` only. A
+/// plain `Content-Type` on a `SignedHeaders=host` presigned PUT is outside
+/// SigV4's signed-header requirement (AWS S3 accepts it too) and must keep
+/// working, so the negative tests above cannot pass by rejecting every
+/// unsigned header.
+#[tokio::test]
+async fn ghsa_g8w9_presigned_put_still_accepts_unsigned_non_amz_headers() -> Result<(), Box<dyn std::error::Error + Send + Sync>>
+{
+    init_logging();
+    let mut env = RustFSTestEnvironment::new().await?;
+    setup(&mut env).await?;
+
+    let key = "presigned-put-unsigned-content-type.txt";
+    let pr = env
+        .create_s3_client()
+        .put_object()
+        .bucket(BUCKET)
+        .key(key)
+        .presigned(valid_config())
+        .await?;
+
+    let unsigned: Vec<(&str, &str)> = vec![("content-type", "text/x-rustfs-test")];
+    let headers = pr.headers().chain(unsigned.iter().copied());
+    let resp = send_raw(pr.method(), pr.uri(), headers, Some(b"plain-header-upload".to_vec())).await?;
+    let status = resp.status();
+    let body = resp.text().await?;
+    assert!(
+        status.is_success(),
+        "presigned PUT with an unsigned Content-Type must succeed, got {status}, body:\n{body}"
+    );
+
+    let head = env.create_s3_client().head_object().bucket(BUCKET).key(key).send().await?;
+    assert_eq!(
+        head.content_type(),
+        Some("text/x-rustfs-test"),
+        "unsigned Content-Type must still be applied"
+    );
+    Ok(())
+}

crates/e2e_test/src/replication_extension_test.rs

@@ -368,23 +368,23 @@ impl Drop for SlowReplicationTargetGuard {
 // Mirrors madmin-go `ResyncTargetsInfo`/`ResyncTarget` json tags — the same
 // shape `mc replicate resync status` decodes.
 #[derive(Debug, Clone, serde::Deserialize)]
-struct ReplicationResetStatusResponse {
+pub(crate) struct ReplicationResetStatusResponse {
     #[serde(rename = "target", default)]
-    targets: Vec<ReplicationResetStatusTarget>,
+    pub(crate) targets: Vec<ReplicationResetStatusTarget>,
 }
 
 #[derive(Debug, Clone, serde::Deserialize)]
-struct ReplicationResetStatusTarget {
+pub(crate) struct ReplicationResetStatusTarget {
     #[serde(rename = "arn", default)]
-    arn: String,
+    pub(crate) arn: String,
     #[serde(rename = "resetid", default)]
-    reset_id: String,
+    pub(crate) reset_id: String,
     #[serde(rename = "resyncStatus", default)]
-    status: String,
+    pub(crate) status: String,
     #[serde(rename = "replicationCount", default)]
-    replicated_count: i64,
+    pub(crate) replicated_count: i64,
     #[serde(rename = "object", default)]
-    object: String,
+    pub(crate) object: String,
 }
 
 fn extract_xml_tag(xml: &str, tag: &str) -> Option<String> {
@@ -512,7 +512,7 @@ pub(crate) async fn put_bucket_replication(
     put_bucket_replication_with_delete_statuses(env, bucket, target_arn, "Enabled", None).await
 }
 
-async fn put_bucket_replication_with_delete_statuses(
+pub(crate) async fn put_bucket_replication_with_delete_statuses(
     env: &RustFSTestEnvironment,
     bucket: &str,
     target_arn: &str,
@@ -627,7 +627,7 @@ async fn put_bucket_replication_rules(
     Ok(())
 }
 
-async fn delete_bucket_replication(
+pub(crate) async fn delete_bucket_replication(
     env: &RustFSTestEnvironment,
     bucket: &str,
 ) -> Result<reqwest::Response, Box<dyn Error + Send + Sync>> {
@@ -2294,7 +2294,7 @@ async fn site_replication_state_edit(
 /// return the target `(arn, reset_id)`, asserting the response carries the
 /// madmin `ResyncTargetsInfo` shape (`target[0].arn` / `target[0].resetid`)
 /// that `mc replicate resync start` decodes.
-async fn start_bucket_replication_reset(
+pub(crate) async fn start_bucket_replication_reset(
     env: &RustFSTestEnvironment,
     bucket: &str,
 ) -> Result<(String, String), Box<dyn Error + Send + Sync>> {
@@ -2314,7 +2314,7 @@ async fn start_bucket_replication_reset(
     Ok((arn, reset_id))
 }
 
-async fn get_replication_reset_status(
+pub(crate) async fn get_replication_reset_status(
     env: &RustFSTestEnvironment,
     bucket: &str,
     arn: &str,
@@ -3837,6 +3837,244 @@ async fn test_bucket_replication_converges_delete_marker_and_version_purge() ->
     Ok(())
 }
 
+/// Regression for rustfs/backlog#2340 (not Wasabi specific): a directory
+/// marker (`prefix/` with a body) in a versioned bucket is stored as the null
+/// version, like MinIO (`putOpts`: "for directory objects skip creating new
+/// versions"), and must still replicate to completion instead of staying
+/// `PENDING`.
+#[tokio::test]
+async fn test_bucket_replication_replicates_directory_marker_in_versioned_bucket() -> TestResult {
+    init_logging();
+
+    let mut source_env = RustFSTestEnvironment::new().await?;
+    let mut source_env_vars = replication_fast_env();
+    source_env_vars.extend_from_slice(LOOPBACK_REPLICATION_TARGET_ENV);
+    source_env.start_rustfs_server_with_env(vec![], &source_env_vars).await?;
+
+    let mut target_env = RustFSTestEnvironment::new().await?;
+    target_env.start_rustfs_server_without_cleanup(vec![]).await?;
+
+    let source_bucket = "replication-dir-marker-src";
+    let target_bucket = "replication-dir-marker-dst";
+    let source_client = source_env.create_s3_client();
+    let target_client = target_env.create_s3_client();
+
+    source_client.create_bucket().bucket(source_bucket).send().await?;
+    target_client.create_bucket().bucket(target_bucket).send().await?;
+    enable_bucket_versioning(&source_env, source_bucket).await?;
+    enable_bucket_versioning(&target_env, target_bucket).await?;
+    let target_arn = set_replication_target(&source_env, source_bucket, &target_env, target_bucket).await?;
+    put_bucket_replication(&source_env, source_bucket, &target_arn).await?;
+
+    let marker_key = "dir/trailing/";
+    let body = b"directory marker body";
+    let put = source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(marker_key)
+        .body(ByteStream::from_static(body))
+        .send()
+        .await?;
+    assert!(
+        put.version_id()
+            .is_none_or(|id| id == "null" || id == uuid::Uuid::nil().to_string()),
+        "a directory marker is the null version even in a versioned bucket: {:?}",
+        put.version_id()
+    );
+
+    wait_for_source_replication_status(&source_client, source_bucket, marker_key, "COMPLETED", false).await?;
+
+    let replica = target_client
+        .get_object()
+        .bucket(target_bucket)
+        .key(marker_key)
+        .send()
+        .await?;
+    assert_eq!(replica.body.collect().await?.into_bytes().as_ref(), body);
+    let listed = target_client
+        .list_object_versions()
+        .bucket(target_bucket)
+        .prefix(marker_key)
+        .send()
+        .await?;
+    let marker_versions: Vec<_> = listed.versions().iter().filter(|v| v.key() == Some(marker_key)).collect();
+    assert_eq!(marker_versions.len(), 1, "the marker must land exactly once: {marker_versions:?}");
+    assert_eq!(
+        marker_versions[0].version_id(),
+        Some("null"),
+        "the replica keeps the null version identity"
+    );
+
+    Ok(())
+}
+
+/// Regression for rustfs/backlog#2340 (not Wasabi specific): permanently
+/// deleting a version whose payload lives in a data dir must leave the source
+/// clean once the purge replicates. Managed-SSE objects are never inlined and a
+/// plain object above the inline threshold takes the same layout. The version
+/// retained with a pending purge used to lose its data dir, so the purge state
+/// could never be applied (`VersionNotFound` on every retry) and the bucket
+/// stayed `BucketNotEmpty` while `ListObjectVersions` was already empty.
+#[tokio::test]
+async fn test_bucket_replication_version_purge_of_non_inline_object_releases_source_bucket() -> TestResult {
+    init_logging();
+
+    let (source_env, target_env, source_bucket, target_bucket) = build_sse_replication_pair("purge-datadir", true, true).await?;
+    let target_arn = wait_for_remote_target_arn(&source_env, &source_bucket).await?;
+    put_bucket_replication_with_delete_statuses(&source_env, &source_bucket, &target_arn, "Enabled", Some("Enabled")).await?;
+    let source_client = source_env.create_s3_client();
+    let target_client = target_env.create_s3_client();
+
+    let sse_key = "sse-object.bin";
+    let large_key = "large-object.bin";
+    let sse_put = source_client
+        .put_object()
+        .bucket(&source_bucket)
+        .key(sse_key)
+        .body(ByteStream::from_static(b"encrypted source payload"))
+        .server_side_encryption(ServerSideEncryption::Aes256)
+        .send()
+        .await?;
+    let large_put = source_client
+        .put_object()
+        .bucket(&source_bucket)
+        .key(large_key)
+        .body(ByteStream::from(vec![0x5a; 2 * 1024 * 1024]))
+        .send()
+        .await?;
+    let purged = [
+        (sse_key, sse_put.version_id().ok_or("SSE PUT omitted version ID")?.to_string()),
+        (large_key, large_put.version_id().ok_or("large PUT omitted version ID")?.to_string()),
+    ];
+    assert_replication_converged(&source_client, &source_bucket, &target_client, &target_bucket).await?;
+
+    for (key, version_id) in &purged {
+        source_client
+            .delete_object()
+            .bucket(&source_bucket)
+            .key(*key)
+            .version_id(version_id)
+            .send()
+            .await?;
+    }
+    assert_replication_converged(&source_client, &source_bucket, &target_client, &target_bucket).await?;
+    let target_state = list_replication_state(&target_client, &target_bucket).await?;
+    assert!(target_state.is_empty(), "target retained an explicitly purged version: {target_state:?}");
+
+    // The purge state is applied on the source asynchronously after the target
+    // acknowledges the delete; only then does the retained version go away and
+    // the bucket become deletable. A listing that is empty while DeleteBucket
+    // keeps answering BucketNotEmpty is exactly the regression.
+    let deadline = tokio::time::Instant::now() + Duration::from_secs(60);
+    loop {
+        let listing = source_client.list_object_versions().bucket(&source_bucket).send().await?;
+        let listed = listing.versions().len() + listing.delete_markers().len();
+        match source_client.delete_bucket().bucket(&source_bucket).send().await {
+            Ok(_) => break,
+            Err(err) if err.code() == Some("BucketNotEmpty") => {
+                if tokio::time::Instant::now() >= deadline {
+                    return Err(format!(
+                        "source bucket stayed BucketNotEmpty after the version purge replicated; \
+                         ListObjectVersions shows {listed} entries"
+                    )
+                    .into());
+                }
+                sleep(Duration::from_millis(500)).await;
+            }
+            Err(err) => return Err(err.into()),
+        }
+    }
+
+    Ok(())
+}
+
+/// Regression for rustfs/backlog#2340 (not Wasabi specific): a single-part
+/// object uploaded with `x-amz-checksum-*` must reach the target with the same
+/// checksum. The outbound options keyed the stored record by algorithm name,
+/// which the target client sent as `x-amz-meta-*` user metadata, so a replica
+/// never carried a checksum although the source HEAD returned one.
+#[tokio::test]
+async fn test_bucket_replication_forwards_single_part_object_checksums() -> TestResult {
+    init_logging();
+
+    let mut source_env = RustFSTestEnvironment::new().await?;
+    let mut source_env_vars = replication_fast_env();
+    source_env_vars.extend_from_slice(LOOPBACK_REPLICATION_TARGET_ENV);
+    source_env.start_rustfs_server_with_env(vec![], &source_env_vars).await?;
+
+    let mut target_env = RustFSTestEnvironment::new().await?;
+    target_env.start_rustfs_server_without_cleanup(vec![]).await?;
+
+    let source_bucket = "replication-checksum-src";
+    let target_bucket = "replication-checksum-dst";
+    let source_client = source_env.create_s3_client();
+    let target_client = target_env.create_s3_client();
+
+    source_client.create_bucket().bucket(source_bucket).send().await?;
+    target_client.create_bucket().bucket(target_bucket).send().await?;
+    enable_bucket_versioning(&source_env, source_bucket).await?;
+    enable_bucket_versioning(&target_env, target_bucket).await?;
+    let target_arn = set_replication_target(&source_env, source_bucket, &target_env, target_bucket).await?;
+    put_bucket_replication(&source_env, source_bucket, &target_arn).await?;
+
+    let body = b"123456789";
+    let crc32_key = "checksum-crc32.txt";
+    let sha256_key = "checksum-sha256.txt";
+    let crc32_put = source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(crc32_key)
+        .body(ByteStream::from_static(body))
+        .checksum_algorithm(aws_sdk_s3::types::ChecksumAlgorithm::Crc32)
+        .send()
+        .await?;
+    let expected_crc32 = crc32_put.checksum_crc32().ok_or("source PUT omitted CRC32")?.to_string();
+    let sha256_put = source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(sha256_key)
+        .body(ByteStream::from_static(body))
+        .checksum_algorithm(aws_sdk_s3::types::ChecksumAlgorithm::Sha256)
+        .send()
+        .await?;
+    let expected_sha256 = sha256_put.checksum_sha256().ok_or("source PUT omitted SHA256")?.to_string();
+
+    for key in [crc32_key, sha256_key] {
+        wait_for_source_replication_status(&source_client, source_bucket, key, "COMPLETED", false).await?;
+    }
+
+    let replica = target_client
+        .head_object()
+        .bucket(target_bucket)
+        .key(crc32_key)
+        .checksum_mode(aws_sdk_s3::types::ChecksumMode::Enabled)
+        .send()
+        .await?;
+    assert_eq!(replica.checksum_crc32(), Some(expected_crc32.as_str()), "replica lost the CRC32 checksum");
+    let replica = target_client
+        .head_object()
+        .bucket(target_bucket)
+        .key(sha256_key)
+        .checksum_mode(aws_sdk_s3::types::ChecksumMode::Enabled)
+        .send()
+        .await?;
+    assert_eq!(
+        replica.checksum_sha256(),
+        Some(expected_sha256.as_str()),
+        "replica lost the SHA256 checksum"
+    );
+    // The bare algorithm name must not leak as user metadata either.
+    assert!(
+        replica
+            .metadata()
+            .is_none_or(|meta| !meta.keys().any(|k| k.eq_ignore_ascii_case("sha256"))),
+        "replica carries the checksum as user metadata: {:?}",
+        replica.metadata()
+    );
+
+    Ok(())
+}
+
 #[tokio::test]
 async fn test_bucket_replication_disabled_delete_marker_does_not_propagate() -> TestResult {
     init_logging();
@@ -8817,9 +9055,11 @@ async fn test_replication_check_flags_multipart_only_version_minting_target() ->
             .is_some_and(|error| error.contains("CreateMultipartUpload")),
         "the failure must name the multipart path: {payload}"
     );
-    // The PutObject leg mirrored, so it is the multipart probe that failed.
+    // The PutObject leg mirrored, so it is the multipart probe that failed;
+    // the mutation phases address the id the PUT reported and still run.
     assert_eq!(target_report["Phases"]["Put"]["Status"], "OK", "{payload}");
-    assert_eq!(target_report["Phases"]["DeleteMarker"]["Status"], "SKIPPED", "{payload}");
+    assert_eq!(target_report["Phases"]["DeleteMarker"]["Status"], "OK", "{payload}");
+    assert_eq!(target_report["Phases"]["VersionDelete"]["Status"], "OK", "{payload}");
     assert_eq!(target_report["Phases"]["Cleanup"]["Status"], "OK", "{payload}");
 
     let probe_key = target
@@ -9007,6 +9247,9 @@ async fn test_replication_check_flags_version_minting_target() -> TestResult {
     let target_bucket = "version-fidelity-dst";
     target.create_bucket(target_bucket);
     target.assign_own_version_ids(true);
+    // Wasabi shape: the probe version the VersionDelete phase removed answers
+    // NoSuchVersion to cleanup's second DELETE, which must count as clean.
+    target.reject_unknown_version_deletes(true);
 
     let mut source_env = RustFSTestEnvironment::new().await?;
     let mut env_vars = replication_fast_env();
@@ -9051,11 +9294,13 @@ async fn test_replication_check_flags_version_minting_target() -> TestResult {
         fidelity["Code"], "BucketRemoteTargetVersionMismatch",
         "the failure must carry a machine-readable code: {payload}"
     );
-    // The probe PUT itself succeeded (fidelity is judged from its response);
+    // The probe PUT itself succeeded (fidelity is judged from its response).
-    // the later mutation phases are pointless against a drifting target and
+    // The mutation phases address the id the target assigned — the ledger
-    // must be skipped, but cleanup still runs.
+    // the worker records per object (rustfs/backlog#2340) — so they run and
+    // pass on a drifting target, and cleanup uses the same id.
     assert_eq!(target_report["Phases"]["Put"]["Status"], "OK", "{payload}");
-    assert_eq!(target_report["Phases"]["DeleteMarker"]["Status"], "SKIPPED", "{payload}");
+    assert_eq!(target_report["Phases"]["DeleteMarker"]["Status"], "OK", "{payload}");
+    assert_eq!(target_report["Phases"]["VersionDelete"]["Status"], "OK", "{payload}");
     assert_eq!(target_report["Phases"]["Cleanup"]["Status"], "OK", "{payload}");
 
     // The probe PUT must carry the source version as `?versionId=` — the
@@ -9945,3 +10190,199 @@ async fn test_get_object_tagging_proxies_unreplicated_object_to_replication_targ
     target.shutdown().await;
     Ok(())
 }
+
+// ---------------------------------------------------------------------------
+// backlog#2363
+// ---------------------------------------------------------------------------
+
+/// Wait until the source reports a terminal replication status for `key`.
+async fn wait_terminal_replication_status(
+    client: &Client,
+    bucket: &str,
+    key: &str,
+    ssec: bool,
+    timeout: Duration,
+) -> Result<String, Box<dyn Error + Send + Sync>> {
+    let customer_key = BASE64_STANDARD.encode_to_string(REPL17_SSEC_KEY);
+    let customer_key_md5 = sse_customer_key_md5_base64(REPL17_SSEC_KEY);
+    let deadline = tokio::time::Instant::now() + timeout;
+    loop {
+        let request = client.head_object().bucket(bucket).key(key);
+        let head = if ssec {
+            request
+                .sse_customer_algorithm("AES256")
+                .sse_customer_key(&customer_key)
+                .sse_customer_key_md5(&customer_key_md5)
+                .send()
+                .await?
+        } else {
+            request.send().await?
+        };
+        let status = head.replication_status().map(|status| status.as_str().to_string());
+        if matches!(status.as_deref(), Some("COMPLETED") | Some("FAILED")) {
+            return Ok(status.unwrap_or_default());
+        }
+        if tokio::time::Instant::now() >= deadline {
+            return Err(format!("{bucket}/{key}: replication never reached a terminal status; last {status:?}").into());
+        }
+        sleep(Duration::from_millis(250)).await;
+    }
+}
+
+/// backlog#2363: SSE-C ciphertext passthrough of objects the source stored
+/// compressed. The replica on a RustFS target must decrypt to the original
+/// bytes for a single PUT and for a multipart upload.
+#[tokio::test]
+async fn test_bucket_replication_sse_c_compressed_passthrough() -> TestResult {
+    init_logging();
+    const PART_SIZE: usize = 5 * 1024 * 1024;
+
+    let mut source_env = RustFSTestEnvironment::new().await?;
+    let mut target_env = RustFSTestEnvironment::new().await?;
+    let mut source_process_env = replication_fast_env();
+    source_process_env.extend_from_slice(LOOPBACK_REPLICATION_TARGET_ENV);
+    source_process_env.extend_from_slice(FAST_SCANNER_ENV);
+    source_process_env.extend_from_slice(&[
+        ("NO_PROXY", "127.0.0.1,localhost"),
+        ("HTTP_PROXY", ""),
+        ("HTTPS_PROXY", ""),
+        ("RUSTFS_COMPRESSION_ENABLED", "true"),
+        ("RUSTFS_COMPRESSION_MULTIPART_ENABLED", "true"),
+    ]);
+    source_env.start_rustfs_server_with_env(vec![], &source_process_env).await?;
+    target_env
+        .start_rustfs_server_without_cleanup_with_env(&[
+            ("NO_PROXY", "127.0.0.1,localhost"),
+            ("HTTP_PROXY", ""),
+            ("HTTPS_PROXY", ""),
+        ])
+        .await?;
+
+    let source_bucket = "ssec-compressed-src";
+    let target_bucket = "ssec-compressed-dst";
+    let source_client = source_env.create_s3_client();
+    let target_client = target_env.create_s3_client();
+    source_client.create_bucket().bucket(source_bucket).send().await?;
+    target_client.create_bucket().bucket(target_bucket).send().await?;
+    enable_bucket_versioning(&source_env, source_bucket).await?;
+    enable_bucket_versioning(&target_env, target_bucket).await?;
+    let target_arn = set_replication_target(&source_env, source_bucket, &target_env, target_bucket).await?;
+    put_bucket_replication(&source_env, source_bucket, &target_arn).await?;
+
+    let customer_key = BASE64_STANDARD.encode_to_string(REPL17_SSEC_KEY);
+    let customer_key_md5 = sse_customer_key_md5_base64(REPL17_SSEC_KEY);
+    let text = |len: usize, seed: u32| -> Vec<u8> {
+        let mut out = Vec::with_capacity(len + 64);
+        let mut line = 0u64;
+        while out.len() < len {
+            out.extend_from_slice(format!("ssec compressed passthrough seed={seed} line={line} lorem ipsum dolor\n").as_bytes());
+            line += 1;
+        }
+        out.truncate(len);
+        out
+    };
+
+    let single_key = "ssec-compressed-single.txt";
+    let single_body = text(1024 * 1024 + 17, 1);
+    source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(single_key)
+        .content_type("text/plain")
+        .body(ByteStream::from(single_body.clone()))
+        .sse_customer_algorithm("AES256")
+        .sse_customer_key(&customer_key)
+        .sse_customer_key_md5(&customer_key_md5)
+        .send()
+        .await?;
+
+    let multipart_key = "ssec-compressed-multipart.txt";
+    let multipart_parts = [text(PART_SIZE, 2), text(1024 * 1024 + 4096, 3)];
+    let multipart_body: Vec<u8> = multipart_parts.concat();
+    let created = source_client
+        .create_multipart_upload()
+        .bucket(source_bucket)
+        .key(multipart_key)
+        .content_type("text/plain")
+        .sse_customer_algorithm("AES256")
+        .sse_customer_key(&customer_key)
+        .sse_customer_key_md5(&customer_key_md5)
+        .send()
+        .await?;
+    let upload_id = created.upload_id().ok_or("missing multipart upload id")?.to_string();
+    let mut completed = Vec::new();
+    for (index, part) in multipart_parts.iter().enumerate() {
+        let part_number = i32::try_from(index + 1)?;
+        let uploaded = source_client
+            .upload_part()
+            .bucket(source_bucket)
+            .key(multipart_key)
+            .upload_id(&upload_id)
+            .part_number(part_number)
+            .body(ByteStream::from(part.clone()))
+            .sse_customer_algorithm("AES256")
+            .sse_customer_key(&customer_key)
+            .sse_customer_key_md5(&customer_key_md5)
+            .send()
+            .await?;
+        completed.push(
+            CompletedPart::builder()
+                .part_number(part_number)
+                .set_e_tag(uploaded.e_tag().map(str::to_string))
+                .build(),
+        );
+    }
+    source_client
+        .complete_multipart_upload()
+        .bucket(source_bucket)
+        .key(multipart_key)
+        .upload_id(&upload_id)
+        .multipart_upload(CompletedMultipartUpload::builder().set_parts(Some(completed)).build())
+        .sse_customer_algorithm("AES256")
+        .sse_customer_key(&customer_key)
+        .sse_customer_key_md5(&customer_key_md5)
+        .send()
+        .await?;
+
+    let mut failures = Vec::new();
+    for (key, body) in [(single_key, &single_body), (multipart_key, &multipart_body)] {
+        let status = wait_terminal_replication_status(&source_client, source_bucket, key, true, Duration::from_secs(120)).await?;
+        if status != "COMPLETED" {
+            failures.push(format!("{key}: source reports {status}"));
+            continue;
+        }
+        let replica = target_client
+            .get_object()
+            .bucket(target_bucket)
+            .key(key)
+            .sse_customer_algorithm("AES256")
+            .sse_customer_key(&customer_key)
+            .sse_customer_key_md5(&customer_key_md5)
+            .send()
+            .await;
+        match replica {
+            Ok(replica) => {
+                let content_length = replica.content_length();
+                match replica.body.collect().await {
+                    Ok(collected) => {
+                        let bytes = collected.into_bytes();
+                        if bytes.as_ref() != body.as_slice() {
+                            failures.push(format!(
+                                "{key}: replica bytes differ (content_length={content_length:?}, got {} bytes, want {})",
+                                bytes.len(),
+                                body.len()
+                            ));
+                        }
+                    }
+                    Err(err) => failures.push(format!("{key}: replica body read failed: {err}")),
+                }
+            }
+            Err(err) => failures.push(format!("{key}: replica GET failed: {err}")),
+        }
+    }
+    assert!(
+        failures.is_empty(),
+        "SSE-C compressed passthrough replicas must decrypt to the source bytes: {failures:?}"
+    );
+    Ok(())
+}

crates/e2e_test/src/replication_target_matrix_test.rs

@@ -31,17 +31,22 @@
 //! Adding a target behavior the fleet has shown: add the mode to the fake
 //! target, add a row here, and record any cell that is red before the fix.
 
-use crate::common::{RustFSTestEnvironment, init_logging, replication_fast_env};
+use crate::common::{init_logging, replication_fast_env};
-use crate::fake_s3_target::{FAKE_ACCESS_KEY, FAKE_SECRET_KEY};
+use crate::fake_s3_target::{BucketMode, FAKE_ACCESS_KEY, FAKE_SECRET_KEY};
-use crate::fake_s3_target::{FakeS3Target, Operation as FakeTargetOperation, RequestRecord};
+use crate::fake_s3_target::{FakeS3Target, FaultAction as FakeTargetFault, Operation as FakeTargetOperation, RequestRecord};
-use crate::on_demand_migration::common::fake_source_client;
+use crate::on_demand_migration::common::{OdmEnvOptions, OdmTestEnv, fake_source_client};
 use crate::replication_extension_test::{
-    LOOPBACK_REPLICATION_TARGET_ENV, ReplicationTargetOptions, enable_bucket_versioning, put_bucket_replication,
+    LOOPBACK_REPLICATION_TARGET_ENV, ReplicationTargetOptions, delete_bucket_replication, enable_bucket_versioning,
-    set_replication_target_with_options,
+    get_replication_reset_status, put_bucket_replication, put_bucket_replication_with_delete_statuses,
+    set_replication_target_with_options, start_bucket_replication_reset,
 };
 use aws_sdk_s3::Client;
+use aws_sdk_s3::error::ProvideErrorMetadata;
 use aws_sdk_s3::primitives::{ByteStream, DateTime};
-use aws_sdk_s3::types::{CompletedMultipartUpload, CompletedPart, ObjectLockLegalHoldStatus, ObjectLockMode};
+use aws_sdk_s3::types::{
+    Checksum, ChecksumAlgorithm, CompletedMultipartUpload, CompletedPart, ObjectAttributes, ObjectLockLegalHold,
+    ObjectLockLegalHoldStatus, ObjectLockMode, ObjectLockRetention, ObjectLockRetentionMode, Tag, Tagging,
+};
 use bytes::Bytes;
 use std::error::Error;
 use std::time::{SystemTime, UNIX_EPOCH};
@@ -63,7 +68,10 @@ enum TargetMode {
     /// Object Lock parameters must carry `Content-MD5` or `x-amz-checksum-*`.
     RequireChecksumWithObjectLock,
     /// AWS S3 / Wasabi / Impossible Cloud: mints its own version ids
-    /// (rustfs/backlog#2085). Data must still land.
+    /// (rustfs/backlog#2085) and, like Wasabi, answers NoSuchVersion to a
+    /// DELETE of an id it never had (rustfs/backlog#2340). Data must still
+    /// land, and every version-addressed mutation must resolve the replica
+    /// through the target-version ledger.
     MintOwnVersionIds,
 }
 
@@ -80,7 +88,10 @@ impl TargetMode {
             TargetMode::Baseline => {}
             TargetMode::RejectAwsChunked => target.reject_aws_chunked_uploads(true),
             TargetMode::RequireChecksumWithObjectLock => target.require_checksum_for_object_lock(true),
-            TargetMode::MintOwnVersionIds => target.assign_own_version_ids(true),
+            TargetMode::MintOwnVersionIds => {
+                target.assign_own_version_ids(true);
+                target.reject_unknown_version_deletes(true);
+            }
         }
     }
 
@@ -110,16 +121,23 @@ enum ObjectShape {
     /// Two-part multipart upload with a GOVERNANCE retention period; the
     /// lock headers travel on CreateMultipartUpload, which has no body.
     LockedMultipart,
+    /// ODM stores two local parts while preserving a single-PUT source's MD5 ETag.
+    OdmPreservedMd5Multipart,
+    /// Single-part object uploaded with `x-amz-checksum-sha256`; the replica
+    /// must carry the same header (rustfs/backlog#2340).
+    Checksummed,
 }
 
 impl ObjectShape {
-    const ALL: [ObjectShape; 6] = [
+    const ALL: [ObjectShape; 8] = [
         ObjectShape::Empty,
         ObjectShape::Plain,
         ObjectShape::Retention,
         ObjectShape::LegalHold,
         ObjectShape::Multipart,
         ObjectShape::LockedMultipart,
+        ObjectShape::OdmPreservedMd5Multipart,
+        ObjectShape::Checksummed,
     ];
 
     fn key(self) -> &'static str {
@@ -130,6 +148,17 @@ impl ObjectShape {
             ObjectShape::LegalHold => "matrix/legal-hold.bin",
             ObjectShape::Multipart => "matrix/multipart.bin",
             ObjectShape::LockedMultipart => "matrix/locked-multipart.bin",
+            ObjectShape::OdmPreservedMd5Multipart => "matrix/odm-preserved-md5.bin",
+            ObjectShape::Checksummed => "matrix/checksummed.bin",
+        }
+    }
+
+    /// The `x-amz-checksum-*` header the source stored and every upload of
+    /// the replica must repeat.
+    fn forwarded_checksum_header(self) -> Option<&'static str> {
+        match self {
+            ObjectShape::Checksummed => Some("x-amz-checksum-sha256"),
+            _ => None,
         }
     }
 
@@ -139,7 +168,8 @@ impl ObjectShape {
 
     /// Upload the shape to the source and return the bytes the target must
     /// end up holding.
-    async fn put(self, client: &Client, bucket: &str) -> Result<Bytes, Box<dyn Error + Send + Sync>> {
+    async fn put(self, env: &OdmTestEnv, bucket: &str) -> Result<Bytes, Box<dyn Error + Send + Sync>> {
+        let client = &env.client;
         let key = self.key();
         match self {
             ObjectShape::Empty => {
@@ -190,6 +220,19 @@ impl ObjectShape {
             }
             ObjectShape::Multipart => multipart_put(client, bucket, key, 0x44, false).await,
             ObjectShape::LockedMultipart => multipart_put(client, bucket, key, 0x55, true).await,
+            ObjectShape::OdmPreservedMd5Multipart => odm_preserved_md5_multipart(env, bucket, key).await,
+            ObjectShape::Checksummed => {
+                let body = payload(40 * 1024, 0x66);
+                client
+                    .put_object()
+                    .bucket(bucket)
+                    .key(key)
+                    .body(ByteStream::from(body.clone()))
+                    .checksum_algorithm(ChecksumAlgorithm::Sha256)
+                    .send()
+                    .await?;
+                Ok(body)
+            }
         }
     }
 }
@@ -219,6 +262,595 @@ fn expectation(mode: TargetMode, shape: ObjectShape) -> Expectation {
         .unwrap_or(Expectation::Completed)
 }
 
+/// rustfs/backlog#2340: a target that mints its own version ids (Wasabi,
+/// AWS S3) answers 404 to a HEAD by the source uuid, which the worker used to
+/// read as "replica missing" and re-drive the PUT — one more target version
+/// per heal, MRF retry or resync. Two re-drive shapes, both must converge on
+/// the single version the first PUT created:
+/// - the first PUT lands but its response is lost, so the object is FAILED
+///   and the scanner heal pass re-drives it;
+/// - an existing-object resync re-drives a COMPLETED object unconditionally.
+#[tokio::test]
+async fn matrix_mint_own_version_ids_redrive_does_not_duplicate() -> TestResult {
+    init_logging();
+
+    let target = FakeS3Target::start().await?;
+    let target_bucket = "matrix-mint-own-redrive-dst".to_string();
+    target.create_bucket_with_object_lock(target_bucket.clone());
+    target.assign_own_version_ids(true);
+
+    let mut env_vars = replication_fast_env();
+    env_vars.extend_from_slice(LOOPBACK_REPLICATION_TARGET_ENV);
+    env_vars.extend_from_slice(&[
+        ("NO_PROXY", "127.0.0.1,localhost"),
+        ("HTTP_PROXY", ""),
+        ("HTTPS_PROXY", ""),
+        // The scanner heal pass is what re-drives a FAILED object.
+        ("RUSTFS_SCANNER_CYCLE", "1"),
+        ("RUSTFS_SCANNER_START_DELAY_SECS", "1"),
+    ]);
+    let env = OdmTestEnv::start_with(OdmEnvOptions {
+        env: env_vars,
+        ..OdmEnvOptions::default()
+    })
+    .await?;
+    let source_env = &env.rustfs;
+
+    let source_bucket = "matrix-mint-own-redrive-src";
+    let source_client = source_env.create_s3_client();
+    source_client
+        .create_bucket()
+        .bucket(source_bucket)
+        .object_lock_enabled_for_bucket(true)
+        .send()
+        .await?;
+    enable_bucket_versioning(source_env, source_bucket).await?;
+    let target_arn = set_replication_target_with_options(
+        source_env,
+        source_bucket,
+        ReplicationTargetOptions {
+            endpoint: &target.address(),
+            access_key: FAKE_ACCESS_KEY,
+            secret_key: FAKE_SECRET_KEY,
+            target_bucket: &target_bucket,
+            secure: false,
+            skip_tls_verify: false,
+            ca_cert_pem: None,
+        },
+    )
+    .await?;
+    put_bucket_replication(source_env, source_bucket, &target_arn).await?;
+
+    // Teach the worker the target's identity contract with one ordinary
+    // write, exactly as production learns it (the PUT response carries the
+    // minted id).
+    let probe_key = "redrive/identity-probe.bin";
+    source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(probe_key)
+        .body(ByteStream::from(payload(4 * 1024, 0x01)))
+        .send()
+        .await?;
+    assert_eq!(
+        wait_for_terminal_replication_status(&source_client, source_bucket, probe_key).await?,
+        "COMPLETED"
+    );
+
+    // Shape 1: the PUT is stored, its response never arrives, heal re-drives.
+    let heal_key = "redrive/heal.bin";
+    target.inject_for_key(FakeTargetOperation::PutObject, heal_key, FakeTargetFault::DisconnectAfterResponse, 1);
+    source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(heal_key)
+        .body(ByteStream::from(payload(8 * 1024, 0x02)))
+        .send()
+        .await?;
+    wait_for_replication_status_and_single_version(&source_client, source_bucket, &target, &target_bucket, heal_key).await?;
+
+    // Shape 2: an existing-object resync re-drives a COMPLETED object.
+    let resync_key = "redrive/resync.bin";
+    source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(resync_key)
+        .body(ByteStream::from(payload(8 * 1024, 0x03)))
+        .send()
+        .await?;
+    assert_eq!(
+        wait_for_terminal_replication_status(&source_client, source_bucket, resync_key).await?,
+        "COMPLETED"
+    );
+    let (reset_arn, _reset_id) = start_bucket_replication_reset(source_env, source_bucket).await?;
+    assert_eq!(reset_arn, target_arn);
+    let resync = async {
+        loop {
+            let status = get_replication_reset_status(source_env, source_bucket, &target_arn).await?;
+            if let Some(entry) = status.targets.iter().find(|entry| entry.arn == target_arn)
+                && entry.status == "Completed"
+            {
+                return Ok::<_, Box<dyn Error + Send + Sync>>(entry.replicated_count);
+            }
+            sleep(Duration::from_millis(250)).await;
+        }
+    };
+    let replicated = timeout(Duration::from_secs(90), resync)
+        .await
+        .map_err(|_| "existing-object resync did not complete within 90 seconds")??;
+    assert!(replicated >= 3, "resync must count the located replicas as replicated, got {replicated}");
+    for key in [probe_key, heal_key, resync_key] {
+        let versions = target.stored_versions(&target_bucket, key);
+        assert_eq!(
+            versions.len(),
+            1,
+            "{key}: a re-drive against a target that mints its own version ids must not mint another one: {versions:?}"
+        );
+    }
+
+    target.shutdown().await;
+    Ok(())
+}
+
+/// rustfs/backlog#2340 (target-version ledger): on a target that mints its own
+/// version ids and answers NoSuchVersion to an unknown id (the Wasabi shape),
+/// every version-addressed mutation must land on the version the target
+/// assigned, which the replication PUT recorded on the source:
+/// - a tag update changes the existing target version, no new version;
+/// - a retention extension and legal hold ON/OFF change that version too;
+/// - a permanent delete of the older of two same-content generations removes
+///   exactly that replica and keeps the live one (content identity alone
+///   could not tell them apart).
+#[tokio::test]
+async fn matrix_mint_own_version_ids_addresses_mutations_through_the_ledger() -> TestResult {
+    init_logging();
+
+    let target = FakeS3Target::start().await?;
+    let target_bucket = "matrix-mint-own-ledger-dst".to_string();
+    target.create_bucket_with_object_lock(target_bucket.clone());
+    TargetMode::MintOwnVersionIds.apply(&target);
+
+    let mut env_vars = replication_fast_env();
+    env_vars.extend_from_slice(LOOPBACK_REPLICATION_TARGET_ENV);
+    env_vars.extend_from_slice(&[
+        ("NO_PROXY", "127.0.0.1,localhost"),
+        ("HTTP_PROXY", ""),
+        ("HTTPS_PROXY", ""),
+        // The scanner heal pass retries a purge the first attempt lost.
+        ("RUSTFS_SCANNER_CYCLE", "1"),
+        ("RUSTFS_SCANNER_START_DELAY_SECS", "1"),
+    ]);
+    let env = OdmTestEnv::start_with(OdmEnvOptions {
+        env: env_vars,
+        ..OdmEnvOptions::default()
+    })
+    .await?;
+    let source_env = &env.rustfs;
+
+    let source_bucket = "matrix-mint-own-ledger-src";
+    let source_client = source_env.create_s3_client();
+    source_client
+        .create_bucket()
+        .bucket(source_bucket)
+        .object_lock_enabled_for_bucket(true)
+        .send()
+        .await?;
+    enable_bucket_versioning(source_env, source_bucket).await?;
+    let target_arn = set_replication_target_with_options(
+        source_env,
+        source_bucket,
+        ReplicationTargetOptions {
+            endpoint: &target.address(),
+            access_key: FAKE_ACCESS_KEY,
+            secret_key: FAKE_SECRET_KEY,
+            target_bucket: &target_bucket,
+            secure: false,
+            skip_tls_verify: false,
+            ca_cert_pem: None,
+        },
+    )
+    .await?;
+    put_bucket_replication_with_delete_statuses(source_env, source_bucket, &target_arn, "Enabled", Some("Enabled")).await?;
+    let target_client = fake_source_client(&target);
+
+    // Tag update on an existing version.
+    let tag_key = "ledger/tags.bin";
+    let tagged = source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(tag_key)
+        .body(ByteStream::from(payload(4 * 1024, 0x01)))
+        .send()
+        .await?;
+    let tag_source_version = tagged.version_id().ok_or("source PUT returned no version id")?.to_string();
+    assert_eq!(
+        wait_for_terminal_replication_status(&source_client, source_bucket, tag_key).await?,
+        "COMPLETED"
+    );
+    let tag_target_version = single_target_version(&target, &target_bucket, tag_key)?;
+    source_client
+        .put_object_tagging()
+        .bucket(source_bucket)
+        .key(tag_key)
+        .version_id(&tag_source_version)
+        .tagging(
+            Tagging::builder()
+                .tag_set(Tag::builder().key("phase").value("after").build()?)
+                .build()?,
+        )
+        .send()
+        .await?;
+    wait_until("tag update on the existing target version", || async {
+        let tags = target_client
+            .get_object_tagging()
+            .bucket(&target_bucket)
+            .key(tag_key)
+            .version_id(&tag_target_version)
+            .send()
+            .await?;
+        Ok(tags
+            .tag_set()
+            .iter()
+            .any(|tag| tag.key() == "phase" && tag.value() == "after"))
+    })
+    .await?;
+    assert_stable_single_version(&target, &target_bucket, tag_key, &tag_target_version).await?;
+
+    // Retention extension and legal hold on an existing version.
+    let lock_key = "ledger/lock.bin";
+    let locked = source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(lock_key)
+        .body(ByteStream::from(payload(4 * 1024, 0x02)))
+        .object_lock_mode(ObjectLockMode::Governance)
+        .object_lock_retain_until_date(retain_until())
+        .send()
+        .await?;
+    let lock_source_version = locked.version_id().ok_or("source PUT returned no version id")?.to_string();
+    assert_eq!(
+        wait_for_terminal_replication_status(&source_client, source_bucket, lock_key).await?,
+        "COMPLETED"
+    );
+    let lock_target_version = single_target_version(&target, &target_bucket, lock_key)?;
+    let extended = DateTime::from_secs(retain_until().secs() + 86_400);
+    source_client
+        .put_object_retention()
+        .bucket(source_bucket)
+        .key(lock_key)
+        .version_id(&lock_source_version)
+        .retention(
+            ObjectLockRetention::builder()
+                .mode(ObjectLockRetentionMode::Governance)
+                .retain_until_date(extended)
+                .build(),
+        )
+        .send()
+        .await?;
+    source_client
+        .put_object_legal_hold()
+        .bucket(source_bucket)
+        .key(lock_key)
+        .version_id(&lock_source_version)
+        .legal_hold(ObjectLockLegalHold::builder().status(ObjectLockLegalHoldStatus::On).build())
+        .send()
+        .await?;
+    wait_until("retention extension and legal hold on the existing target version", || async {
+        let head = target_client
+            .head_object()
+            .bucket(&target_bucket)
+            .key(lock_key)
+            .version_id(&lock_target_version)
+            .send()
+            .await?;
+        Ok(head.object_lock_retain_until_date().map(|date| date.secs()) == Some(extended.secs())
+            && head.object_lock_legal_hold_status() == Some(&ObjectLockLegalHoldStatus::On))
+    })
+    .await?;
+    source_client
+        .put_object_legal_hold()
+        .bucket(source_bucket)
+        .key(lock_key)
+        .version_id(&lock_source_version)
+        .legal_hold(ObjectLockLegalHold::builder().status(ObjectLockLegalHoldStatus::Off).build())
+        .send()
+        .await?;
+    wait_until("legal hold removal on the existing target version", || async {
+        let head = target_client
+            .head_object()
+            .bucket(&target_bucket)
+            .key(lock_key)
+            .version_id(&lock_target_version)
+            .send()
+            .await?;
+        Ok(head.object_lock_legal_hold_status() == Some(&ObjectLockLegalHoldStatus::Off))
+    })
+    .await?;
+    assert_stable_single_version(&target, &target_bucket, lock_key, &lock_target_version).await?;
+
+    // Permanent delete of the older of two same-content generations.
+    let generations_key = "ledger/generations.bin";
+    let body = payload(4 * 1024, 0x03);
+    let older = source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(generations_key)
+        .body(ByteStream::from(body.clone()))
+        .send()
+        .await?;
+    let older_version = older.version_id().ok_or("source PUT returned no version id")?.to_string();
+    assert_eq!(
+        wait_for_terminal_replication_status(&source_client, source_bucket, generations_key).await?,
+        "COMPLETED"
+    );
+    let older_replica = single_target_version(&target, &target_bucket, generations_key)?;
+    source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(generations_key)
+        .body(ByteStream::from(body))
+        .send()
+        .await?;
+    assert_eq!(
+        wait_for_terminal_replication_status(&source_client, source_bucket, generations_key).await?,
+        "COMPLETED"
+    );
+    wait_until("both generations replicated", || async {
+        Ok(target.stored_versions(&target_bucket, generations_key).len() == 2)
+    })
+    .await?;
+    let newer_replica = target
+        .stored_versions(&target_bucket, generations_key)
+        .into_iter()
+        .map(|(version_id, _)| version_id)
+        .find(|version_id| version_id != &older_replica)
+        .ok_or("the second generation must have its own target version")?;
+
+    source_client
+        .delete_object()
+        .bucket(source_bucket)
+        .key(generations_key)
+        .version_id(&older_version)
+        .send()
+        .await?;
+    wait_until("permanent delete of the older generation's replica", || async {
+        let versions: Vec<String> = target
+            .stored_versions(&target_bucket, generations_key)
+            .into_iter()
+            .map(|(version_id, _)| version_id)
+            .collect();
+        Ok(versions == [newer_replica.clone()])
+    })
+    .await?;
+    assert_stable_single_version(&target, &target_bucket, generations_key, &newer_replica).await?;
+
+    // No mutation above may have gone out as a re-PUT: one upload per key.
+    for key in [tag_key, lock_key] {
+        let puts = target
+            .requests()
+            .iter()
+            .filter(|record| record.key.as_deref() == Some(key) && record.operation == FakeTargetOperation::PutObject)
+            .count();
+        assert_eq!(
+            puts, 1,
+            "{key}: a metadata update must not re-PUT the object on a target that mints its own ids"
+        );
+    }
+
+    target.shutdown().await;
+    Ok(())
+}
+
+/// rustfs/backlog#2340 (pending purge lifecycle): a permanent delete whose
+/// replication keeps failing leaves the version in xl.meta as a PENDING purge,
+/// hidden from listings. Once the bucket's replication configuration is
+/// removed nothing can ever confirm that purge remotely, so the delete worker
+/// must settle it locally (abandoned, with the replica left on the former
+/// target) — otherwise the bucket stays `BucketNotEmpty` forever with a
+/// residue the client cannot see.
+#[tokio::test]
+async fn matrix_removed_replication_config_abandons_pending_purge() -> TestResult {
+    init_logging();
+
+    let target = FakeS3Target::start().await?;
+    let target_bucket = "matrix-abandoned-purge-dst".to_string();
+    target.create_bucket_with_object_lock(target_bucket.clone());
+    TargetMode::MintOwnVersionIds.apply(&target);
+
+    let mut env_vars = replication_fast_env();
+    env_vars.extend_from_slice(LOOPBACK_REPLICATION_TARGET_ENV);
+    env_vars.extend_from_slice(&[
+        ("NO_PROXY", "127.0.0.1,localhost"),
+        ("HTTP_PROXY", ""),
+        ("HTTPS_PROXY", ""),
+        // The scanner heal pass is what revisits a pending purge.
+        ("RUSTFS_SCANNER_CYCLE", "1"),
+        ("RUSTFS_SCANNER_START_DELAY_SECS", "1"),
+    ]);
+    let env = OdmTestEnv::start_with(OdmEnvOptions {
+        env: env_vars,
+        ..OdmEnvOptions::default()
+    })
+    .await?;
+    let source_env = &env.rustfs;
+
+    let source_bucket = "matrix-abandoned-purge-src";
+    let source_client = source_env.create_s3_client();
+    source_client
+        .create_bucket()
+        .bucket(source_bucket)
+        .object_lock_enabled_for_bucket(true)
+        .send()
+        .await?;
+    enable_bucket_versioning(source_env, source_bucket).await?;
+    let target_arn = set_replication_target_with_options(
+        source_env,
+        source_bucket,
+        ReplicationTargetOptions {
+            endpoint: &target.address(),
+            access_key: FAKE_ACCESS_KEY,
+            secret_key: FAKE_SECRET_KEY,
+            target_bucket: &target_bucket,
+            secure: false,
+            skip_tls_verify: false,
+            ca_cert_pem: None,
+        },
+    )
+    .await?;
+    put_bucket_replication_with_delete_statuses(source_env, source_bucket, &target_arn, "Enabled", Some("Enabled")).await?;
+
+    let key = "purge/orphaned.bin";
+    let put = source_client
+        .put_object()
+        .bucket(source_bucket)
+        .key(key)
+        .body(ByteStream::from(payload(4 * 1024, 0x07)))
+        .send()
+        .await?;
+    let source_version = put.version_id().ok_or("source PUT returned no version id")?.to_string();
+    assert_eq!(
+        wait_for_terminal_replication_status(&source_client, source_bucket, key).await?,
+        "COMPLETED"
+    );
+    let replica = single_target_version(&target, &target_bucket, key)?;
+
+    // The target refuses every purge: the version stays a pending purge.
+    // More refusals than any scanner cycle can consume within the test.
+    target.inject_for_key(FakeTargetOperation::DeleteObject, key, FakeTargetFault::ResponseStatus(503), 4_000);
+    source_client
+        .delete_object()
+        .bucket(source_bucket)
+        .key(key)
+        .version_id(&source_version)
+        .send()
+        .await?;
+    wait_until("the refused purge to reach the target at least once", || async {
+        Ok(target.count_requests(FakeTargetOperation::DeleteObject, key) >= 1)
+    })
+    .await?;
+    let listed = source_client.list_object_versions().bucket(source_bucket).send().await?;
+    assert!(
+        listed.versions().is_empty() && listed.delete_markers().is_empty(),
+        "a pending purge is hidden from listings: {listed:?}"
+    );
+    let blocked = source_client.delete_bucket().bucket(source_bucket).send().await;
+    assert!(
+        blocked
+            .as_ref()
+            .err()
+            .and_then(|err| err.as_service_error())
+            .is_some_and(|err| err.code() == Some("BucketNotEmpty")),
+        "the hidden pending purge must block DeleteBucket while the target is still configured: {blocked:?}"
+    );
+
+    // Removing the replication configuration orphans the purge; the scanner
+    // heal pass must settle it locally so the bucket becomes deletable.
+    let response = delete_bucket_replication(source_env, source_bucket).await?;
+    assert!(response.status().is_success(), "DeleteBucketReplication: {}", response.status());
+    wait_until("DeleteBucket to succeed once the orphaned purge is abandoned", || async {
+        match source_client.delete_bucket().bucket(source_bucket).send().await {
+            Ok(_) => Ok(true),
+            Err(err) if err.as_service_error().is_some_and(|err| err.code() == Some("BucketNotEmpty")) => Ok(false),
+            Err(err) => Err(err.into()),
+        }
+    })
+    .await?;
+    // Abandoned means abandoned: the replica stays on the former target and,
+    // once the attempts in flight at removal time have drained, no further
+    // purge attempts are sent to it.
+    assert_eq!(
+        single_target_version(&target, &target_bucket, key)?,
+        replica,
+        "an abandoned purge must not touch the replica on the former target"
+    );
+    sleep(Duration::from_secs(3)).await;
+    let settled = target.count_requests(FakeTargetOperation::DeleteObject, key);
+    sleep(Duration::from_secs(3)).await;
+    assert_eq!(
+        target.count_requests(FakeTargetOperation::DeleteObject, key),
+        settled,
+        "purge attempts must stop once the target is no longer configured"
+    );
+
+    target.shutdown().await;
+    Ok(())
+}
+
+fn single_target_version(target: &FakeS3Target, target_bucket: &str, key: &str) -> Result<String, Box<dyn Error + Send + Sync>> {
+    let versions = target.stored_versions(target_bucket, key);
+    match versions.as_slice() {
+        [(version_id, false)] => Ok(version_id.clone()),
+        other => Err(format!("{key}: expected exactly one live target version, got {other:?}").into()),
+    }
+}
+
+/// The target keeps holding exactly `version_id` for a few scanner cycles: a
+/// re-driven PUT or a wrong delete would show up here.
+async fn assert_stable_single_version(target: &FakeS3Target, target_bucket: &str, key: &str, version_id: &str) -> TestResult {
+    for _ in 0..8 {
+        let versions = target.stored_versions(target_bucket, key);
+        if versions.len() != 1 || versions[0].0 != version_id {
+            return Err(
+                format!("{key}: target versions drifted from the single expected replica {version_id}: {versions:?}").into(),
+            );
+        }
+        sleep(Duration::from_millis(500)).await;
+    }
+    Ok(())
+}
+
+async fn wait_until<F, Fut>(what: &str, mut probe: F) -> TestResult
+where
+    F: FnMut() -> Fut,
+    Fut: std::future::Future<Output = Result<bool, Box<dyn Error + Send + Sync>>>,
+{
+    let wait = async {
+        loop {
+            if probe().await? {
+                return Ok::<_, Box<dyn Error + Send + Sync>>(());
+            }
+            sleep(Duration::from_millis(250)).await;
+        }
+    };
+    timeout(Duration::from_secs(90), wait)
+        .await
+        .map_err(|_| format!("{what} did not happen within 90 seconds"))?
+}
+
+/// Wait until `key` is COMPLETED on the source and, for the observation
+/// window after that, the target still holds exactly one live version of it.
+async fn wait_for_replication_status_and_single_version(
+    source_client: &Client,
+    source_bucket: &str,
+    target: &FakeS3Target,
+    target_bucket: &str,
+    key: &str,
+) -> TestResult {
+    // The lost PUT response first settles the object FAILED; only the next
+    // scanner heal pass can turn that into COMPLETED, so FAILED is transient
+    // here and the wait is for COMPLETED alone.
+    let converged = async {
+        loop {
+            let head = source_client.head_object().bucket(source_bucket).key(key).send().await?;
+            if head.replication_status().is_some_and(|status| status.as_str() == "COMPLETED") {
+                return Ok::<_, Box<dyn Error + Send + Sync>>(());
+            }
+            sleep(Duration::from_millis(250)).await;
+        }
+    };
+    timeout(Duration::from_secs(90), converged)
+        .await
+        .map_err(|_| format!("{key}: heal re-drive did not converge to COMPLETED within 90 seconds"))??;
+    // The heal pass keeps visiting the key for a few scanner cycles; a
+    // duplicate would show up here as a second stored version.
+    for _ in 0..12 {
+        let versions = target.stored_versions(target_bucket, key);
+        assert_eq!(versions.len(), 1, "{key}: target minted another version on re-drive: {versions:?}");
+        sleep(Duration::from_millis(500)).await;
+    }
+    Ok(())
+}
+
 #[tokio::test]
 async fn matrix_baseline_target() -> TestResult {
     run_row(TargetMode::Baseline).await
@@ -270,11 +902,15 @@ async fn run_row(mode: TargetMode) -> TestResult {
     target.create_bucket_with_object_lock(target_bucket.clone());
     mode.apply(&target);
 
-    let mut source_env = RustFSTestEnvironment::new().await?;
     let mut env_vars = replication_fast_env();
     env_vars.extend_from_slice(LOOPBACK_REPLICATION_TARGET_ENV);
     env_vars.extend_from_slice(&[("NO_PROXY", "127.0.0.1,localhost"), ("HTTP_PROXY", ""), ("HTTPS_PROXY", "")]);
-    source_env.start_rustfs_server_with_env(vec![], &env_vars).await?;
+    let env = OdmTestEnv::start_with(OdmEnvOptions {
+        env: env_vars,
+        ..OdmEnvOptions::default()
+    })
+    .await?;
+    let source_env = &env.rustfs;
 
     let source_bucket = format!("matrix-{}-src", mode.slug());
     let source_client = source_env.create_s3_client();
@@ -284,9 +920,9 @@ async fn run_row(mode: TargetMode) -> TestResult {
         .object_lock_enabled_for_bucket(true)
         .send()
         .await?;
-    enable_bucket_versioning(&source_env, &source_bucket).await?;
+    enable_bucket_versioning(source_env, &source_bucket).await?;
     let target_arn = set_replication_target_with_options(
-        &source_env,
+        source_env,
         &source_bucket,
         ReplicationTargetOptions {
             endpoint: &target.address(),
@@ -299,14 +935,21 @@ async fn run_row(mode: TargetMode) -> TestResult {
         },
     )
     .await?;
-    put_bucket_replication(&source_env, &source_bucket, &target_arn).await?;
+    put_bucket_replication(source_env, &source_bucket, &target_arn).await?;
 
     let target_client = fake_source_client(&target);
     let mut failures = Vec::new();
     for shape in ObjectShape::ALL {
         let cell = format!("{}/{:?}", mode.slug(), shape);
-        let expected_body = shape.put(&source_client, &source_bucket).await?;
+        let expected_body = shape.put(&env, &source_bucket).await?;
         let status = wait_for_terminal_replication_status(&source_client, &source_bucket, shape.key()).await?;
+        if shape == ObjectShape::OdmPreservedMd5Multipart {
+            assert_eq!(
+                env.source.count_requests(FakeTargetOperation::GetObject, shape.key()),
+                2,
+                "one passthrough GET plus one background pull; replication must read the persisted local parts"
+            );
+        }
         let journal = target.requests();
         let outcome = match expectation(mode, shape) {
             Expectation::Completed => {
@@ -379,6 +1022,36 @@ async fn check_completed_cell(
     if uploads.is_empty() {
         return Err("no upload reached the target although the source reports COMPLETED".into());
     }
+    if shape == ObjectShape::OdmPreservedMd5Multipart {
+        let key_requests: Vec<_> = journal
+            .iter()
+            .filter(|record| record.key.as_deref() == Some(shape.key()))
+            .collect();
+        for operation in [
+            FakeTargetOperation::CreateMultipartUpload,
+            FakeTargetOperation::CompleteMultipartUpload,
+        ] {
+            if !key_requests.iter().any(|record| record.operation == operation) {
+                return Err(format!("preserved-MD5 multipart object did not use {operation:?}").into());
+            }
+        }
+        if key_requests
+            .iter()
+            .any(|record| record.operation == FakeTargetOperation::PutObject)
+        {
+            return Err("preserved-MD5 multipart object used a single PutObject".into());
+        }
+        let mut part_numbers: Vec<_> = key_requests
+            .iter()
+            .filter(|record| record.operation == FakeTargetOperation::UploadPart)
+            .map(|record| record.part_number)
+            .collect();
+        part_numbers.sort_unstable();
+        part_numbers.dedup();
+        if part_numbers != [Some(1), Some(2)] {
+            return Err(format!("preserved-MD5 multipart object uploaded unexpected parts: {part_numbers:?}").into());
+        }
+    }
     if let Some(framed) = uploads.iter().find(|record| record.transport.aws_chunked) {
         return Err(format!("{cell}: an upload went out aws-chunked (rustfs#6853 framing): {framed:?}").into());
     }
@@ -401,6 +1074,19 @@ async fn check_completed_cell(
     }) {
         return Err(format!("a locked PutObject went out without any integrity header (rustfs#7082): {bare:?}").into());
     }
+    // rustfs/backlog#2340 contract: a source checksum reaches the target as
+    // the `x-amz-checksum-*` header, not as user metadata; every PutObject of
+    // the shape carries it.
+    if let Some(header) = shape.forwarded_checksum_header()
+        && let Some(missing) = uploads.iter().find(|record| {
+            record.operation == FakeTargetOperation::PutObject
+                && !record.transport.checksum_headers.iter().any(|name| name == header)
+        })
+    {
+        return Err(
+            format!("a PutObject went out without the source's {header} header (rustfs/backlog#2340): {missing:?}").into(),
+        );
+    }
     Ok(())
 }
 
@@ -455,6 +1141,71 @@ async fn wait_for_terminal_replication_status(
     }
 }
 
+async fn odm_preserved_md5_multipart(env: &OdmTestEnv, bucket: &str, key: &str) -> Result<Bytes, Box<dyn Error + Send + Sync>> {
+    const PART_SIZE: usize = 5 * 1024 * 1024;
+    let origin_bucket = format!("{bucket}-origin");
+    env.source.create_bucket_with_mode(&origin_bucket, BucketMode::Unversioned);
+    let mut spec = env.fake_source_spec(&origin_bucket);
+    // Below the 16 MiB inline default the pull is one tee'd PUT with a single
+    // part; force the passthrough + background multipart write-back instead.
+    spec.policy.inline_max_bytes = 4096;
+    spec.policy.multipart_part_size_bytes = PART_SIZE as u64;
+    spec.policy.preserve_etag = true;
+    env.configure_and_wait(bucket, &spec).await?;
+
+    // A normal source PUT produces the MD5 ETag; only ODM chooses the local parts.
+    let body = payload(PART_SIZE + 4096, 0x66);
+    let source_put = env
+        .source_client()
+        .put_object()
+        .bucket(&origin_bucket)
+        .key(key)
+        .body(ByteStream::from(body.clone()))
+        .send()
+        .await?;
+    let source_etag = source_put.e_tag().ok_or("source PUT omitted its ETag")?.trim_matches('"');
+    assert_eq!(source_etag.len(), 32, "source fixture must have a single-PUT MD5 ETag");
+    assert!(source_etag.bytes().all(|byte| byte.is_ascii_hexdigit()));
+
+    let pulled = env.raw_get(bucket, key).await?;
+    assert_eq!(pulled.status, 200, "{}", String::from_utf8_lossy(&pulled.body));
+    assert_eq!(pulled.body, body);
+    assert!(
+        env.wait_local_listed(bucket, key, Duration::from_secs(30)).await?,
+        "ODM must persist the object"
+    );
+    let attributes = env
+        .client
+        .get_object_attributes()
+        .bucket(bucket)
+        .key(key)
+        .object_attributes(ObjectAttributes::Etag)
+        .object_attributes(ObjectAttributes::ObjectParts)
+        .object_attributes(ObjectAttributes::Checksum)
+        .send()
+        .await?;
+    assert_eq!(attributes.e_tag().map(|etag| etag.trim_matches('"')), Some(source_etag));
+    let parts = attributes
+        .object_parts()
+        .ok_or("the ODM copy must expose its two local parts")?;
+    assert_eq!(parts.total_parts_count(), Some(2));
+    assert_eq!(
+        parts
+            .parts()
+            .iter()
+            .map(|part| (part.part_number(), part.size()))
+            .collect::<Vec<_>>(),
+        [(Some(1), Some(PART_SIZE as i64)), (Some(2), Some(4096))]
+    );
+    assert!(
+        attributes
+            .checksum()
+            .is_none_or(|checksum| checksum == &Checksum::builder().build()),
+        "multipart routing must work without an object checksum record"
+    );
+    Ok(body)
+}
+
 async fn multipart_put(
     client: &Client,
     bucket: &str,

crates/e2e_test/src/upgrade_compatibility_test.rs

@@ -14,9 +14,10 @@
 
 use crate::common::{
     RustFSTestClusterEnvironment, RustFSTestEnvironment, admin_request, init_logging, replication_fast_env, rustfs_binary_path,
+    signed_request,
 };
-use crate::fake_s3_target::{BucketMode, FAKE_ACCESS_KEY, FAKE_SECRET_KEY, FakeS3Target};
+use crate::fake_s3_target::{BucketMode, FAKE_ACCESS_KEY, FAKE_SECRET_KEY, FakeS3Target, Operation as FakeTargetOperation};
-use crate::on_demand_migration::common::{ODM_SERVER_ENV, OdmTestEnv, SeedObject};
+use crate::on_demand_migration::common::{ODM_SERVER_ENV, OdmTestEnv, SeedObject, fake_source_client};
 use crate::replication_extension_test::{
     LOOPBACK_REPLICATION_TARGET_ENV, ReplicationTargetOptions, put_bucket_replication, set_replication_target_with_options,
 };
@@ -25,9 +26,10 @@ use aws_sdk_s3::error::ProvideErrorMetadata;
 use aws_sdk_s3::primitives::ByteStream;
 use aws_sdk_s3::types::{
     BucketLifecycleConfiguration, BucketVersioningStatus, CompletedMultipartUpload, CompletedPart, DefaultRetention,
-    ExpirationStatus, LifecycleExpiration, LifecycleRule, LifecycleRuleFilter, ObjectLockConfiguration, ObjectLockEnabled,
+    ExpirationStatus, LifecycleExpiration, LifecycleRule, LifecycleRuleFilter, ObjectAttributes, ObjectLockConfiguration,
-    ObjectLockRetentionMode, ObjectLockRule, PublicAccessBlockConfiguration, ServerSideEncryption, ServerSideEncryptionByDefault,
+    ObjectLockEnabled, ObjectLockRetentionMode, ObjectLockRule, PublicAccessBlockConfiguration, ServerSideEncryption,
-    ServerSideEncryptionConfiguration, ServerSideEncryptionRule, Tag, Tagging, VersioningConfiguration,
+    ServerSideEncryptionByDefault, ServerSideEncryptionConfiguration, ServerSideEncryptionRule, Tag, Tagging,
+    VersioningConfiguration,
 };
 use http::{Method, StatusCode};
 use std::path::{Path, PathBuf};
@@ -1205,3 +1207,770 @@ async fn rollback_to_previous_release_reads_current_bucket_metadata() -> TestRes
     replication_target.shutdown().await;
     Ok(())
 }
+
+// ---------------------------------------------------------------------------
+// rc.5 multipart layouts under the current build (backlog#2147 follow-up to
+// rustfs#7305)
+// ---------------------------------------------------------------------------
+//
+// rustfs#7305 changed `ObjectInfo::is_multipart` to consult the stored part
+// list before the ETag shape. Every earlier check of that change used
+// synthetic metadata; this scenario writes the layouts with the published
+// rc.5 binary and then reads, describes, and replicates them with the
+// current build on the same data directory.
+
+const LAYOUT_PLAIN_BUCKET: &str = "upgrade-layout-plain";
+const LAYOUT_ENCRYPTED_BUCKET: &str = "upgrade-layout-encrypted";
+const LAYOUT_REPLICA_BUCKET: &str = "upgrade-layout-replica";
+const LAYOUT_PART_SIZE: usize = 5 * 1024 * 1024;
+const LAYOUT_TAIL_SIZE: usize = 1024 * 1024 + 4096;
+const LAYOUT_SSEC_KEY: &str = "0123456789abcdef0123456789abcdef";
+const LAYOUT_REPLICATION_TIMEOUT: Duration = Duration::from_secs(180);
+
+struct LayoutCase {
+    bucket: &'static str,
+    key: &'static str,
+    /// Empty for a single PUT.
+    part_sizes: Vec<usize>,
+    body: Vec<u8>,
+    ssec: bool,
+    /// `false` for layouts whose replication is a known pre-existing failure;
+    /// their outcome is logged, not asserted.
+    assert_replication: bool,
+    /// Recorded from the rc.5 writer.
+    rc5_etag: String,
+    /// Whether rc.5 reported `ObjectParts` for the object.
+    rc5_reported_parts: Option<usize>,
+}
+
+impl LayoutCase {
+    fn is_multipart_layout(&self) -> bool {
+        self.part_sizes.len() > 1
+    }
+
+    fn label(&self) -> String {
+        format!("{}/{}", self.bucket, self.key)
+    }
+}
+
+fn layout_noise(len: usize, seed: u64) -> Vec<u8> {
+    let mut state = seed ^ 0x9E37_79B9_7F4A_7C15;
+    (0..len)
+        .map(|_| {
+            state ^= state << 13;
+            state ^= state >> 7;
+            state ^= state << 17;
+            (state >> 24) as u8
+        })
+        .collect()
+}
+
+fn layout_text(len: usize, seed: u64) -> Vec<u8> {
+    let mut out = Vec::with_capacity(len + 64);
+    let mut line = 0u64;
+    while out.len() < len {
+        out.extend_from_slice(format!("rc5 legacy layout seed={seed} line={line} lorem ipsum dolor sit amet\n").as_bytes());
+        line += 1;
+    }
+    out.truncate(len);
+    out
+}
+
+fn layout_ssec_key_md5() -> String {
+    use md5::{Digest as _, Md5};
+    let mut hasher = Md5::new();
+    hasher.update(LAYOUT_SSEC_KEY.as_bytes());
+    base64_simd::STANDARD.encode_to_string(hasher.finalize())
+}
+
+fn layout_ssec_key() -> String {
+    base64_simd::STANDARD.encode_to_string(LAYOUT_SSEC_KEY)
+}
+
+async fn layout_head(
+    client: &Client,
+    case: &LayoutCase,
+) -> Result<aws_sdk_s3::operation::head_object::HeadObjectOutput, BoxError> {
+    let request = client.head_object().bucket(case.bucket).key(case.key);
+    let request = if case.ssec {
+        request
+            .sse_customer_algorithm("AES256")
+            .sse_customer_key(layout_ssec_key())
+            .sse_customer_key_md5(layout_ssec_key_md5())
+    } else {
+        request
+    };
+    Ok(request.send().await?)
+}
+
+async fn layout_get(
+    client: &Client,
+    case: &LayoutCase,
+    range: Option<String>,
+    part_number: Option<i32>,
+) -> Result<aws_sdk_s3::operation::get_object::GetObjectOutput, BoxError> {
+    let request = client.get_object().bucket(case.bucket).key(case.key);
+    let request = if case.ssec {
+        request
+            .sse_customer_algorithm("AES256")
+            .sse_customer_key(layout_ssec_key())
+            .sse_customer_key_md5(layout_ssec_key_md5())
+    } else {
+        request
+    };
+    let request = request.set_range(range).set_part_number(part_number);
+    Ok(request.send().await?)
+}
+
+async fn layout_attributes(
+    client: &Client,
+    case: &LayoutCase,
+) -> Result<aws_sdk_s3::operation::get_object_attributes::GetObjectAttributesOutput, BoxError> {
+    let request = client
+        .get_object_attributes()
+        .bucket(case.bucket)
+        .key(case.key)
+        .object_attributes(ObjectAttributes::Etag)
+        .object_attributes(ObjectAttributes::ObjectParts)
+        .object_attributes(ObjectAttributes::ObjectSize)
+        .max_parts(100);
+    let request = if case.ssec {
+        request
+            .sse_customer_algorithm("AES256")
+            .sse_customer_key(layout_ssec_key())
+            .sse_customer_key_md5(layout_ssec_key_md5())
+    } else {
+        request
+    };
+    Ok(request.send().await?)
+}
+
+/// Write `case` with the rc.5 client; single PUT when `part_sizes` is empty.
+async fn layout_write(client: &Client, case: &LayoutCase) -> Result<(), BoxError> {
+    let content_type = "text/plain";
+    if case.part_sizes.is_empty() {
+        let request = client
+            .put_object()
+            .bucket(case.bucket)
+            .key(case.key)
+            .content_type(content_type)
+            .body(ByteStream::from(case.body.clone()));
+        let request = if case.ssec {
+            request
+                .sse_customer_algorithm("AES256")
+                .sse_customer_key(layout_ssec_key())
+                .sse_customer_key_md5(layout_ssec_key_md5())
+        } else {
+            request
+        };
+        request.send().await?;
+        return Ok(());
+    }
+    let create = client
+        .create_multipart_upload()
+        .bucket(case.bucket)
+        .key(case.key)
+        .content_type(content_type);
+    let create = if case.ssec {
+        create
+            .sse_customer_algorithm("AES256")
+            .sse_customer_key(layout_ssec_key())
+            .sse_customer_key_md5(layout_ssec_key_md5())
+    } else {
+        create
+    };
+    let created = create.send().await?;
+    let upload_id = created.upload_id().ok_or("CreateMultipartUpload omitted upload ID")?;
+    let mut completed = Vec::with_capacity(case.part_sizes.len());
+    let mut offset = 0usize;
+    for (index, size) in case.part_sizes.iter().enumerate() {
+        let part_number = i32::try_from(index + 1)?;
+        let chunk = case.body[offset..offset + size].to_vec();
+        offset += size;
+        let upload = client
+            .upload_part()
+            .bucket(case.bucket)
+            .key(case.key)
+            .upload_id(upload_id)
+            .part_number(part_number)
+            .body(ByteStream::from(chunk));
+        let upload = if case.ssec {
+            upload
+                .sse_customer_algorithm("AES256")
+                .sse_customer_key(layout_ssec_key())
+                .sse_customer_key_md5(layout_ssec_key_md5())
+        } else {
+            upload
+        };
+        let uploaded = upload.send().await?;
+        completed.push(
+            CompletedPart::builder()
+                .part_number(part_number)
+                .e_tag(uploaded.e_tag().ok_or("UploadPart omitted ETag")?)
+                .build(),
+        );
+    }
+    client
+        .complete_multipart_upload()
+        .bucket(case.bucket)
+        .key(case.key)
+        .upload_id(upload_id)
+        .multipart_upload(CompletedMultipartUpload::builder().set_parts(Some(completed)).build())
+        .send()
+        .await?;
+    Ok(())
+}
+
+fn layout_cases() -> Vec<LayoutCase> {
+    let two = vec![LAYOUT_PART_SIZE, LAYOUT_TAIL_SIZE];
+    let three = vec![LAYOUT_PART_SIZE, LAYOUT_PART_SIZE, 4096];
+    let total = |sizes: &[usize]| sizes.iter().sum::<usize>();
+    let case = |bucket, key, part_sizes: Vec<usize>, body: Vec<u8>, ssec| LayoutCase {
+        bucket,
+        key,
+        part_sizes,
+        body,
+        ssec,
+        assert_replication: true,
+        rc5_etag: String::new(),
+        rc5_reported_parts: None,
+    };
+    vec![
+        case(LAYOUT_PLAIN_BUCKET, "plain/single.bin", vec![], layout_noise(1024 * 1024 + 17, 1), false),
+        case(
+            LAYOUT_PLAIN_BUCKET,
+            "plain/multipart-2.bin",
+            two.clone(),
+            layout_noise(total(&two), 2),
+            false,
+        ),
+        case(
+            LAYOUT_PLAIN_BUCKET,
+            "plain/multipart-3.bin",
+            three.clone(),
+            layout_noise(total(&three), 3),
+            false,
+        ),
+        case(
+            LAYOUT_PLAIN_BUCKET,
+            "plain/compressed-single.txt",
+            vec![],
+            layout_text(1024 * 1024 + 17, 4),
+            false,
+        ),
+        case(
+            LAYOUT_PLAIN_BUCKET,
+            "plain/compressed-multipart-2.txt",
+            two.clone(),
+            layout_text(total(&two), 5),
+            false,
+        ),
+        case(
+            LAYOUT_PLAIN_BUCKET,
+            "plain/ssec-multipart-2.bin",
+            two.clone(),
+            layout_noise(total(&two), 6),
+            true,
+        ),
+        // SSE-C passthrough replicates the stored ciphertext part by part; a
+        // compressible first part is stored well below 5 MiB, so the sender
+        // declares each part's plaintext length and the target validates the
+        // 5 MiB minimum against it (rustfs/backlog#2363). rc.5 as the sender
+        // still fails this layout (see `rc5_baseline_replicates_multipart_layouts`).
+        case(
+            LAYOUT_PLAIN_BUCKET,
+            "plain/ssec-compressed-multipart-2.txt",
+            two.clone(),
+            layout_text(total(&two), 7),
+            true,
+        ),
+        case(
+            LAYOUT_ENCRYPTED_BUCKET,
+            "encrypted/single.bin",
+            vec![],
+            layout_noise(1024 * 1024 + 17, 8),
+            false,
+        ),
+        case(
+            LAYOUT_ENCRYPTED_BUCKET,
+            "encrypted/multipart-2.bin",
+            two.clone(),
+            layout_noise(total(&two), 9),
+            false,
+        ),
+        case(
+            LAYOUT_ENCRYPTED_BUCKET,
+            "encrypted/multipart-3.bin",
+            three.clone(),
+            layout_noise(total(&three), 10),
+            false,
+        ),
+        case(
+            LAYOUT_ENCRYPTED_BUCKET,
+            "encrypted/compressed-multipart-2.txt",
+            two.clone(),
+            layout_text(total(&two), 11),
+            false,
+        ),
+    ]
+}
+
+fn layout_server_env() -> Vec<(&'static str, &'static str)> {
+    let mut env = bucket_config_server_env();
+    env.push(("RUSTFS_COMPRESSION_ENABLED", "true"));
+    env.push(("RUSTFS_COMPRESSION_MULTIPART_ENABLED", "true"));
+    env
+}
+
+fn layout_reported_parts(attributes: &aws_sdk_s3::operation::get_object_attributes::GetObjectAttributesOutput) -> Option<usize> {
+    attributes.object_parts().map(|parts| parts.parts().len())
+}
+
+async fn assert_layout_readable(client: &Client, case: &LayoutCase, context: &str) -> TestResult {
+    let label = case.label();
+    let head = layout_head(client, case).await?;
+    assert_eq!(
+        head.e_tag().map(|etag| etag.trim_matches('"')),
+        Some(case.rc5_etag.as_str()),
+        "{context}: {label}: the ETag written by rc.5 must be reported unchanged"
+    );
+    assert_eq!(
+        head.content_length(),
+        Some(i64::try_from(case.body.len())?),
+        "{context}: {label}: HEAD content length"
+    );
+
+    let full = layout_get(client, case, None, None).await?.body.collect().await?.into_bytes();
+    assert_eq!(full.len(), case.body.len(), "{context}: {label}: full GET length");
+    assert!(full == case.body, "{context}: {label}: full GET body must equal the rc.5 upload");
+
+    if case.is_multipart_layout() {
+        let first = case.part_sizes[0];
+        let range = format!("bytes={}-{}", first - 32, first + 31);
+        let crossing = layout_get(client, case, Some(range), None)
+            .await?
+            .body
+            .collect()
+            .await?
+            .into_bytes();
+        assert!(
+            crossing == case.body[first - 32..first + 32],
+            "{context}: {label}: range across the first part boundary"
+        );
+        let tail_start: usize = case.part_sizes[..case.part_sizes.len() - 1].iter().sum();
+        let last_number = i32::try_from(case.part_sizes.len())?;
+        let last = layout_get(client, case, None, Some(last_number)).await?;
+        assert_eq!(
+            last.content_length(),
+            Some(i64::try_from(case.part_sizes[case.part_sizes.len() - 1])?),
+            "{context}: {label}: partNumber={last_number} length"
+        );
+        let last_body = last.body.collect().await?.into_bytes();
+        assert!(
+            last_body == case.body[tail_start..],
+            "{context}: {label}: partNumber={last_number} body must be the stored last part"
+        );
+    }
+    Ok(())
+}
+
+async fn assert_layout_attributes(client: &Client, case: &LayoutCase, context: &str) -> TestResult {
+    let label = case.label();
+    let attributes = layout_attributes(client, case).await?;
+    assert_eq!(
+        attributes.e_tag().map(|etag| etag.trim_matches('"')),
+        Some(case.rc5_etag.as_str()),
+        "{context}: {label}: attributes ETag"
+    );
+    assert_eq!(
+        attributes.object_size(),
+        Some(i64::try_from(case.body.len())?),
+        "{context}: {label}: attributes ObjectSize"
+    );
+    if case.is_multipart_layout() {
+        let parts = attributes
+            .object_parts()
+            .ok_or_else(|| format!("{context}: {label}: multipart layout must expose ObjectParts"))?;
+        assert_eq!(
+            parts.total_parts_count(),
+            Some(i32::try_from(case.part_sizes.len())?),
+            "{context}: {label}: TotalPartsCount"
+        );
+        let observed: Vec<(Option<i32>, Option<i64>)> =
+            parts.parts().iter().map(|part| (part.part_number(), part.size())).collect();
+        let expected: Vec<(Option<i32>, Option<i64>)> = case
+            .part_sizes
+            .iter()
+            .enumerate()
+            .map(|(index, size)| (Some(index as i32 + 1), Some(*size as i64)))
+            .collect();
+        assert_eq!(
+            observed, expected,
+            "{context}: {label}: ObjectParts must report the plaintext part layout"
+        );
+    } else {
+        assert!(
+            attributes.object_parts().is_none_or(|parts| parts.parts().is_empty()),
+            "{context}: {label}: a single PUT must not report stored parts"
+        );
+    }
+    Ok(())
+}
+
+async fn put_layout_replication_rule(env: &RustFSTestEnvironment, bucket: &str, arn: &str) -> TestResult {
+    let body = format!(
+        r#"<ReplicationConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/">
+  <Role></Role>
+  <Rule>
+    <ID>legacy-layouts</ID>
+    <Priority>1</Priority>
+    <Status>Enabled</Status>
+    <Filter><Prefix></Prefix></Filter>
+    <DeleteMarkerReplication><Status>Enabled</Status></DeleteMarkerReplication>
+    <DeleteReplication><Status>Enabled</Status></DeleteReplication>
+    <ExistingObjectReplication><Status>Enabled</Status></ExistingObjectReplication>
+    <Destination><Bucket>{arn}</Bucket></Destination>
+  </Rule>
+</ReplicationConfiguration>"#
+    );
+    let url = format!("{}/{bucket}?replication", env.url);
+    let response = signed_request(
+        Method::PUT,
+        &url,
+        &env.access_key,
+        &env.secret_key,
+        Some(body.into_bytes()),
+        Some("application/xml"),
+    )
+    .await?;
+    if response.status() != StatusCode::OK {
+        let status = response.status();
+        let body = response.text().await.unwrap_or_default();
+        return Err(format!("put replication rule on {bucket} failed: {status} {body}").into());
+    }
+    Ok(())
+}
+
+/// Wait for the existing-object replication of `case` to reach a terminal
+/// status and return it (`COMPLETED` or `FAILED`).
+async fn wait_layout_replication_terminal(client: &Client, case: &LayoutCase) -> Result<String, BoxError> {
+    let deadline = Instant::now() + LAYOUT_REPLICATION_TIMEOUT;
+    loop {
+        let head = layout_head(client, case).await?;
+        let status = head.replication_status().map(|status| status.as_str().to_string());
+        if matches!(status.as_deref(), Some("COMPLETED") | Some("FAILED")) {
+            return Ok(status.unwrap_or_default());
+        }
+        if Instant::now() >= deadline {
+            return Err(format!(
+                "{}: existing-object replication never reached a terminal status; last {status:?}",
+                case.label()
+            )
+            .into());
+        }
+        sleep(Duration::from_millis(500)).await;
+    }
+}
+
+#[derive(Debug)]
+struct LayoutTransport {
+    status: String,
+    uploaded_parts: Vec<i32>,
+    single_puts: usize,
+    completes: usize,
+    /// Raw per-key journal in target order: (sequence, operation, part number,
+    /// upload id), so duplicate drives can be told apart from retries.
+    journal: Vec<(u64, String, Option<i32>, Option<String>)>,
+}
+
+/// Configure every layout bucket to replicate its existing objects to a fresh
+/// fake target, wait for each case to settle, and report the transport the
+/// target observed per case.
+async fn replicate_layouts(
+    env: &RustFSTestEnvironment,
+    client: &Client,
+    cases: &[LayoutCase],
+) -> Result<(FakeS3Target, Vec<LayoutTransport>), BoxError> {
+    let target = FakeS3Target::start().await?;
+    target.create_bucket(LAYOUT_REPLICA_BUCKET);
+    for bucket in [LAYOUT_PLAIN_BUCKET, LAYOUT_ENCRYPTED_BUCKET] {
+        let arn = set_replication_target_with_options(
+            env,
+            bucket,
+            ReplicationTargetOptions {
+                endpoint: &target.address(),
+                access_key: FAKE_ACCESS_KEY,
+                secret_key: FAKE_SECRET_KEY,
+                target_bucket: LAYOUT_REPLICA_BUCKET,
+                secure: false,
+                skip_tls_verify: false,
+                ca_cert_pem: None,
+            },
+        )
+        .await?;
+        put_layout_replication_rule(env, bucket, &arn).await?;
+    }
+    let mut statuses = Vec::with_capacity(cases.len());
+    for case in cases {
+        statuses.push(wait_layout_replication_terminal(client, case).await?);
+    }
+    let journal = target.requests();
+    let mut transports = Vec::with_capacity(cases.len());
+    for (case, status) in cases.iter().zip(statuses) {
+        let key_requests: Vec<_> = journal
+            .iter()
+            .filter(|record| record.key.as_deref() == Some(case.key))
+            .collect();
+        let mut uploaded_parts: Vec<i32> = key_requests
+            .iter()
+            .filter(|record| record.operation == FakeTargetOperation::UploadPart)
+            .filter_map(|record| record.part_number)
+            .collect();
+        uploaded_parts.sort_unstable();
+        uploaded_parts.dedup();
+        let transport = LayoutTransport {
+            status,
+            uploaded_parts,
+            single_puts: key_requests
+                .iter()
+                .filter(|record| record.operation == FakeTargetOperation::PutObject)
+                .count(),
+            completes: key_requests
+                .iter()
+                .filter(|record| record.operation == FakeTargetOperation::CompleteMultipartUpload)
+                .count(),
+            journal: key_requests
+                .iter()
+                .map(|record| {
+                    (
+                        record.sequence,
+                        format!("{:?}", record.operation),
+                        record.part_number,
+                        record.upload_id.as_ref().map(|id| id.chars().take(12).collect()),
+                    )
+                })
+                .collect(),
+        };
+        tracing::info!(
+            target: "e2e_test::upgrade_compatibility_test",
+            object = %case.label(),
+            ?transport,
+            "replication transport observed on the target"
+        );
+        transports.push(transport);
+    }
+    Ok((target, transports))
+}
+
+/// rc.5 writes single-PUT, multipart, compressed, SSE-C and SSE-S3 layouts;
+/// the current build must read every byte, expose the stored part layout
+/// through GetObjectAttributes and partNumber reads, and replicate the objects
+/// with the transport that matches their stored parts.
+#[tokio::test]
+#[ignore = "requires the pinned 1.0.0-rc.5 release binary"]
+async fn direct_upgrade_from_rc5_preserves_multipart_layouts() -> TestResult {
+    init_logging();
+    let previous_binary = source_binary()?;
+    let server_env = layout_server_env();
+
+    let mut env = RustFSTestEnvironment::new().await?;
+    env.start_rustfs_server_from_binary(&previous_binary, vec![], &server_env)
+        .await?;
+    let old_client = env.create_s3_client();
+    env.create_test_bucket(LAYOUT_PLAIN_BUCKET).await?;
+    env.create_test_bucket(LAYOUT_ENCRYPTED_BUCKET).await?;
+    enable_versioning(&old_client, LAYOUT_PLAIN_BUCKET).await?;
+    enable_versioning(&old_client, LAYOUT_ENCRYPTED_BUCKET).await?;
+    put_default_sse_s3_encryption(&old_client, LAYOUT_ENCRYPTED_BUCKET).await?;
+    assert_default_sse_s3_encryption(&old_client, LAYOUT_ENCRYPTED_BUCKET, "rc.5").await?;
+
+    let mut cases = layout_cases();
+    for case in cases.iter_mut() {
+        layout_write(&old_client, case).await?;
+        let head = layout_head(&old_client, case).await?;
+        case.rc5_etag = head
+            .e_tag()
+            .ok_or_else(|| format!("{}: rc.5 HEAD omitted the ETag", case.label()))?
+            .trim_matches('"')
+            .to_string();
+        case.rc5_reported_parts = layout_attributes(&old_client, case)
+            .await
+            .ok()
+            .and_then(|a| layout_reported_parts(&a));
+        tracing::info!(
+            target: "e2e_test::upgrade_compatibility_test",
+            object = %case.label(),
+            parts = case.part_sizes.len(),
+            etag = %case.rc5_etag,
+            rc5_reported_parts = ?case.rc5_reported_parts,
+            "rc.5 wrote a legacy layout"
+        );
+    }
+    // The rc.5 writer must itself still read what it wrote, so a later
+    // failure is attributable to the upgrade rather than to the fixture.
+    for case in &cases {
+        assert_layout_readable(&old_client, case, "rc.5").await?;
+    }
+
+    // Upgrade in place.
+    env.restart_server_preserving_data(vec![], &server_env).await?;
+    let client = env.create_s3_client();
+    for case in &cases {
+        assert_layout_readable(&client, case, "upgraded").await?;
+        assert_layout_attributes(&client, case, "upgraded").await?;
+    }
+
+    // Replicate the pre-existing objects with the current build.
+    let (target, transports) = replicate_layouts(&env, &client, &cases).await?;
+    let replica_client = fake_source_client(&target);
+    for (case, transport) in cases.iter().zip(&transports) {
+        let label = case.label();
+        if !case.assert_replication {
+            continue;
+        }
+        assert_eq!(transport.status, "COMPLETED", "{label}: existing-object replication must complete");
+        if case.is_multipart_layout() {
+            let expected: Vec<i32> = (1..=i32::try_from(case.part_sizes.len())?).collect();
+            assert_eq!(
+                transport.uploaded_parts, expected,
+                "{label}: stored parts must replicate as the same multipart layout"
+            );
+            // An object still PENDING when the next scanner cycle arrives is
+            // not driven a second time (rustfs/backlog#2362); the journal is
+            // logged so a duplicate round is visible if this ever regresses.
+            assert_eq!(
+                transport.completes, 1,
+                "{label}: exactly one CompleteMultipartUpload; journal {:?}",
+                transport.journal
+            );
+            assert_eq!(
+                transport.single_puts, 0,
+                "{label}: a multipart layout must not go out as a single PutObject"
+            );
+        } else {
+            assert_eq!(transport.single_puts, 1, "{label}: a single PUT replicates as exactly one PutObject");
+            assert!(transport.uploaded_parts.is_empty(), "{label}: a single PUT must not go out as multipart");
+        }
+        if !case.ssec {
+            let replica = replica_client
+                .get_object()
+                .bucket(LAYOUT_REPLICA_BUCKET)
+                .key(case.key)
+                .send()
+                .await
+                .map_err(|err| format!("{label}: replica missing on the target: {err}"))?
+                .body
+                .collect()
+                .await?
+                .into_bytes();
+            assert_eq!(replica.len(), case.body.len(), "{label}: replica length");
+            assert!(replica == case.body, "{label}: replica body must equal the rc.5 upload");
+        }
+    }
+    Ok(())
+}
+
+/// The same layouts replicated by rc.5 itself, without an upgrade. This is the
+/// baseline that tells a pre-existing transport failure apart from one the
+/// current build introduced; it records the outcome per layout and only fails
+/// when the fixture cannot run.
+#[tokio::test]
+#[ignore = "requires the pinned 1.0.0-rc.5 release binary"]
+async fn rc5_baseline_replicates_multipart_layouts() -> TestResult {
+    init_logging();
+    let previous_binary = source_binary()?;
+    let server_env = layout_server_env();
+
+    let mut env = RustFSTestEnvironment::new().await?;
+    env.start_rustfs_server_from_binary(&previous_binary, vec![], &server_env)
+        .await?;
+    let client = env.create_s3_client();
+    env.create_test_bucket(LAYOUT_PLAIN_BUCKET).await?;
+    env.create_test_bucket(LAYOUT_ENCRYPTED_BUCKET).await?;
+    enable_versioning(&client, LAYOUT_PLAIN_BUCKET).await?;
+    enable_versioning(&client, LAYOUT_ENCRYPTED_BUCKET).await?;
+    put_default_sse_s3_encryption(&client, LAYOUT_ENCRYPTED_BUCKET).await?;
+
+    let mut cases = layout_cases();
+    for case in cases.iter_mut() {
+        layout_write(&client, case).await?;
+        let head = layout_head(&client, case).await?;
+        case.rc5_etag = head
+            .e_tag()
+            .ok_or_else(|| format!("{}: rc.5 HEAD omitted the ETag", case.label()))?
+            .trim_matches('"')
+            .to_string();
+    }
+    let (_target, transports) = replicate_layouts(&env, &client, &cases).await?;
+    let summary: Vec<String> = cases
+        .iter()
+        .zip(&transports)
+        .map(|(case, transport)| {
+            format!(
+                "{}: {} parts={:?} puts={}",
+                case.label(),
+                transport.status,
+                transport.uploaded_parts,
+                transport.single_puts
+            )
+        })
+        .collect();
+    tracing::info!(target: "e2e_test::upgrade_compatibility_test", ?summary, "rc.5 baseline replication outcomes");
+    Ok(())
+}
+
+/// backlog#2362 under the same conditions that reproduced it with the rc.5
+/// writer, but with the workspace build on both sides so it runs in the
+/// ordinary lane: every pre-existing layout is driven through exactly one
+/// upload round even though the scanner re-scans it every second while the
+/// first round is still in flight.
+#[tokio::test]
+async fn existing_object_replication_drives_each_layout_once() -> TestResult {
+    init_logging();
+    let server_env = layout_server_env();
+
+    let mut env = RustFSTestEnvironment::new().await?;
+    env.start_rustfs_server_with_env(vec![], &server_env).await?;
+    let writer = env.create_s3_client();
+    env.create_test_bucket(LAYOUT_PLAIN_BUCKET).await?;
+    env.create_test_bucket(LAYOUT_ENCRYPTED_BUCKET).await?;
+    enable_versioning(&writer, LAYOUT_PLAIN_BUCKET).await?;
+    enable_versioning(&writer, LAYOUT_ENCRYPTED_BUCKET).await?;
+    put_default_sse_s3_encryption(&writer, LAYOUT_ENCRYPTED_BUCKET).await?;
+
+    let mut cases = layout_cases();
+    for case in cases.iter_mut() {
+        layout_write(&writer, case).await?;
+        let head = layout_head(&writer, case).await?;
+        case.rc5_etag = head
+            .e_tag()
+            .ok_or_else(|| format!("{}: HEAD omitted the ETag", case.label()))?
+            .trim_matches('"')
+            .to_string();
+    }
+
+    // The objects come from an earlier process lifetime: the scanner starts
+    // cold and every object is a candidate at once.
+    env.restart_server_preserving_data(vec![], &server_env).await?;
+    let client = env.create_s3_client();
+    let (_target, transports) = replicate_layouts(&env, &client, &cases).await?;
+    let mut duplicates = Vec::new();
+    for (case, transport) in cases.iter().zip(&transports) {
+        assert_eq!(
+            transport.status,
+            "COMPLETED",
+            "{}: existing-object replication must complete",
+            case.label()
+        );
+        let rounds = if case.is_multipart_layout() {
+            transport.completes
+        } else {
+            transport.single_puts
+        };
+        if rounds != 1 {
+            duplicates.push(format!("{}: {rounds} upload rounds; journal {:?}", case.label(), transport.journal));
+        }
+    }
+    assert!(duplicates.is_empty(), "each existing object must be driven exactly once: {duplicates:?}");
+    Ok(())
+}

crates/ecstore/Cargo.toml

@@ -228,7 +228,7 @@ metrics = { workspace = true }
 # crates.io. The guard scripts/check_no_tokio_io_uring.sh allows an explicit
 # io-uring integration; only the tokio "io-uring" runtime feature is banned.
 [target.'cfg(target_os = "linux")'.dependencies]
-rustfs-uring = "0.2.1"
+rustfs-uring = "0.2.2"
 
 [target.'cfg(windows)'.dependencies]
 winapi-util.workspace = true

crates/ecstore/src/api/mod.rs

@@ -32,7 +32,8 @@ pub mod bucket {
     pub mod bucket_target_sys {
         pub use crate::bucket::bucket_target_sys::{
             AdvancedPutOptions, BucketTargetError, BucketTargetSys, PutObjectOptions, RemoveObjectOptions, S3ClientError,
-            SsecPassthroughCapability, TargetClient, append_version_id_query,
+            SsecPassthroughCapability, TargetClient, VersionIdentityCapability, append_version_id_query,
+            resolve_delete_api_version_id,
         };
     }
 
@@ -76,11 +77,26 @@ pub mod bucket {
             };
         }
 
+        pub mod recovery_disposition {
+            pub use crate::bucket::lifecycle::recovery_disposition::{
+                IlmRecoveryDispositionExecutionOutcome, IlmRecoveryDispositionReasonCode, IlmRecoveryDispositionState,
+                dry_run_recovery_disposition, execute_recovery_disposition,
+            };
+        }
+
+        pub mod recovery_export {
+            pub use crate::bucket::lifecycle::recovery_export::{
+                IlmRecoveryExportCreated, IlmRecoveryExportObservation, create_recovery_export,
+                inspect_recovery_export_observation, load_recovery_export,
+            };
+        }
+
         pub mod transition_transaction {
             pub use crate::bucket::lifecycle::transition_transaction::{
                 TransitionOperatorDeleteResult, TransitionOperatorError, TransitionOperatorProbe, TransitionOperatorStatus,
-                delete_transition_candidate_for_operator, finalize_missing_transition_transaction_for_operator,
+                TransitionRecoveryRetryResult, TransitionRecoveryRetryStatus, delete_transition_candidate_for_operator,
-                inspect_transition_transaction_for_operator,
+                finalize_missing_transition_transaction_for_operator, inspect_transition_recovery_retry_for_operator,
+                inspect_transition_transaction_for_operator, retry_transition_recovery_for_operator,
             };
             #[cfg(feature = "test-util")]
             pub use crate::bucket::lifecycle::transition_transaction::{
@@ -293,7 +309,7 @@ pub mod cache {
 pub mod capacity {
     pub use crate::core::pools::{
         DecommissionUnresolvedEntry, PoolDecommissionInfo, PoolStatus, get_total_usable_capacity, get_total_usable_capacity_free,
-        path2_bucket_object, path2_bucket_object_with_base_path,
+        is_pool_activation_fleet_proof_error, path2_bucket_object, path2_bucket_object_with_base_path,
     };
     pub use crate::store::utils::is_reserved_or_invalid_bucket;
 }
@@ -402,8 +418,8 @@ pub mod disk {
 
 pub mod error {
     pub use crate::error::{
-        Error, Result, StorageError, classify_system_path_failure_reason, is_err_bucket_not_found, is_err_object_not_found,
+        Error, PoolMetadataError, PoolMetadataFailure, Result, StorageError, classify_system_path_failure_reason,
-        is_err_version_not_found,
+        is_err_bucket_not_found, is_err_object_not_found, is_err_version_not_found,
     };
 }
 
@@ -448,9 +464,12 @@ pub mod notification {
     #[cfg(any(test, feature = "test-util"))]
     pub use crate::services::notification_sys::rotate_cross_pool_fence_fleet_proof_for_test;
     pub use crate::services::notification_sys::{
-        ClusterTierDailyStats, CrossPoolFenceFleetProofToken, LegacyTransitionStateReconcileFleetProofToken, NotificationPeerErr,
+        ClusterTierDailyStats, CrossPoolFenceFleetProofToken, IlmRecoveryExportFleetProofToken,
-        NotificationSys, ScannerPublicationLeaseGrant, acquire_cross_pool_fence_fleet_proof,
+        LegacyTransitionStateReconcileFleetProofToken, NotificationPeerErr, NotificationSys, ScannerPublicationLeaseGrant,
+        acquire_cross_pool_fence_fleet_proof, acquire_ilm_recovery_export_fleet_proof,
         acquire_legacy_transition_state_reconcile_fleet_proof, cross_pool_fence_fleet_proof_matches, get_global_notification_sys,
+        ilm_recovery_export_fleet_proof_matches, ilm_recovery_export_local_process_epoch,
+        ilm_recovery_export_member_epochs_sha256, ilm_recovery_export_topology_generation,
         legacy_transition_state_reconcile_fleet_proof_matches, new_global_notification_sys,
         scanner_peer_transport_error_message_is_retryable, start_remote_version_state_fleet_probe,
     };
@@ -503,7 +522,8 @@ pub mod rpc {
     pub use crate::cluster::rpc::{
         AuthenticatedChannel, KMS_SIGNAL_SUBSYSTEM, LocalPeerS3Client, PEER_RESTDRY_RUN, PEER_RESTSIGNAL, PEER_RESTSUB_SYS,
         PeerRestClient, PeerS3Client, S3PeerSys, SERVICE_SIGNAL_REFRESH_CONFIG, SERVICE_SIGNAL_RELOAD_DYNAMIC,
-        ScannerBucketListing, ScannerPeerActivity, ScannerPeerDirtyUsageSnapshot, ScannerPublicationLease, TONIC_RPC_PREFIX,
+        ScannerBucketListing, ScannerDirtyUsageAcknowledgement, ScannerPeerActivity, ScannerPeerDirtyUsageBucket,
+        ScannerPeerDirtyUsageSnapshot, ScannerPublicationLease, ScannerScopedDirtyUsageAckEntry, TONIC_RPC_PREFIX,
         TonicInterceptor, build_put_file_auth_trailer, check_and_record_signed_rpc_nonce, decode_heal_bucket_rpc_options,
         encode_heal_bucket_rpc_options, gen_signature_headers, gen_tonic_replay_scope_headers, gen_tonic_signature_headers,
         gen_tonic_signature_interceptor, node_service_time_out_client, node_service_time_out_client_no_auth,

crates/ecstore/src/bucket/bucket_target_sys.rs

@@ -18,7 +18,7 @@ use crate::bucket::metadata_sys::get_replication_config;
 use crate::bucket::remote_s3_client::{
     PathStyle, REPLICATION_TARGET_RETRY_POLICY, RemoteCredentials, RemoteS3EndpointSpec, build_remote_s3_client,
 };
-use crate::bucket::replication::{ObjectLockIntegrity, object_lock_put_integrity};
+use crate::bucket::replication::{ObjectLockIntegrity, object_lock_put_integrity, replication_etags_match};
 use crate::bucket::replication::{ReplicationStatusType, ReplicationTargetConfigBridge};
 use crate::bucket::target::ARN;
 use crate::bucket::target::BucketTargetType;
@@ -33,6 +33,8 @@ use aws_sdk_s3::operation::get_object::{GetObjectError, GetObjectOutput};
 use aws_sdk_s3::operation::get_object_tagging::{GetObjectTaggingError, GetObjectTaggingOutput};
 use aws_sdk_s3::operation::head_bucket::HeadBucketError;
 use aws_sdk_s3::operation::head_object::HeadObjectError;
+use aws_sdk_s3::operation::put_object_legal_hold::{PutObjectLegalHoldError, PutObjectLegalHoldOutput};
+use aws_sdk_s3::operation::put_object_retention::{PutObjectRetentionError, PutObjectRetentionOutput};
 use aws_sdk_s3::operation::put_object_tagging::{PutObjectTaggingError, PutObjectTaggingOutput};
 use aws_sdk_s3::operation::upload_part::UploadPartOutput;
 use aws_sdk_s3::primitives::ByteStream;
@@ -42,6 +44,7 @@ use aws_sdk_s3::types::{
     ChecksumAlgorithm, ChecksumMode, CompletedMultipartUpload, CompletedPart, ObjectLockLegalHoldStatus, ObjectLockRetentionMode,
     ServerSideEncryption,
 };
+use aws_sdk_s3::types::{ObjectLockLegalHold, ObjectLockRetention};
 use aws_sdk_s3::{Client as S3Client, operation::head_object::HeadObjectOutput};
 use aws_smithy_runtime_api::client::orchestrator::HttpRequest;
 use futures::{StreamExt, stream};
@@ -126,6 +129,25 @@ impl From<&BucketTarget> for RemoteS3EndpointSpec {
 }
 
 pub type HeadObjectSdkError = Box<SdkError<HeadObjectError>>;
+
+/// Whether an edited bucket target still addresses the same remote service
+/// (endpoint, bucket, path style, TLS and identity), so a verdict learned
+/// about that service stays valid across the edit.
+fn same_replication_service(edited: &BucketTarget, previous: &BucketTarget) -> bool {
+    let access_key = |target: &BucketTarget| target.credentials.as_ref().map(|credentials| credentials.access_key.clone());
+    edited.endpoint == previous.endpoint
+        && edited.target_bucket == previous.target_bucket
+        && edited.secure == previous.secure
+        && edited.path == previous.path
+        && access_key(edited) == access_key(previous)
+}
+
+/// Page size and page budget for [`TargetClient::locate_replica_by_etag`].
+const FIND_VERSION_BY_ETAG_PAGE_SIZE: i32 = 1000;
+const FIND_VERSION_BY_ETAG_MAX_PAGES: usize = 8;
+/// Candidate cap for [`TargetClient::replica_candidates_by_etag`]: more than
+/// this many same-content versions of one key is ambiguity by any measure.
+const FIND_VERSION_BY_ETAG_MAX_MATCHES: usize = 16;
 pub type GetObjectSdkError = Box<SdkError<GetObjectError>>;
 pub type GetObjectTaggingSdkError = Box<SdkError<GetObjectTaggingError>>;
 pub type PutObjectTaggingSdkError = Box<SdkError<PutObjectTaggingError>>;
@@ -349,6 +371,13 @@ struct TargetClientBuildProbe {
 /// their import path while the verdict vocabulary lives with the
 /// replication decision logic.
 pub use crate::bucket::replication::SsecPassthroughCapability;
+/// Version-identity verdicts (see the enum's own docs in
+/// `rustfs-replication`) are cached here per target ARN and follow the same
+/// `arn_remotes_map` lifecycle. They carry no TTL: the verdict is refreshed
+/// by every replication write's response, so it can only go stale on a
+/// target that receives no writes — and a stale `MintsOwn` costs one extra
+/// content-identity lookup before a PUT, never a lost replica.
+pub use crate::bucket::replication::VersionIdentityCapability;
 
 /// How long an audited SSE-C passthrough verdict stays authoritative.
 ///
@@ -375,6 +404,11 @@ pub struct BucketTargetSys {
     /// SSE-C passthrough capability verdicts keyed by target ARN. See
     /// [`SsecPassthroughCapability`]; reset alongside `arn_remotes_map`.
     ssec_passthrough_map: Arc<RwLock<HashMap<String, SsecPassthroughRecord>>>,
+    /// Version-identity verdicts keyed by target ARN. See
+    /// [`VersionIdentityCapability`]; reset alongside `arn_remotes_map`. A std
+    /// lock (never held across an await) so the replication worker can record
+    /// a verdict from inside its synchronous PUT-response audit.
+    version_identity_map: Arc<std::sync::RwLock<HashMap<String, VersionIdentityCapability>>>,
     pub targets_map: Arc<RwLock<HashMap<String, Vec<BucketTarget>>>>,
     /// Buckets whose persisted `bucket-targets.json` exists but cannot be
     /// decoded (rustfs/backlog#2282). Written under the bucket's update mutex
@@ -423,6 +457,7 @@ impl BucketTargetSys {
         Self {
             arn_remotes_map: Arc::new(RwLock::new(HashMap::new())),
             ssec_passthrough_map: Arc::new(RwLock::new(HashMap::new())),
+            version_identity_map: Arc::new(std::sync::RwLock::new(HashMap::new())),
             targets_map: Arc::new(RwLock::new(HashMap::new())),
             unreadable_targets: Arc::new(RwLock::new(HashSet::new())),
             h_mutex: Arc::new(RwLock::new(HashMap::new())),
@@ -746,10 +781,40 @@ impl BucketTargetSys {
                 arn_remotes_map.remove(&target.arn);
                 health_map.remove(&target.arn);
                 ssec_map.remove(&target.arn);
+                self.forget_version_identity_capability(&target.arn);
             }
         }
     }
 
+    /// Cached version-identity verdict for a target ARN; `Unknown` until a
+    /// replication write or a replication-check VersionFidelity probe judged
+    /// it since the target was built.
+    pub fn version_identity_capability(&self, arn: &str) -> VersionIdentityCapability {
+        self.version_identity_map
+            .read()
+            .unwrap_or_else(|poisoned| poisoned.into_inner())
+            .get(arn)
+            .copied()
+            .unwrap_or_default()
+    }
+
+    /// Record a version-identity verdict for a target ARN. Written by the
+    /// replication worker after every PutObject / CompleteMultipartUpload
+    /// response and by the replication-check VersionFidelity phase.
+    pub fn record_version_identity_capability(&self, arn: &str, capability: VersionIdentityCapability) {
+        self.version_identity_map
+            .write()
+            .unwrap_or_else(|poisoned| poisoned.into_inner())
+            .insert(arn.to_string(), capability);
+    }
+
+    fn forget_version_identity_capability(&self, arn: &str) {
+        self.version_identity_map
+            .write()
+            .unwrap_or_else(|poisoned| poisoned.into_inner())
+            .remove(arn);
+    }
+
     /// Cached SSE-C passthrough capability for a target ARN, plus whether the
     /// verdict is older than [`SSEC_PASSTHROUGH_CAPABILITY_TTL`]. `(Unknown,
     /// false)` when no verdict has been recorded since the target was built.
@@ -1162,12 +1227,32 @@ impl BucketTargetSys {
         // Remove existing targets
         if let Some(existing_targets) = targets_map.remove(bucket) {
             let mut ssec_map = self.ssec_passthrough_map.write().await;
+            let unchanged_service: HashMap<&str, &BucketTarget> = targets
+                .map(|new_targets| {
+                    new_targets
+                        .targets
+                        .iter()
+                        .map(|target| (target.arn.as_str(), target))
+                        .collect()
+                })
+                .unwrap_or_default();
             for target in existing_targets {
                 arn_remotes_map.remove(&target.arn);
                 health_map.remove(&target.arn);
                 // A rebuilt/edited target may point at a different service:
                 // the SSE-C passthrough verdict must be re-audited from Unknown.
                 ssec_map.remove(&target.arn);
+                // The version-identity verdict survives an edit that keeps the
+                // same remote service (a resync start or a bandwidth change
+                // rewrites the entry in place): forgetting it there would make
+                // the very resync that follows re-drive every object as a
+                // duplicate on a target that mints its own version ids.
+                if unchanged_service
+                    .get(target.arn.as_str())
+                    .is_none_or(|edited| !same_replication_service(edited, &target))
+                {
+                    self.forget_version_identity_capability(&target.arn);
+                }
                 self.update_bandwidth_limit(bucket, &target.arn, 0);
             }
         }
@@ -1284,6 +1369,7 @@ fn generate_arn(t: &BucketTarget, depl_id: &str) -> String {
     arn.to_string()
 }
 
+#[derive(Debug, Clone)]
 pub struct RemoveObjectOptions {
     pub force_delete: bool,
     pub governance_bypass: bool,
@@ -1339,7 +1425,12 @@ fn build_remove_object_headers(version_id: Option<&str>, opts: &RemoveObjectOpti
 /// and silently creates a delete marker instead of removing the version, while
 /// the source stamps `VersionPurgeStatus=Complete` (backlog#799 B8 / #857).
 /// Non-replication callers always pass the version through unchanged.
-fn resolve_delete_api_version_id(version_id: Option<String>, opts: &RemoveObjectOptions) -> Option<String> {
+/// The `versionId` a replicated DELETE puts on the wire: none for a
+/// delete-marker creation (the target mints the marker; the source version
+/// travels in the internal headers for RustFS peers), the addressed version
+/// otherwise. A generic S3 target given the version id on a marker-creation
+/// DELETE would permanently delete that version instead.
+pub fn resolve_delete_api_version_id(version_id: Option<String>, opts: &RemoveObjectOptions) -> Option<String> {
     if opts.replication_request && opts.replication_delete_marker {
         None
     } else {
@@ -1892,6 +1983,125 @@ impl TargetClient {
             .map_err(Box::new)
     }
 
+    /// Candidate replicas by content identity on a target that mints its own
+    /// version ids: page `ListObjectVersions` under the exact key and report
+    /// the live versions whose ETag matches `source_etag`, newest first.
+    /// Delete markers and prefix siblings never match. Bounded to
+    /// [`FIND_VERSION_BY_ETAG_MAX_PAGES`] pages and
+    /// [`FIND_VERSION_BY_ETAG_MAX_MATCHES`] candidates so a key with a very
+    /// deep history cannot turn one convergence check into an unbounded scan;
+    /// a replica beyond that window reads as missing, which only costs a
+    /// re-PUT (today's behaviour), never a lost object.
+    ///
+    /// Content identity is not version identity: two source generations with
+    /// the same bytes have the same ETag. Callers drop the candidates other
+    /// source versions already claim through their ledgers and refuse an
+    /// [`ReplicaLocation::Ambiguous`] remainder before mutating or deleting.
+    pub async fn replica_candidates_by_etag(
+        &self,
+        bucket: &str,
+        object: &str,
+        source_etag: &str,
+    ) -> Result<Vec<String>, Box<SdkError<aws_sdk_s3::operation::list_object_versions::ListObjectVersionsError>>> {
+        let mut key_marker: Option<String> = None;
+        let mut version_id_marker: Option<String> = None;
+        let mut matches: Vec<String> = Vec::new();
+        for _ in 0..FIND_VERSION_BY_ETAG_MAX_PAGES {
+            let page = self
+                .client
+                .list_object_versions()
+                .bucket(bucket)
+                .prefix(object)
+                .max_keys(FIND_VERSION_BY_ETAG_PAGE_SIZE)
+                .set_key_marker(key_marker.take())
+                .set_version_id_marker(version_id_marker.take())
+                .send()
+                .await
+                .map_err(Box::new)?;
+            matches.extend(
+                page.versions()
+                    .iter()
+                    .filter(|version| {
+                        version.key() == Some(object)
+                            && version.version_id().is_some_and(|id| !id.is_empty())
+                            && replication_etags_match(Some(source_etag), version.e_tag())
+                    })
+                    .filter_map(|version| version.version_id().map(str::to_string)),
+            );
+            // A listing that moved past the exact key (every listed key is >=
+            // the prefix), ended, or already filled the candidate cap decides.
+            if matches.len() >= FIND_VERSION_BY_ETAG_MAX_MATCHES
+                || page
+                    .versions()
+                    .iter()
+                    .any(|version| version.key().is_some_and(|key| key > object))
+                || !page.is_truncated().unwrap_or(false)
+            {
+                break;
+            }
+            key_marker = page.next_key_marker().map(str::to_string);
+            version_id_marker = page.next_version_id_marker().map(str::to_string);
+            if key_marker.is_none() {
+                break;
+            }
+        }
+        matches.truncate(FIND_VERSION_BY_ETAG_MAX_MATCHES);
+        Ok(matches)
+    }
+
+    /// PutObjectRetention against a replica version on a target that does not
+    /// take retention through the replication PUT's own headers (it mints its
+    /// own version ids, so a re-PUT would create another version instead of
+    /// updating this one). Anti-loop marker always added.
+    pub async fn put_object_retention(
+        &self,
+        bucket: &str,
+        object: &str,
+        version_id: Option<String>,
+        mode: ObjectLockRetentionMode,
+        retain_until: aws_sdk_s3::primitives::DateTime,
+    ) -> Result<PutObjectRetentionOutput, Box<SdkError<PutObjectRetentionError>>> {
+        let headers = proxy_outbound_headers(HeaderMap::new());
+        self.client
+            .put_object_retention()
+            .bucket(bucket)
+            .key(object)
+            .set_version_id(resolve_read_api_version_id(version_id))
+            .retention(
+                ObjectLockRetention::builder()
+                    .mode(mode)
+                    .retain_until_date(retain_until)
+                    .build(),
+            )
+            .customize()
+            .map_request(move |req| apply_extra_headers(req, &headers))
+            .send()
+            .await
+            .map_err(Box::new)
+    }
+
+    /// PutObjectLegalHold counterpart of [`Self::put_object_retention`].
+    pub async fn put_object_legal_hold(
+        &self,
+        bucket: &str,
+        object: &str,
+        version_id: Option<String>,
+        status: ObjectLockLegalHoldStatus,
+    ) -> Result<PutObjectLegalHoldOutput, Box<SdkError<PutObjectLegalHoldError>>> {
+        let headers = proxy_outbound_headers(HeaderMap::new());
+        self.client
+            .put_object_legal_hold()
+            .bucket(bucket)
+            .key(object)
+            .set_version_id(resolve_read_api_version_id(version_id))
+            .legal_hold(ObjectLockLegalHold::builder().status(status).build())
+            .customize()
+            .map_request(move |req| apply_extra_headers(req, &headers))
+            .send()
+            .await
+            .map_err(Box::new)
+    }
+
     /// HEAD used by the read-proxy path (GET/HEAD of an object not yet
     /// replicated locally, MinIO `proxyHeadToRepTarget`).
     ///
@@ -2064,7 +2274,15 @@ impl TargetClient {
             }
         }
 
-        match builder
+        // A forwarded source checksum is this PUT's integrity header. In
+        // streaming-checksum mode (`RUSTFS_REPLICATION_STREAMING_CHECKSUMS`)
+        // the SDK would still add its default CRC32 trailer, and a target that
+        // receives both keeps the trailer's algorithm: a forwarded SHA256
+        // vanished from the replica while the source reported COMPLETED. Pin
+        // this request to WhenRequired so nothing is sent beside the source's
+        // own checksum.
+        let forwards_source_checksum = headers.keys().any(|name| name.as_str().starts_with("x-amz-checksum-"));
+        let mut operation = builder
             .bucket(bucket)
             .key(object)
             .content_length(size)
@@ -2084,10 +2302,14 @@ impl TargetClient {
                 }
 
                 Result::<_, aws_smithy_types::error::operation::BuildError>::Ok(req)
-            })
+            });
-            .send()
+        if forwards_source_checksum {
-            .await
+            operation = operation.config_override(
-        {
+                aws_sdk_s3::config::Builder::new()
+                    .request_checksum_calculation(aws_sdk_s3::config::RequestChecksumCalculation::WhenRequired),
+            );
+        }
+        match operation.send().await {
             Ok(output) => {
                 // Under SSE-KMS/DSSE or SSE-C the target's ETag is not the MD5
                 // of the stored plaintext, so it cannot be compared against the
@@ -2331,6 +2553,45 @@ impl TargetClient {
     }
 }
 
+/// Where a replica stands on a target that mints its own version ids, by
+/// content identity (exact key + ETag) after the candidates other source
+/// versions claim were removed. See
+/// [`TargetClient::replica_candidates_by_etag`].
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum ReplicaLocation {
+    /// No live version under the key carries the source ETag.
+    Missing,
+    /// Exactly one live version carries it: safe to address.
+    Unique(String),
+    /// More than one live version carries it (same bytes replicated for
+    /// several source generations). `newest` is the most recently listed
+    /// one — good enough to prove the replica exists, never good enough to
+    /// pick which one to mutate or delete.
+    Ambiguous { newest: String },
+}
+
+impl ReplicaLocation {
+    /// `matches` newest first, as the target listed them.
+    pub fn from_matches(mut matches: Vec<String>) -> Self {
+        match matches.len() {
+            0 => Self::Missing,
+            1 => Self::Unique(matches.remove(0)),
+            _ => Self::Ambiguous {
+                newest: matches.remove(0),
+            },
+        }
+    }
+
+    /// The version to read for existence/ETag checks, where an ambiguous
+    /// match is still a located replica.
+    pub fn any_version_id(&self) -> Option<&str> {
+        match self {
+            Self::Missing => None,
+            Self::Unique(version_id) | Self::Ambiguous { newest: version_id } => Some(version_id),
+        }
+    }
+}
+
 #[derive(Debug)]
 pub enum BucketTargetError {
     BucketRemoteTargetNotFound {
@@ -2507,13 +2768,21 @@ mod tests {
     }
 
     fn header_recording_target_client(response_headers: Vec<(String, String)>) -> (TargetClient, RecordedHeaders) {
+        header_recording_target_client_with_checksums(response_headers, replication_request_checksum_calculation())
+    }
+
+    fn header_recording_target_client_with_checksums(
+        response_headers: Vec<(String, String)>,
+        checksums: RequestChecksumCalculation,
+    ) -> (TargetClient, RecordedHeaders) {
         let request_headers: RecordedHeaders = Arc::new(std::sync::Mutex::new(Vec::new()));
         let connector = SharedHttpConnector::new(RecordingHeaderConnector {
             request_headers: Arc::clone(&request_headers),
             response_headers,
         });
         let http_client = http_client_fn(move |_settings, _components| connector.clone());
-        let client = s3_client_for_test(443, Some(http_client));
+        let client =
+            s3_client_for_endpoint_test_with_checksums("https://localhost:443".to_string(), Some(http_client), checksums);
         (
             TargetClient {
                 endpoint: "https://localhost:443".to_string(),
@@ -2680,6 +2949,47 @@ mod tests {
         }
     }
 
+    /// With streaming checksums enabled the SDK adds a CRC32 trailer to every
+    /// upload. A PUT that forwards the source's checksum must not get that
+    /// second algorithm: a target that receives both keeps the trailer's and
+    /// the forwarded SHA256 never reaches the replica (rustfs/backlog#2340).
+    #[tokio::test]
+    async fn streaming_put_object_with_forwarded_checksum_sends_no_sdk_checksum() {
+        let (client, recorded) =
+            header_recording_target_client_with_checksums(Vec::new(), RequestChecksumCalculation::WhenSupported);
+        let mut forwarded = PutObjectOptions::default();
+        forwarded.user_metadata.insert(
+            "x-amz-checksum-sha256".to_string(),
+            "OoJ3yNhRwv3wwtZoGqEIPrPX9xwTnfLl+ka0wStN1g0=".to_string(),
+        );
+        client
+            .put_object("target-bucket", "object", 4, streaming_test_body(b"data"), &forwarded)
+            .await
+            .expect("recorded put_object should succeed");
+        client
+            .put_object("target-bucket", "object", 4, streaming_test_body(b"data"), &PutObjectOptions::default())
+            .await
+            .expect("recorded put_object should succeed");
+        let recorded = recorded.lock().expect("recorded header lock should not be poisoned");
+        let with_forwarded = &recorded[0];
+        assert_eq!(
+            recorded_header(with_forwarded, "x-amz-checksum-sha256"),
+            Some("OoJ3yNhRwv3wwtZoGqEIPrPX9xwTnfLl+ka0wStN1g0=")
+        );
+        assert_eq!(
+            recorded_header(with_forwarded, "x-amz-trailer"),
+            None,
+            "the SDK must not add a trailer checksum"
+        );
+        assert_eq!(recorded_header(with_forwarded, "x-amz-sdk-checksum-algorithm"), None);
+        // Control: the same client still streams a trailer when nothing is forwarded.
+        let without_forwarded = &recorded[1];
+        assert!(
+            recorded_header(without_forwarded, "x-amz-trailer").is_some(),
+            "streaming mode must still apply to uploads without a forwarded checksum: {without_forwarded:?}"
+        );
+    }
+
     /// A forwarded source checksum already satisfies the rule; nothing is added.
     #[tokio::test]
     async fn locked_put_object_keeps_a_forwarded_source_checksum() {
@@ -3045,6 +3355,14 @@ mod tests {
     }
 
     fn s3_client_for_endpoint_test(endpoint: String, http_client: Option<SharedHttpClient>) -> S3Client {
+        s3_client_for_endpoint_test_with_checksums(endpoint, http_client, replication_request_checksum_calculation())
+    }
+
+    fn s3_client_for_endpoint_test_with_checksums(
+        endpoint: String,
+        http_client: Option<SharedHttpClient>,
+        checksums: RequestChecksumCalculation,
+    ) -> S3Client {
         let credentials = SdkCredentials::builder()
             .access_key_id("test-access")
             .secret_access_key("test-secret")
@@ -3058,7 +3376,7 @@ mod tests {
             .behavior_version(aws_sdk_s3::config::BehaviorVersion::latest())
             // Mirror the production remote-target builder so recorded requests
             // exercise the same checksum/framing behavior (#6853).
-            .request_checksum_calculation(replication_request_checksum_calculation());
+            .request_checksum_calculation(checksums);
         if let Some(http_client) = http_client {
             config = config.http_client(http_client);
         }
@@ -3221,6 +3539,64 @@ mod tests {
         assert!(message.contains("connection refused"));
     }
 
+    #[test]
+    fn same_replication_service_ignores_resync_and_bandwidth_edits() {
+        let base = BucketTarget {
+            endpoint: "target.example:9000".to_string(),
+            target_bucket: "replica".to_string(),
+            secure: true,
+            path: "on".to_string(),
+            arn: "arn:rustfs:replication:us-east-1:bucket:same".to_string(),
+            credentials: Some(Credentials {
+                access_key: "access".to_string(),
+                ..Default::default()
+            }),
+            ..Default::default()
+        };
+        let resync_edit = BucketTarget {
+            reset_id: "reset-1".to_string(),
+            bandwidth_limit: 1024,
+            ..base.clone()
+        };
+        assert!(same_replication_service(&resync_edit, &base));
+        for moved in [
+            BucketTarget {
+                endpoint: "other.example:9000".to_string(),
+                ..base.clone()
+            },
+            BucketTarget {
+                target_bucket: "other".to_string(),
+                ..base.clone()
+            },
+            BucketTarget {
+                secure: false,
+                ..base.clone()
+            },
+            BucketTarget {
+                credentials: Some(Credentials {
+                    access_key: "rotated".to_string(),
+                    ..Default::default()
+                }),
+                ..base.clone()
+            },
+        ] {
+            assert!(!same_replication_service(&moved, &base));
+        }
+    }
+
+    #[test]
+    fn version_identity_verdict_is_per_arn_and_forgotten_with_the_target() {
+        let sys = BucketTargetSys::default();
+        let arn = "arn:rustfs:replication:us-east-1:bucket:identity";
+        assert_eq!(sys.version_identity_capability(arn), VersionIdentityCapability::Unknown);
+        sys.record_version_identity_capability(arn, VersionIdentityCapability::MintsOwn);
+        assert_eq!(sys.version_identity_capability(arn), VersionIdentityCapability::MintsOwn);
+        assert_eq!(sys.version_identity_capability("other"), VersionIdentityCapability::Unknown);
+        // A rebuilt target may point at a different service.
+        sys.forget_version_identity_capability(arn);
+        assert_eq!(sys.version_identity_capability(arn), VersionIdentityCapability::Unknown);
+    }
+
     #[test]
     fn endpoint_health_key_preserves_explicit_port() {
         let url = Url::parse("https://remote.example:9443").expect("url should parse");

crates/ecstore/src/bucket/lifecycle/bucket_lifecycle_ops.rs

@@ -32,6 +32,7 @@ use crate::bucket::lifecycle::manual_transition_job::{
     record_manual_transition_worker_result_with_reason, renew_manual_transition_job_lease_if_owned,
     save_manual_transition_job_record_if_current, save_manual_transition_task_if_absent, update_manual_transition_job_record,
 };
+use crate::bucket::lifecycle::recovery_disposition_runtime::run_recovery_disposition_maintenance_loop;
 use crate::bucket::lifecycle::replication_sink;
 use crate::bucket::lifecycle::replication_sink::{
     DeleteReplicationConfigSnapshot, ReplicationObjectBridge, ReplicationStatusType, replication_state_to_filemeta,
@@ -149,6 +150,7 @@ pub type ExpiryOpType = Box<dyn ExpiryOp + Send + Sync + 'static>;
 static XXHASH_SEED: u64 = 0;
 static TIER_FREE_VERSION_RECOVERY_STARTED: OnceLock<()> = OnceLock::new();
 static MANUAL_TRANSITION_JOB_RECOVERY_STARTED: OnceLock<()> = OnceLock::new();
+static RECOVERY_DISPOSITION_MAINTENANCE_STARTED: OnceLock<()> = OnceLock::new();
 
 #[cfg(test)]
 #[derive(Default)]
@@ -2398,9 +2400,20 @@ pub async fn init_background_expiry(api: Arc<ECStore>) {
     let _ = spawn_tier_free_version_recovery_once(api.clone(), &TIER_FREE_VERSION_RECOVERY_STARTED);
     spawn_tier_delete_journal_recovery_once(api.clone());
     spawn_transition_transaction_recovery_once(api.clone());
+    spawn_recovery_disposition_maintenance_once(api.clone());
     spawn_manual_transition_job_recovery_once(api);
 }
 
+fn spawn_recovery_disposition_maintenance_once(api: Arc<ECStore>) -> Option<JoinHandle<()>> {
+    let cancel_token = api.ctx.background_cancel_token()?;
+    if RECOVERY_DISPOSITION_MAINTENANCE_STARTED.set(()).is_err() {
+        return None;
+    }
+    Some(tokio::spawn(async move {
+        run_recovery_disposition_maintenance_loop(api, cancel_token).await;
+    }))
+}
+
 fn spawn_manual_transition_job_recovery_once(api: Arc<ECStore>) -> Option<JoinHandle<()>> {
     if MANUAL_TRANSITION_JOB_RECOVERY_STARTED.set(()).is_err() {
         return None;
@@ -4023,6 +4036,10 @@ impl ManualTransitionRunReport {
             || self.skipped_queue_timeout > 0
     }
 
+    fn has_enqueue_backpressure(&self) -> bool {
+        self.skipped_queue_full > 0 || self.skipped_queue_closed > 0 || self.skipped_queue_timeout > 0
+    }
+
     pub fn was_truncated(&self) -> bool {
         self.truncated_by_limit || self.truncated_by_duration || self.cancelled
     }
@@ -4238,7 +4255,7 @@ pub async fn enqueue_transition_for_existing_objects_scoped(
             }
             report.scanned = report.scanned.saturating_add(1);
             enqueue_transition_with_lifecycle_report(Some(api.clone()), object, &lc, &src, &options, &mut report).await;
-            if report.has_partial_enqueue() {
+            if report.has_enqueue_backpressure() {
                 report.next_marker.clone_from(&previous_marker);
                 report.next_version_idmarker.clone_from(&previous_version_marker);
                 report.continuation_token =
@@ -9937,6 +9954,18 @@ mod tests {
         assert_eq!(report.skipped_queue_closed, 0);
         assert_eq!(report.skipped_queue_timeout, 0);
         assert!(report.has_partial_enqueue());
+        assert!(report.has_enqueue_backpressure());
+    }
+
+    #[test]
+    fn manual_transition_in_flight_skip_does_not_stop_the_scan() {
+        let options = ManualTransitionRunOptions::default();
+        let mut report = ManualTransitionRunReport::new("bucket", &options);
+
+        report.record_enqueue_outcome(TransitionEnqueueOutcome::AlreadyInFlight);
+
+        assert!(report.has_partial_enqueue());
+        assert!(!report.has_enqueue_backpressure());
     }
 
     #[test]

crates/ecstore/src/bucket/lifecycle/config_boundary.rs

@@ -41,6 +41,21 @@ where
     com::read_config(api, file).await
 }
 
+pub(crate) async fn read_config_limited_preserve_empty<S>(api: Arc<S>, file: &str, max_bytes: usize) -> Result<Vec<u8>>
+where
+    S: ObjectIO<
+            Error = Error,
+            RangeSpec = HTTPRangeSpec,
+            HeaderMap = HeaderMap,
+            ObjectOptions = ObjectOptions,
+            ObjectInfo = ObjectInfo,
+            GetObjectReader = GetObjectReader,
+            PutObjectReader = PutObjReader,
+        >,
+{
+    com::read_config_limited_preserve_empty(api, file, max_bytes).await
+}
+
 pub(crate) async fn read_config_with_metadata<S>(api: Arc<S>, file: &str, opts: &ObjectOptions) -> Result<(Vec<u8>, ObjectInfo)>
 where
     S: ObjectIO<
@@ -56,6 +71,26 @@ where
     com::read_config_with_metadata(api, file, opts).await
 }
 
+pub(crate) async fn read_config_limited_preserve_empty_with_metadata<S>(
+    api: Arc<S>,
+    file: &str,
+    opts: &ObjectOptions,
+    max_bytes: usize,
+) -> Result<(Vec<u8>, ObjectInfo)>
+where
+    S: ObjectIO<
+            Error = Error,
+            RangeSpec = HTTPRangeSpec,
+            HeaderMap = HeaderMap,
+            ObjectOptions = ObjectOptions,
+            ObjectInfo = ObjectInfo,
+            GetObjectReader = GetObjectReader,
+            PutObjectReader = PutObjReader,
+        >,
+{
+    com::read_config_limited_preserve_empty_with_metadata_opts(api, file, opts, max_bytes).await
+}
+
 pub(crate) async fn save_config<S>(api: Arc<S>, file: &str, data: Vec<u8>) -> Result<()>
 where
     S: ObjectIO<
@@ -126,20 +161,30 @@ where
             DeletedObject = DeletedObject,
         >,
 {
-    match api
+    delete_config_if_match_with_opts(api, file, etag, ObjectOptions::default()).await
-        .delete_object(
+}
-            RUSTFS_META_BUCKET,
+
-            file,
+pub(crate) async fn delete_config_if_match_with_opts<S>(
-            ObjectOptions {
+    api: Arc<S>,
-                http_preconditions: Some(HTTPPreconditions {
+    file: &str,
-                    if_match: Some(etag.to_string()),
+    etag: &str,
-                    ..Default::default()
+    mut options: ObjectOptions,
-                }),
+) -> Result<()>
-                ..Default::default()
+where
-            },
+    S: ObjectOperations<
-        )
+            Error = Error,
-        .await
+            ObjectInfo = ObjectInfo,
-    {
+            ObjectOptions = ObjectOptions,
+            FileInfo = FileInfo,
+            ObjectToDelete = ObjectToDelete,
+            DeletedObject = DeletedObject,
+        >,
+{
+    options.http_preconditions = Some(HTTPPreconditions {
+        if_match: Some(etag.to_string()),
+        ..Default::default()
+    });
+    match api.delete_object(RUSTFS_META_BUCKET, file, options).await {
         Ok(_) => Ok(()),
         Err(err) => {
             if err == Error::FileNotFound || matches!(err, Error::ObjectNotFound(_, _)) {

crates/ecstore/src/bucket/lifecycle/durable_namespace.rs

@@ -22,7 +22,7 @@ use super::{
     bucket_lifecycle_ops::{
         ManualTransitionQueueSnapshot, ManualTransitionRunReport, decode_manual_transition_continuation_token,
     },
-    manual_transition_job, recovery_control, tier_delete_journal, transition_transaction,
+    manual_transition_job, recovery_control, recovery_disposition, recovery_export, tier_delete_journal, transition_transaction,
 };
 use crate::error::{Error, Result};
 use crate::services::tier::tier_probe_intent;
@@ -42,6 +42,8 @@ pub(crate) enum DurableIlmRecordKind {
     ManualTransitionTask,
     ManualTransitionWorkerResult,
     RecoveryControl,
+    RecoveryExport,
+    RecoveryDisposition,
 }
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
@@ -112,8 +114,20 @@ pub(crate) const RECOVERY_CONTROL_NAMESPACE: DurableIlmNamespace = DurableIlmNam
     max_record_size: recovery_control::MAX_ILM_RECOVERY_CONTROL_SIZE,
     kind: DurableIlmRecordKind::RecoveryControl,
 };
+pub(crate) const RECOVERY_EXPORT_NAMESPACE: DurableIlmNamespace = DurableIlmNamespace {
+    name: "recovery-export",
+    prefix: recovery_export::ILM_RECOVERY_EXPORT_PREFIX,
+    max_record_size: recovery_export::MAX_ILM_RECOVERY_EXPORT_SIZE,
+    kind: DurableIlmRecordKind::RecoveryExport,
+};
+pub(crate) const RECOVERY_DISPOSITION_NAMESPACE: DurableIlmNamespace = DurableIlmNamespace {
+    name: "recovery-disposition",
+    prefix: recovery_disposition::ILM_RECOVERY_DISPOSITION_PREFIX,
+    max_record_size: recovery_disposition::MAX_ILM_RECOVERY_DISPOSITION_SIZE,
+    kind: DurableIlmRecordKind::RecoveryDisposition,
+};
 
-pub(crate) const DURABLE_ILM_NAMESPACES: [DurableIlmNamespace; 10] = [
+pub(crate) const DURABLE_ILM_NAMESPACES: [DurableIlmNamespace; 12] = [
     TIER_DELETE_JOURNAL_NAMESPACE,
     TIER_DELETE_JOURNAL_V6_NAMESPACE,
     TIER_DELETE_DISPATCH_MANIFEST_NAMESPACE,
@@ -124,6 +138,8 @@ pub(crate) const DURABLE_ILM_NAMESPACES: [DurableIlmNamespace; 10] = [
     MANUAL_TRANSITION_TASK_NAMESPACE,
     MANUAL_TRANSITION_WORKER_RESULT_NAMESPACE,
     RECOVERY_CONTROL_NAMESPACE,
+    RECOVERY_EXPORT_NAMESPACE,
+    RECOVERY_DISPOSITION_NAMESPACE,
 ];
 
 #[derive(Debug, Clone, PartialEq, Eq)]
@@ -261,6 +277,28 @@ pub(crate) enum DurableIlmRecordCheckpoint {
         #[serde(default, skip_serializing_if = "Option::is_none")]
         owner_fence_sha256: Option<String>,
     },
+    RecoveryExport {
+        content_sha256: String,
+        source_generation_sha256: String,
+        topology_generation: String,
+        member_epochs_sha256: String,
+        creator_sha256: String,
+        retain_until_unix_nanos: i64,
+    },
+    RecoveryDisposition {
+        content_sha256: String,
+        identity_sha256: String,
+        copy_manifest_sha256: String,
+        copy_manifest_count: usize,
+        created_at_unix_nanos: i64,
+        revision: u64,
+        state: recovery_disposition::IlmRecoveryDispositionState,
+        owner_fence_sha256: Option<String>,
+        owner_lease_acquired_at_unix_nanos: Option<i64>,
+        owner_lease_expires_at_unix_nanos: Option<i64>,
+        confirmed_absent_sha256: Vec<String>,
+        retain_until_unix_nanos: i64,
+    },
 }
 
 impl DurableIlmRecordCheckpoint {
@@ -275,7 +313,9 @@ impl DurableIlmRecordCheckpoint {
             | Self::ManualTransitionScope { content_sha256, .. }
             | Self::ManualTransitionTask { content_sha256 }
             | Self::ManualTransitionWorkerResult { content_sha256 }
-            | Self::RecoveryControl { content_sha256, .. } => content_sha256,
+            | Self::RecoveryControl { content_sha256, .. }
+            | Self::RecoveryExport { content_sha256, .. }
+            | Self::RecoveryDisposition { content_sha256, .. } => content_sha256,
         }
     }
 
@@ -316,6 +356,9 @@ impl DurableIlmRecordCheckpoint {
             {
                 return Err(Error::other("durable ILM tier delete journal checkpoint is invalid"));
             }
+            if !recovery_disposition_checkpoint_is_valid(checkpoint) {
+                return Err(Error::other("durable ILM recovery disposition checkpoint is invalid"));
+            }
         }
         if self == next {
             if let Self::ManualTransitionJob {
@@ -456,9 +499,7 @@ impl DurableIlmRecordCheckpoint {
                 },
             ) => {
                 previous_identity == next_identity
-                    && transition_state_distance(*previous_state, *next_state)
+                    && transition_state_revision_is_successor(*previous_state, *previous_revision, *next_state, *next_revision)
-                        .and_then(|distance| previous_revision.checked_add(distance))
-                        .is_some_and(|expected_revision| *next_revision == expected_revision)
                     && (!previous_remote_version_known || previous_remote_version == next_remote_version)
             }
             (
@@ -594,6 +635,83 @@ impl DurableIlmRecordCheckpoint {
                     && previous_attempts == next_attempts;
                 adjacent && (claim || source_refresh || completion)
             }
+            (
+                Self::RecoveryDisposition {
+                    identity_sha256: previous_identity,
+                    copy_manifest_sha256: previous_manifest,
+                    copy_manifest_count: previous_manifest_count,
+                    created_at_unix_nanos: previous_created_at,
+                    revision: previous_revision,
+                    state: previous_state,
+                    owner_fence_sha256: previous_owner,
+                    owner_lease_acquired_at_unix_nanos: previous_owner_acquired,
+                    owner_lease_expires_at_unix_nanos: previous_owner_expires,
+                    confirmed_absent_sha256: previous_confirmed,
+                    retain_until_unix_nanos: previous_retain_until,
+                    ..
+                },
+                Self::RecoveryDisposition {
+                    identity_sha256: next_identity,
+                    copy_manifest_sha256: next_manifest,
+                    copy_manifest_count: next_manifest_count,
+                    created_at_unix_nanos: next_created_at,
+                    revision: next_revision,
+                    state: next_state,
+                    owner_fence_sha256: next_owner,
+                    owner_lease_acquired_at_unix_nanos: next_owner_acquired,
+                    owner_lease_expires_at_unix_nanos: next_owner_expires,
+                    confirmed_absent_sha256: next_confirmed,
+                    retain_until_unix_nanos: next_retain_until,
+                    ..
+                },
+            ) => {
+                use recovery_disposition::IlmRecoveryDispositionState::{Applying, Completed, Prepared};
+
+                let immutable_identity_matches = previous_identity == next_identity
+                    && previous_manifest == next_manifest
+                    && previous_manifest_count == next_manifest_count
+                    && previous_created_at == next_created_at
+                    && previous_retain_until == next_retain_until;
+                let adjacent = previous_revision.checked_add(1) == Some(*next_revision);
+                let progress_is_monotonic = sorted_sha256_set_is_subset(previous_confirmed, next_confirmed);
+                let legal_edge = match (previous_state, next_state) {
+                    (Prepared, Prepared) => {
+                        let claim = previous_owner.is_none() && next_owner.is_some();
+                        let takeover = previous_owner.is_some()
+                            && previous_owner != next_owner
+                            && previous_owner_expires
+                                .zip(*next_owner_acquired)
+                                .is_some_and(|(expires, acquired)| acquired >= expires);
+                        previous_confirmed == next_confirmed && (claim || takeover)
+                    }
+                    (Prepared, Applying) => {
+                        previous_confirmed == next_confirmed
+                            && previous_owner.is_some()
+                            && previous_owner == next_owner
+                            && previous_owner_acquired == next_owner_acquired
+                            && previous_owner_expires == next_owner_expires
+                    }
+                    (Applying, Applying) => {
+                        let progress = previous_owner == next_owner
+                            && previous_owner_acquired == next_owner_acquired
+                            && previous_owner_expires == next_owner_expires
+                            && previous_confirmed.len().checked_add(1) == Some(next_confirmed.len());
+                        let takeover = previous_owner.is_some()
+                            && previous_owner != next_owner
+                            && previous_confirmed == next_confirmed
+                            && previous_owner_expires
+                                .zip(*next_owner_acquired)
+                                .is_some_and(|(expires, acquired)| acquired >= expires);
+                        progress || takeover
+                    }
+                    (Applying, Completed) => {
+                        previous_owner.is_some() && next_owner.is_none() && previous_confirmed == next_confirmed
+                    }
+                    _ => false,
+                };
+
+                immutable_identity_matches && adjacent && progress_is_monotonic && legal_edge
+            }
             _ => false,
         };
 
@@ -611,6 +729,11 @@ impl DurableIlmRecordCheckpoint {
     /// after the exact terminal ETag and terminal receipt were committed, to
     /// purge older object versions exposed by that deletion.
     pub(crate) fn is_predecessor_of_terminal(&self, terminal: &Self) -> bool {
+        for checkpoint in [self, terminal] {
+            if !recovery_disposition_checkpoint_is_valid(checkpoint) {
+                return false;
+            }
+        }
         if let Self::TierProbeIntent { state, .. } = terminal
             && !matches!(
                 state,
@@ -627,6 +750,11 @@ impl DurableIlmRecordCheckpoint {
         {
             return false;
         }
+        if let Self::RecoveryDisposition { state, .. } = terminal
+            && state != &recovery_disposition::IlmRecoveryDispositionState::Completed
+        {
+            return false;
+        }
         if self == terminal || self.validate_successor(terminal).is_ok() {
             return true;
         }
@@ -752,11 +880,121 @@ impl DurableIlmRecordCheckpoint {
                     && terminal_revision > previous_revision
                     && terminal_attempts >= previous_attempts
             }
+            (
+                Self::RecoveryDisposition {
+                    identity_sha256: previous_identity,
+                    copy_manifest_sha256: previous_manifest,
+                    copy_manifest_count: previous_manifest_count,
+                    created_at_unix_nanos: previous_created_at,
+                    revision: previous_revision,
+                    state: previous_state,
+                    owner_fence_sha256: previous_owner,
+                    confirmed_absent_sha256: previous_confirmed,
+                    retain_until_unix_nanos: previous_retain_until,
+                    ..
+                },
+                Self::RecoveryDisposition {
+                    identity_sha256: terminal_identity,
+                    copy_manifest_sha256: terminal_manifest,
+                    copy_manifest_count: terminal_manifest_count,
+                    created_at_unix_nanos: terminal_created_at,
+                    revision: terminal_revision,
+                    state: recovery_disposition::IlmRecoveryDispositionState::Completed,
+                    confirmed_absent_sha256: terminal_confirmed,
+                    retain_until_unix_nanos: terminal_retain_until,
+                    ..
+                },
+            ) => {
+                matches!(
+                    previous_state,
+                    recovery_disposition::IlmRecoveryDispositionState::Prepared
+                        | recovery_disposition::IlmRecoveryDispositionState::Applying
+                ) && previous_identity == terminal_identity
+                    && previous_manifest == terminal_manifest
+                    && previous_manifest_count == terminal_manifest_count
+                    && previous_created_at == terminal_created_at
+                    && previous_retain_until == terminal_retain_until
+                    && terminal_revision.checked_sub(*previous_revision).is_some_and(|distance| {
+                        let minimum_distance = match previous_state {
+                            recovery_disposition::IlmRecoveryDispositionState::Prepared if previous_owner.is_some() => 3,
+                            recovery_disposition::IlmRecoveryDispositionState::Prepared => 4,
+                            recovery_disposition::IlmRecoveryDispositionState::Applying
+                                if previous_confirmed.len() == *previous_manifest_count =>
+                            {
+                                1
+                            }
+                            recovery_disposition::IlmRecoveryDispositionState::Applying => 2,
+                            recovery_disposition::IlmRecoveryDispositionState::Completed => u64::MAX,
+                        };
+                        distance >= minimum_distance
+                    })
+                    && sorted_sha256_set_is_subset(previous_confirmed, terminal_confirmed)
+            }
             _ => false,
         }
     }
 }
 
+fn recovery_disposition_checkpoint_is_valid(checkpoint: &DurableIlmRecordCheckpoint) -> bool {
+    use recovery_disposition::IlmRecoveryDispositionState::{Applying, Completed, Prepared};
+
+    let DurableIlmRecordCheckpoint::RecoveryDisposition {
+        content_sha256,
+        identity_sha256,
+        copy_manifest_sha256,
+        copy_manifest_count,
+        created_at_unix_nanos,
+        revision,
+        state,
+        owner_fence_sha256,
+        owner_lease_acquired_at_unix_nanos,
+        owner_lease_expires_at_unix_nanos,
+        confirmed_absent_sha256,
+        retain_until_unix_nanos,
+    } = checkpoint
+    else {
+        return true;
+    };
+    let owner_fence_sha256 = owner_fence_sha256.as_deref();
+
+    is_canonical_sha256(content_sha256)
+        && is_canonical_sha256(identity_sha256)
+        && is_canonical_sha256(copy_manifest_sha256)
+        && *copy_manifest_count > 0
+        && *created_at_unix_nanos > 0
+        && *revision > 0
+        && *retain_until_unix_nanos > 0
+        && owner_fence_sha256.is_none_or(is_canonical_sha256)
+        && match (
+            owner_fence_sha256,
+            *owner_lease_acquired_at_unix_nanos,
+            *owner_lease_expires_at_unix_nanos,
+        ) {
+            (None, None, None) => true,
+            (Some(_), Some(acquired), Some(expires)) => acquired > 0 && expires > acquired,
+            _ => false,
+        }
+        && confirmed_absent_sha256.len() <= *copy_manifest_count
+        && confirmed_absent_sha256.iter().all(|digest| is_canonical_sha256(digest))
+        && confirmed_absent_sha256.windows(2).all(|pair| pair[0] < pair[1])
+        && match *state {
+            Prepared => confirmed_absent_sha256.is_empty(),
+            Applying => owner_fence_sha256.is_some(),
+            Completed => owner_fence_sha256.is_none() && confirmed_absent_sha256.len() == *copy_manifest_count,
+        }
+}
+
+fn is_canonical_sha256(value: &str) -> bool {
+    is_sha256_checksum(value)
+        && !value
+            .bytes()
+            .any(|byte| byte.is_ascii_hexdigit() && byte.is_ascii_uppercase())
+}
+
+fn sorted_sha256_set_is_subset(subset: &[String], superset: &[String]) -> bool {
+    subset.iter().all(|candidate| superset.binary_search(candidate).is_ok())
+}
+
 fn tier_delete_dispatch_parent_progress_delta(
     previous_sequence: u64,
     previous_completed_journals: u64,
@@ -790,6 +1028,22 @@ fn transition_state_distance(
     }
 }
 
+fn transition_state_revision_is_successor(
+    from: transition_transaction::TransitionTransactionState,
+    from_revision: u64,
+    to: transition_transaction::TransitionTransactionState,
+    to_revision: u64,
+) -> bool {
+    use transition_transaction::TransitionTransactionState::{LocalCommitStarted, UploadOutcomeUnknown};
+
+    if from == UploadOutcomeUnknown && from_revision == 1 && to == LocalCommitStarted {
+        return to_revision == 2;
+    }
+    transition_state_distance(from, to)
+        .and_then(|distance| from_revision.checked_add(distance))
+        .is_some_and(|expected_revision| to_revision == expected_revision)
+}
+
 fn tier_probe_state_reaches(
     from: tier_probe_intent::TierProbeIntentState,
     to: tier_probe_intent::TierProbeIntentState,
@@ -1348,6 +1602,55 @@ pub(crate) fn validate_durable_ilm_record(path: &str, data: &[u8]) -> Result<Val
                 },
             )
         }
+        DurableIlmRecordKind::RecoveryExport => {
+            let (protocol, export_id) = recovery_export::recovery_export_id_from_record_object_name(path)?;
+            let export = recovery_export::IlmRecoveryExport::decode(&export_id, data)?;
+            let canonical = recovery_export::recovery_export_record_object_name(protocol, &export_id)?;
+            if canonical != path || export.protocol != protocol {
+                return Err(Error::other("ILM recovery export path is not canonical"));
+            }
+            let source_generation_sha256 = checkpoint_hash(&export.source_generation)?;
+            (
+                "export_id",
+                export_id,
+                DurableIlmRecordCheckpoint::RecoveryExport {
+                    content_sha256,
+                    source_generation_sha256,
+                    topology_generation: export.topology_generation,
+                    member_epochs_sha256: export.member_epochs_sha256,
+                    creator_sha256: export.creator_sha256,
+                    retain_until_unix_nanos: export.retain_until_unix_nanos,
+                },
+            )
+        }
+        DurableIlmRecordKind::RecoveryDisposition => {
+            // The disposition module owns strict schema, checksum, canonical
+            // path, immutable-manifest, and state-specific validation. Keep
+            // this boundary limited to decommission identity/checkpoint
+            // projection so the two readers cannot accept different records.
+            let disposition = recovery_disposition::decode_recovery_disposition_checkpoint(path, data)?;
+            if disposition.content_sha256 != content_sha256 {
+                return Err(Error::other("ILM recovery disposition checkpoint content digest is invalid"));
+            }
+            (
+                "disposition_id",
+                disposition.disposition_id,
+                DurableIlmRecordCheckpoint::RecoveryDisposition {
+                    content_sha256: disposition.content_sha256,
+                    identity_sha256: disposition.identity_sha256,
+                    copy_manifest_sha256: disposition.copy_manifest_sha256,
+                    copy_manifest_count: disposition.copy_manifest_count,
+                    created_at_unix_nanos: disposition.created_at_unix_nanos,
+                    revision: disposition.revision,
+                    state: disposition.state,
+                    owner_fence_sha256: disposition.owner_fence_sha256,
+                    owner_lease_acquired_at_unix_nanos: disposition.owner_lease_acquired_at_unix_nanos,
+                    owner_lease_expires_at_unix_nanos: disposition.owner_lease_expires_at_unix_nanos,
+                    confirmed_absent_sha256: disposition.confirmed_absent_sha256,
+                    retain_until_unix_nanos: disposition.retain_until_unix_nanos,
+                },
+            )
+        }
         DurableIlmRecordKind::ManualTransitionJob => {
             let job_id = manual_transition_job::manual_transition_job_id_from_record_object_name(path)
                 .map_err(|err| Error::other(err.to_string()))?;
@@ -1503,6 +1806,203 @@ mod tests {
         }
     }
 
+    fn recovery_disposition_checkpoint(
+        revision: u64,
+        state: recovery_disposition::IlmRecoveryDispositionState,
+        owner_fence: Option<&str>,
+        confirmed_absent_sha256: Vec<String>,
+    ) -> DurableIlmRecordCheckpoint {
+        let (owner_lease_acquired_at_unix_nanos, owner_lease_expires_at_unix_nanos) = match owner_fence {
+            Some("f") => (Some(10), Some(20)),
+            Some(_) => (Some(1), Some(10)),
+            None => (None, None),
+        };
+        DurableIlmRecordCheckpoint::RecoveryDisposition {
+            content_sha256: format!("{revision:064x}"),
+            identity_sha256: "a".repeat(64),
+            copy_manifest_sha256: "d".repeat(64),
+            copy_manifest_count: 2,
+            created_at_unix_nanos: 1_700_000_000_000_000_000,
+            revision,
+            state,
+            owner_fence_sha256: owner_fence.map(|digest| digest.repeat(64)),
+            owner_lease_acquired_at_unix_nanos,
+            owner_lease_expires_at_unix_nanos,
+            confirmed_absent_sha256,
+            retain_until_unix_nanos: 1_820_000_000_000_000_000,
+        }
+    }
+
+    #[test]
+    fn recovery_disposition_namespace_is_registered_without_shadowing_its_root() {
+        let disposition_id = "a".repeat(64);
+        let path = format!(
+            "{}/tier_delete_journal/{}/{}/{}.json",
+            recovery_disposition::ILM_RECOVERY_DISPOSITION_PREFIX,
+            &disposition_id[..2],
+            &disposition_id[2..4],
+            disposition_id
+        );
+        let namespace = classify_durable_ilm_record(&path)
+            .expect("recovery disposition path should classify")
+            .expect("recovery disposition should be durable");
+
+        assert_eq!(namespace, &RECOVERY_DISPOSITION_NAMESPACE);
+        assert!(classify_durable_ilm_record(recovery_disposition::ILM_RECOVERY_DISPOSITION_PREFIX).is_err());
+    }
+
+    #[test]
+    fn recovery_disposition_checkpoint_accepts_only_monotonic_progress() {
+        use recovery_disposition::IlmRecoveryDispositionState::{Applying, Completed, Prepared};
+
+        let first_copy = "b".repeat(64);
+        let second_copy = "c".repeat(64);
+        let prepared = recovery_disposition_checkpoint(1, Prepared, None, Vec::new());
+        let claimed = recovery_disposition_checkpoint(2, Prepared, Some("e"), Vec::new());
+        let applying = recovery_disposition_checkpoint(3, Applying, Some("e"), Vec::new());
+        let first_absent = recovery_disposition_checkpoint(4, Applying, Some("e"), vec![first_copy.clone()]);
+        let taken_over = recovery_disposition_checkpoint(5, Applying, Some("f"), vec![first_copy.clone()]);
+        let all_absent = recovery_disposition_checkpoint(6, Applying, Some("f"), vec![first_copy.clone(), second_copy.clone()]);
+        let completed = recovery_disposition_checkpoint(7, Completed, None, vec![first_copy.clone(), second_copy.clone()]);
+
+        prepared
+            .validate_successor(&claimed)
+            .expect("Prepared should record an owner claim without absence progress");
+        claimed
+            .validate_successor(&applying)
+            .expect("Prepared should advance to Applying without folding in deletion progress");
+        applying
+            .validate_successor(&first_absent)
+            .expect("Applying should append newly confirmed absent copies");
+        first_absent
+            .validate_successor(&taken_over)
+            .expect("Applying should record a fenced owner takeover without losing progress");
+        taken_over
+            .validate_successor(&all_absent)
+            .expect("Applying should preserve every earlier confirmation while making progress");
+        all_absent
+            .validate_successor(&completed)
+            .expect("a fully confirmed manifest should advance to Completed");
+
+        assert!(
+            prepared.validate_successor(&completed).is_err(),
+            "adjacent receipt updates must not skip Applying"
+        );
+        assert!(
+            first_absent
+                .validate_successor(&recovery_disposition_checkpoint(5, Applying, Some("e"), Vec::new()))
+                .is_err(),
+            "confirmed-absent progress must not move backwards"
+        );
+        assert!(
+            applying
+                .validate_successor(&recovery_disposition_checkpoint(4, Completed, None, vec![first_copy.clone()]))
+                .is_err(),
+            "Completed must cover the complete immutable copy manifest"
+        );
+        assert!(
+            completed
+                .validate_successor(&recovery_disposition_checkpoint(7, Applying, Some("e"), vec![second_copy]))
+                .is_err(),
+            "Completed is terminal"
+        );
+        assert!(
+            first_absent
+                .validate_successor(&recovery_disposition_checkpoint(5, Applying, Some("e"), vec![first_copy.clone()]))
+                .is_err(),
+            "a same-state revision bump must change the owner fence or absence progress"
+        );
+        assert!(
+            applying
+                .validate_successor(&recovery_disposition_checkpoint(4, Applying, None, vec![first_copy]))
+                .is_err(),
+            "Applying must retain a fenced owner"
+        );
+
+        let mut noncanonical_identity = claimed.clone();
+        if let DurableIlmRecordCheckpoint::RecoveryDisposition { identity_sha256, .. } = &mut noncanonical_identity {
+            *identity_sha256 = "A".repeat(64);
+        }
+        assert!(prepared.validate_successor(&noncanonical_identity).is_err());
+        let mut changed_created_at = claimed;
+        if let DurableIlmRecordCheckpoint::RecoveryDisposition {
+            created_at_unix_nanos, ..
+        } = &mut changed_created_at
+        {
+            *created_at_unix_nanos += 1;
+        }
+        assert!(prepared.validate_successor(&changed_created_at).is_err());
+
+        let mut early_takeover = taken_over;
+        if let DurableIlmRecordCheckpoint::RecoveryDisposition {
+            owner_lease_acquired_at_unix_nanos,
+            ..
+        } = &mut early_takeover
+        {
+            *owner_lease_acquired_at_unix_nanos = Some(9);
+        }
+        assert!(first_absent.validate_successor(&early_takeover).is_err());
+        assert!(
+            applying
+                .validate_successor(&recovery_disposition_checkpoint(
+                    4,
+                    Applying,
+                    Some("e"),
+                    vec!["c".repeat(64), "b".repeat(64)],
+                ))
+                .is_err(),
+            "confirmed-absent entries must be a canonical sorted set"
+        );
+    }
+
+    #[test]
+    fn recovery_disposition_terminal_predecessor_requires_exact_identity_and_full_manifest() {
+        use recovery_disposition::IlmRecoveryDispositionState::{Applying, Completed, Prepared};
+
+        let first_copy = "b".repeat(64);
+        let second_copy = "c".repeat(64);
+        let prepared = recovery_disposition_checkpoint(1, Prepared, None, Vec::new());
+        let applying = recovery_disposition_checkpoint(3, Applying, Some("e"), vec![first_copy.clone()]);
+        let completed = recovery_disposition_checkpoint(5, Completed, None, vec![first_copy.clone(), second_copy]);
+
+        assert!(prepared.is_predecessor_of_terminal(&completed));
+        assert!(applying.is_predecessor_of_terminal(&completed));
+        assert!(
+            !prepared.is_predecessor_of_terminal(&recovery_disposition_checkpoint(2, Applying, Some("e"), Vec::new())),
+            "a nonterminal disposition must not authorize terminal cleanup"
+        );
+        assert!(
+            !prepared.is_predecessor_of_terminal(&recovery_disposition_checkpoint(
+                4,
+                Completed,
+                None,
+                vec![first_copy.clone(), "c".repeat(64)],
+            )),
+            "terminal proof must leave enough revisions for claim, apply, progress, and completion"
+        );
+        assert!(
+            !applying.is_predecessor_of_terminal(&recovery_disposition_checkpoint(
+                4,
+                Completed,
+                None,
+                vec![first_copy.clone(), "c".repeat(64)],
+            )),
+            "an incomplete Applying checkpoint cannot complete without a progress generation"
+        );
+
+        let mut other_identity = completed;
+        if let DurableIlmRecordCheckpoint::RecoveryDisposition { identity_sha256, .. } = &mut other_identity {
+            *identity_sha256 = "e".repeat(64);
+        }
+        assert!(!prepared.is_predecessor_of_terminal(&other_identity));
+
+        let incomplete_terminal = recovery_disposition_checkpoint(4, Completed, None, vec![first_copy]);
+        assert!(
+            !prepared.is_predecessor_of_terminal(&incomplete_terminal),
+            "a partial confirmed-absent set must not become terminal proof"
+        );
+    }
+
     fn tier_probe_intent_fixture() -> tier_probe_intent::TierProbeIntent {
         let probe_id = Uuid::parse_str("36e2220e-9ad2-495b-b3bc-c4d2caf70a31").expect("fixture uuid should parse");
         tier_probe_intent::TierProbeIntent {
@@ -1687,6 +2187,64 @@ mod tests {
         );
     }
 
+    #[test]
+    fn transition_checkpoint_accepts_only_the_distinguishable_compact_edge() {
+        let identity_sha256 = "a".repeat(64);
+        let unknown_remote_sha256 = "b".repeat(64);
+        let known_remote_sha256 = "c".repeat(64);
+        let checkpoint = |revision, state, remote_version_sha256: String, remote_version_known| {
+            DurableIlmRecordCheckpoint::TransitionTransaction {
+                content_sha256: format!("{revision:064x}"),
+                identity_sha256: identity_sha256.clone(),
+                remote_version_sha256,
+                remote_version_known,
+                revision,
+                state,
+            }
+        };
+        let compact_unknown = checkpoint(
+            1,
+            transition_transaction::TransitionTransactionState::UploadOutcomeUnknown,
+            unknown_remote_sha256.clone(),
+            false,
+        );
+        let compact_local_commit = checkpoint(
+            2,
+            transition_transaction::TransitionTransactionState::LocalCommitStarted,
+            known_remote_sha256.clone(),
+            true,
+        );
+        compact_unknown
+            .validate_successor(&compact_local_commit)
+            .expect("compact pre-upload fence should advance directly to the exact local-commit fence");
+
+        let legacy_unknown = checkpoint(
+            2,
+            transition_transaction::TransitionTransactionState::UploadOutcomeUnknown,
+            unknown_remote_sha256,
+            false,
+        );
+        let invalid_legacy_skip = checkpoint(
+            3,
+            transition_transaction::TransitionTransactionState::LocalCommitStarted,
+            known_remote_sha256.clone(),
+            true,
+        );
+        assert!(
+            legacy_unknown.validate_successor(&invalid_legacy_skip).is_err(),
+            "legacy UploadOutcomeUnknown@2 must not masquerade as the compact edge"
+        );
+        let valid_legacy_skip = checkpoint(
+            4,
+            transition_transaction::TransitionTransactionState::LocalCommitStarted,
+            known_remote_sha256,
+            true,
+        );
+        legacy_unknown
+            .validate_successor(&valid_legacy_skip)
+            .expect("legacy receipts may still observe the existing two-edge state advance");
+    }
+
     #[test]
     fn tier_delete_dispatch_manifest_namespace_validates_monotonic_branches() {
         use tier_delete_journal::TierDeleteDispatchManifestState::{Aborted, Aborting, Completed, DispatchAuthorized, Preparing};

crates/ecstore/src/bucket/lifecycle/mod.rs

@@ -25,6 +25,9 @@ mod object_handlers_common;
 mod object_lock_boundary;
 pub use self::core as lifecycle;
 pub mod recovery_control;
+pub mod recovery_disposition;
+pub(crate) mod recovery_disposition_runtime;
+pub mod recovery_export;
 mod replication_sink;
 pub mod rule;
 mod runtime_boundary;

crates/ecstore/src/bucket/lifecycle/recovery_control.rs

@@ -168,7 +168,7 @@ impl IlmRecoverySourceGeneration {
         Ok(generation)
     }
 
-    fn validate(&self) -> Result<()> {
+    pub(crate) fn validate(&self) -> Result<()> {
         if self.source_schema.trim().is_empty() {
             return Err(IlmRecoveryControlError::Corrupt("source schema is empty"));
         }
@@ -485,6 +485,40 @@ impl IlmRecoveryControl {
         self.validate()
     }
 
+    pub fn abandon_for_operator(&mut self, expected_source_generation: &IlmRecoverySourceGeneration) -> Result<()> {
+        if self.owner.is_some()
+            || self.classification != IlmRecoveryClassification::RetainedAmbiguous
+            || &self.observed_source_generation != expected_source_generation
+        {
+            return Err(IlmRecoveryControlError::InvalidSuccessor(
+                "operator abandonment requires the exact ownerless retained source generation",
+            ));
+        }
+        self.bump_revision()?;
+        self.classification = IlmRecoveryClassification::Abandoned;
+        self.validate()
+    }
+
+    pub fn retry_for_operator(&mut self, expected_source_generation: &IlmRecoverySourceGeneration) -> Result<()> {
+        if self.owner.is_some()
+            || !matches!(
+                self.classification,
+                IlmRecoveryClassification::RetainedAmbiguous | IlmRecoveryClassification::OperatorRequired
+            )
+            || self.attempt_count == u64::MAX
+            || &self.observed_source_generation != expected_source_generation
+        {
+            return Err(IlmRecoveryControlError::InvalidSuccessor(
+                "operator retry requires the exact ownerless retained source generation",
+            ));
+        }
+        self.bump_revision()?;
+        self.classification = IlmRecoveryClassification::Retrying;
+        self.consecutive_failure_count = 0;
+        self.next_attempt_at_unix_nanos = None;
+        self.validate()
+    }
+
     pub fn validate_successor(&self, next: &Self) -> Result<()> {
         self.validate()?;
         next.validate()?;
@@ -508,12 +542,58 @@ impl IlmRecoveryControl {
                 self.validate_failure_successor(next)
             }
             (Some(_), None) => self.validate_finish_successor(next),
+            (None, None)
+                if self.classification == IlmRecoveryClassification::RetainedAmbiguous
+                    && next.classification == IlmRecoveryClassification::Abandoned =>
+            {
+                self.validate_operator_abandon_successor(next)
+            }
+            (None, None)
+                if matches!(
+                    self.classification,
+                    IlmRecoveryClassification::RetainedAmbiguous | IlmRecoveryClassification::OperatorRequired
+                ) && next.classification == IlmRecoveryClassification::Retrying =>
+            {
+                self.validate_operator_retry_successor(next)
+            }
             (None, None) => Err(IlmRecoveryControlError::InvalidSuccessor(
                 "ownerless control cannot advance without a claim",
             )),
         }
     }
 
+    fn validate_operator_abandon_successor(&self, next: &Self) -> Result<()> {
+        if next.observed_source_generation != self.observed_source_generation
+            || next.attempt_count != self.attempt_count
+            || next.consecutive_failure_count != self.consecutive_failure_count
+            || next.first_failure_at_unix_nanos != self.first_failure_at_unix_nanos
+            || next.last_failure_at_unix_nanos != self.last_failure_at_unix_nanos
+            || next.next_attempt_at_unix_nanos != self.next_attempt_at_unix_nanos
+            || next.last_error_code != self.last_error_code
+        {
+            return Err(IlmRecoveryControlError::InvalidSuccessor(
+                "operator abandonment changed recovery history or source generation",
+            ));
+        }
+        Ok(())
+    }
+
+    fn validate_operator_retry_successor(&self, next: &Self) -> Result<()> {
+        if next.observed_source_generation != self.observed_source_generation
+            || next.attempt_count != self.attempt_count
+            || next.consecutive_failure_count != 0
+            || next.first_failure_at_unix_nanos != self.first_failure_at_unix_nanos
+            || next.last_failure_at_unix_nanos != self.last_failure_at_unix_nanos
+            || next.next_attempt_at_unix_nanos.is_some()
+            || next.last_error_code != self.last_error_code
+        {
+            return Err(IlmRecoveryControlError::InvalidSuccessor(
+                "operator retry changed recovery history or source generation",
+            ));
+        }
+        Ok(())
+    }
+
     fn validate_claim_successor(&self, next: &Self) -> Result<()> {
         if self.classification != IlmRecoveryClassification::Retrying
             || next.classification != IlmRecoveryClassification::Retrying
@@ -827,6 +907,23 @@ pub async fn observe_recovery_source(
     api: Arc<ECStore>,
     canonical_path: &str,
     source_schema: &str,
+) -> EcstoreResult<ObservedIlmRecoverySource> {
+    observe_recovery_source_with_options(api, canonical_path, source_schema, false).await
+}
+
+pub(crate) async fn observe_recovery_source_no_lock(
+    api: Arc<ECStore>,
+    canonical_path: &str,
+    source_schema: &str,
+) -> EcstoreResult<ObservedIlmRecoverySource> {
+    observe_recovery_source_with_options(api, canonical_path, source_schema, true).await
+}
+
+async fn observe_recovery_source_with_options(
+    api: Arc<ECStore>,
+    canonical_path: &str,
+    source_schema: &str,
+    no_lock: bool,
 ) -> EcstoreResult<ObservedIlmRecoverySource> {
     validate_canonical_source_path(canonical_path).map_err(recovery_control_store_error)?;
     if source_schema.trim().is_empty() {
@@ -837,7 +934,16 @@ pub async fn observe_recovery_source(
     let mut observations = Vec::new();
     for set in api.all_set_disks() {
         let authority = format!("pool-{}/set-{}", set.pool_index, set.set_index);
-        match config_boundary::read_config_with_metadata(set, canonical_path, &ObjectOptions::default()).await {
+        match config_boundary::read_config_with_metadata(
+            set,
+            canonical_path,
+            &ObjectOptions {
+                no_lock,
+                ..Default::default()
+            },
+        )
+        .await
+        {
             Ok((data, metadata)) => {
                 let etag = metadata
                     .etag
@@ -1255,6 +1361,99 @@ mod tests {
         ));
     }
 
+    #[test]
+    fn operator_abandonment_is_an_exact_ownerless_retained_successor() {
+        let mut retained = IlmRecoveryControl::new(
+            control().identity,
+            generation(),
+            IlmRecoveryClassification::RetainedAmbiguous,
+            1_000_000_000,
+            IlmRecoveryErrorCode::OperatorDispositionRequired,
+        )
+        .expect("retained control should build");
+        let previous = retained.clone();
+        retained
+            .abandon_for_operator(&previous.observed_source_generation)
+            .expect("exact retained generation should be abandonable");
+        previous
+            .validate_successor(&retained)
+            .expect("operator abandonment should be a valid successor");
+        assert_eq!(retained.classification, IlmRecoveryClassification::Abandoned);
+        assert_eq!(retained.revision, previous.revision + 1);
+
+        let mut wrong_generation = previous.clone();
+        let mut generation = previous.observed_source_generation.clone();
+        generation.source_etag = "different".to_string();
+        assert!(wrong_generation.abandon_for_operator(&generation).is_err());
+
+        let mut mutated_history = retained.clone();
+        mutated_history.attempt_count += 1;
+        assert!(previous.validate_successor(&mutated_history).is_err());
+    }
+
+    #[test]
+    fn operator_retry_rearms_exact_retained_generation_without_resetting_history() {
+        for classification in [
+            IlmRecoveryClassification::RetainedAmbiguous,
+            IlmRecoveryClassification::OperatorRequired,
+        ] {
+            let mut retained = control();
+            retained
+                .claim("node-a", Uuid::new_v4(), 2_000_000_000, 1)
+                .expect("attempt should claim");
+            retained
+                .record_retryable_failure(2_000_000_001, IlmRecoveryErrorCode::BackendTimeout)
+                .expect("failure should persist");
+            retained.classification = classification;
+            retained.next_attempt_at_unix_nanos = None;
+            if classification == IlmRecoveryClassification::OperatorRequired {
+                retained.attempt_count = u64::from(MAX_RECOVERY_ATTEMPTS);
+                retained.consecutive_failure_count = MAX_RECOVERY_ATTEMPTS;
+            }
+            retained.validate().expect("retained control should remain valid");
+
+            let previous = retained.clone();
+            retained
+                .retry_for_operator(&previous.observed_source_generation)
+                .expect("exact retained generation should be retryable");
+            previous
+                .validate_successor(&retained)
+                .expect("operator retry should be a valid successor");
+            assert_eq!(retained.classification, IlmRecoveryClassification::Retrying);
+            assert_eq!(retained.revision, previous.revision + 1);
+            assert_eq!(retained.attempt_count, previous.attempt_count);
+            assert_eq!(retained.first_failure_at_unix_nanos, previous.first_failure_at_unix_nanos);
+            assert_eq!(retained.last_failure_at_unix_nanos, previous.last_failure_at_unix_nanos);
+            assert_eq!(retained.last_error_code, previous.last_error_code);
+            assert_eq!(retained.consecutive_failure_count, 0);
+            assert_eq!(retained.next_attempt_at_unix_nanos, None);
+            assert!(retained.should_attempt_at(2_000_000_002));
+
+            if classification == IlmRecoveryClassification::OperatorRequired {
+                retained
+                    .claim("node-b", Uuid::new_v4(), 2_000_000_002, 1)
+                    .expect("operator retry should authorize one new bounded attempt");
+                retained
+                    .record_retryable_failure(2_000_000_003, IlmRecoveryErrorCode::BackendTimeout)
+                    .expect("the bounded attempt failure should persist");
+                assert_eq!(retained.classification, IlmRecoveryClassification::OperatorRequired);
+                assert_eq!(retained.attempt_count, u64::from(MAX_RECOVERY_ATTEMPTS) + 1);
+                assert_eq!(retained.consecutive_failure_count, 1);
+            }
+
+            let mut wrong_generation = previous;
+            let mut changed_generation = wrong_generation.observed_source_generation.clone();
+            changed_generation.source_etag = "changed".to_string();
+            assert!(wrong_generation.retry_for_operator(&changed_generation).is_err());
+        }
+
+        let mut exhausted = control();
+        exhausted.classification = IlmRecoveryClassification::OperatorRequired;
+        exhausted.attempt_count = u64::MAX;
+        let generation = exhausted.observed_source_generation.clone();
+        assert!(exhausted.retry_for_operator(&generation).is_err());
+    }
+
     #[test]
     fn recovery_control_view_redacts_source_and_owner_details() {
         let mut control = control();

crates/ecstore/src/bucket/lifecycle/recovery_export.rs

@@ -0,0 +1,840 @@
+// Copyright 2024 RustFS Team
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+use std::{collections::HashSet, sync::Arc};
+
+use rustfs_utils::crypto::{hex_sha256, is_sha256_checksum};
+use serde::{Deserialize, Serialize};
+
+use super::config_boundary;
+use super::recovery_control::{
+    IlmRecoveryClassification, IlmRecoveryControl, IlmRecoveryProtocol, IlmRecoverySourceCopy, IlmRecoverySourceGeneration,
+    MAX_ILM_RECOVERY_CONTROL_SIZE, ObservedIlmRecoveryControl, ObservedIlmRecoverySource, recovery_control_record_object_name,
+};
+use super::tier_delete_journal::{
+    TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA, TIER_DELETE_JOURNAL_V2_RECOVERY_SCHEMA, validate_legacy_tier_delete_recovery_source,
+};
+use crate::disk::RUSTFS_META_BUCKET;
+use crate::error::{Error, Result};
+use crate::object_api::{ObjectOptions, WriteCompletion};
+use crate::services::notification_sys::{
+    acquire_ilm_recovery_export_fleet_proof, ilm_recovery_export_fleet_proof_matches, ilm_recovery_export_member_epochs_sha256,
+    ilm_recovery_export_topology_generation,
+};
+use crate::storage_api_contracts::{list::ListOperations as _, namespace::NamespaceLocking as _, object::HTTPPreconditions};
+use crate::store::ECStore;
+
+pub const ILM_RECOVERY_EXPORT_SCHEMA: &str = "rustfs-ilm-recovery-export-v1";
+pub const ILM_RECOVERY_EXPORT_PREFIX: &str = "ilm/recovery-exports";
+pub const MAX_ILM_RECOVERY_EXPORT_SIZE: usize = 128 * 1024;
+const MAX_ILM_RECOVERY_EXPORTS: usize = 10_000;
+const MAX_ILM_RECOVERY_EXPORT_BYTES: u64 = 1024 * 1024 * 1024;
+const MAX_ACTOR_EXPORTS_PER_MINUTE: usize = 10;
+const MAX_CLUSTER_EXPORTS_PER_MINUTE: usize = 100;
+const EXPORT_RETENTION_NANOS: i64 = 90 * 24 * 60 * 60 * 1_000_000_000;
+const EXPORT_ADMISSION_LOCK: &str = "ilm/recovery-admission/export.lock";
+const MAX_LEGACY_TIER_DELETE_SOURCE_SIZE: usize = 64 * 1024;
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub struct IlmRecoveryExportObservation {
+    pub control_id: String,
+    pub protocol: IlmRecoveryProtocol,
+    pub control_etag: String,
+    pub control_revision: u64,
+    pub classification: IlmRecoveryClassification,
+    pub canonical_source_path: String,
+    pub source_generation: IlmRecoverySourceGeneration,
+    pub topology_generation: String,
+    pub member_epochs_sha256: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub struct IlmRecoveryExport {
+    pub export_id: String,
+    pub control_id: String,
+    pub protocol: IlmRecoveryProtocol,
+    pub control_etag: String,
+    pub control_revision: u64,
+    pub classification: IlmRecoveryClassification,
+    pub canonical_source_path: String,
+    pub source_generation: IlmRecoverySourceGeneration,
+    pub topology_generation: String,
+    pub member_epochs_sha256: String,
+    pub creator_sha256: String,
+    pub created_at_unix_nanos: i64,
+    pub retain_until_unix_nanos: i64,
+    pub source_bytes_base64: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+struct PersistedIlmRecoveryExport {
+    schema: String,
+    content_sha256: String,
+    export: IlmRecoveryExport,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
+pub struct IlmRecoveryExportCreated {
+    pub export_id: String,
+    pub content_sha256: String,
+    pub encoded: Vec<u8>,
+    pub replayed: bool,
+}
+
+impl IlmRecoveryExport {
+    fn validate(&self) -> Result<()> {
+        self.source_generation.validate().map_err(Error::other)?;
+        validate_sha256(&self.export_id, "ILM recovery export ID is invalid")?;
+        validate_sha256(&self.control_id, "ILM recovery export control ID is invalid")?;
+        validate_sha256(&self.topology_generation, "ILM recovery export topology generation is invalid")?;
+        validate_sha256(&self.member_epochs_sha256, "ILM recovery export member epoch digest is invalid")?;
+        validate_sha256(&self.creator_sha256, "ILM recovery export creator digest is invalid")?;
+        if self.protocol != IlmRecoveryProtocol::TierDeleteJournal
+            || self.classification != IlmRecoveryClassification::RetainedAmbiguous
+            || !is_legacy_export_schema(&self.source_generation.source_schema)
+        {
+            return Err(Error::other("ILM recovery export source is not an exportable legacy journal"));
+        }
+        if self.control_etag.trim().is_empty() || self.control_revision == 0 {
+            return Err(Error::other("ILM recovery export control generation is invalid"));
+        }
+        if self.canonical_source_path.is_empty()
+            || self.canonical_source_path.starts_with('/')
+            || self.canonical_source_path.ends_with('/')
+            || self.canonical_source_path.split('/').any(str::is_empty)
+        {
+            return Err(Error::other("ILM recovery export source path is invalid"));
+        }
+        if self.created_at_unix_nanos <= 0
+            || self.retain_until_unix_nanos < self.created_at_unix_nanos.saturating_add(EXPORT_RETENTION_NANOS)
+        {
+            return Err(Error::other("ILM recovery export retention is invalid"));
+        }
+        let source = base64_simd::STANDARD
+            .decode_to_vec(self.source_bytes_base64.as_bytes())
+            .map_err(|_| Error::other("ILM recovery export source encoding is invalid"))?;
+        validate_legacy_tier_delete_recovery_source(&self.canonical_source_path, &self.source_generation.source_schema, &source)?;
+        let encoded_len = u64::try_from(source.len()).map_err(|_| Error::other("ILM recovery export source length overflow"))?;
+        if source.is_empty()
+            || source.len() > MAX_LEGACY_TIER_DELETE_SOURCE_SIZE
+            || hex_sha256(&source, ToOwned::to_owned) != self.source_generation.content_sha256
+            || self.source_generation.copies.iter().any(|copy| {
+                copy.canonical_path != self.canonical_source_path
+                    || copy.etag != self.source_generation.source_etag
+                    || copy.content_sha256 != self.source_generation.content_sha256
+                    || copy.encoded_len != encoded_len
+            })
+        {
+            return Err(Error::other("ILM recovery export source bytes do not match the observed generation"));
+        }
+        if recovery_export_id(&self.control_id, &self.source_generation)? != self.export_id {
+            return Err(Error::other("ILM recovery export ID does not match its source generation"));
+        }
+        Ok(())
+    }
+
+    pub fn encode(&self) -> Result<Vec<u8>> {
+        self.validate()?;
+        let export_bytes = serde_json::to_vec(self).map_err(Error::other)?;
+        let persisted = PersistedIlmRecoveryExport {
+            schema: ILM_RECOVERY_EXPORT_SCHEMA.to_string(),
+            content_sha256: hex_sha256(&export_bytes, ToOwned::to_owned),
+            export: self.clone(),
+        };
+        let encoded = serde_json::to_vec(&persisted).map_err(Error::other)?;
+        if encoded.len() > MAX_ILM_RECOVERY_EXPORT_SIZE {
+            return Err(Error::other("encoded ILM recovery export exceeds maximum size"));
+        }
+        Ok(encoded)
+    }
+
+    pub fn decode(expected_export_id: &str, data: &[u8]) -> Result<Self> {
+        validate_sha256(expected_export_id, "ILM recovery export ID is invalid")?;
+        if data.len() > MAX_ILM_RECOVERY_EXPORT_SIZE {
+            return Err(Error::other("encoded ILM recovery export exceeds maximum size"));
+        }
+        let persisted: PersistedIlmRecoveryExport = serde_json::from_slice(data).map_err(Error::other)?;
+        if persisted.schema != ILM_RECOVERY_EXPORT_SCHEMA {
+            return Err(Error::other("ILM recovery export schema is unsupported"));
+        }
+        validate_sha256(&persisted.content_sha256, "ILM recovery export checksum is invalid")?;
+        let export_bytes = serde_json::to_vec(&persisted.export).map_err(Error::other)?;
+        if hex_sha256(&export_bytes, ToOwned::to_owned) != persisted.content_sha256 {
+            return Err(Error::other("ILM recovery export checksum mismatch"));
+        }
+        persisted.export.validate()?;
+        if persisted.export.export_id != expected_export_id {
+            return Err(Error::other("ILM recovery export ID does not match record key"));
+        }
+        Ok(persisted.export)
+    }
+}
+
+pub fn recovery_export_record_object_name(protocol: IlmRecoveryProtocol, export_id: &str) -> Result<String> {
+    validate_sha256(export_id, "ILM recovery export ID is invalid")?;
+    Ok(format!(
+        "{}/{}/{}/{}/{}.json",
+        ILM_RECOVERY_EXPORT_PREFIX,
+        protocol.as_str(),
+        &export_id[..2],
+        &export_id[2..4],
+        export_id
+    ))
+}
+
+pub fn recovery_export_id_from_record_object_name(object: &str) -> Result<(IlmRecoveryProtocol, String)> {
+    let suffix = object
+        .strip_prefix(ILM_RECOVERY_EXPORT_PREFIX)
+        .and_then(|suffix| suffix.strip_prefix('/'))
+        .ok_or_else(|| Error::other("ILM recovery export path has wrong prefix"))?;
+    let mut parts = suffix.split('/');
+    let protocol = match parts.next() {
+        Some("tier_delete_journal") => IlmRecoveryProtocol::TierDeleteJournal,
+        _ => return Err(Error::other("ILM recovery export protocol is invalid")),
+    };
+    let shard_a = parts
+        .next()
+        .ok_or_else(|| Error::other("ILM recovery export path is incomplete"))?;
+    let shard_b = parts
+        .next()
+        .ok_or_else(|| Error::other("ILM recovery export path is incomplete"))?;
+    let export_id = parts
+        .next()
+        .and_then(|name| name.strip_suffix(".json"))
+        .ok_or_else(|| Error::other("ILM recovery export suffix is invalid"))?;
+    if parts.next().is_some() {
+        return Err(Error::other("ILM recovery export path is not canonical"));
+    }
+    validate_sha256(export_id, "ILM recovery export ID is invalid")?;
+    if shard_a != &export_id[..2] || shard_b != &export_id[2..4] {
+        return Err(Error::other("ILM recovery export shard does not match export ID"));
+    }
+    Ok((protocol, export_id.to_string()))
+}
+
+pub async fn inspect_recovery_export_observation(api: Arc<ECStore>, control_id: &str) -> Result<IlmRecoveryExportObservation> {
+    let proof = acquire_ilm_recovery_export_fleet_proof()
+        .await
+        .ok_or_else(|| Error::other("ILM recovery export fleet proof is unavailable"))?;
+    let observed_control = load_exportable_control(api.clone(), control_id).await?;
+    let observed_source = observe_export_source(
+        api,
+        &observed_control.control.identity.canonical_source_path,
+        &observed_control.control.observed_source_generation.source_schema,
+    )
+    .await?;
+    if !observed_source.is_consistent()
+        || observed_source.generation != observed_control.control.observed_source_generation
+        || !ilm_recovery_export_fleet_proof_matches(&proof).await
+    {
+        return Err(Error::other("ILM recovery export observation changed or is incomplete"));
+    }
+    Ok(IlmRecoveryExportObservation {
+        control_id: control_id.to_string(),
+        protocol: observed_control.control.identity.protocol,
+        control_etag: observed_control.etag,
+        control_revision: observed_control.control.revision,
+        classification: observed_control.control.classification,
+        canonical_source_path: observed_control.control.identity.canonical_source_path,
+        source_generation: observed_source.generation,
+        topology_generation: ilm_recovery_export_topology_generation(&proof),
+        member_epochs_sha256: ilm_recovery_export_member_epochs_sha256(&proof),
+    })
+}
+
+pub async fn create_recovery_export(
+    api: Arc<ECStore>,
+    observation: &IlmRecoveryExportObservation,
+    creator_sha256: &str,
+) -> Result<IlmRecoveryExportCreated> {
+    validate_sha256(creator_sha256, "ILM recovery export creator digest is invalid")?;
+    let lock = api.new_ns_lock(RUSTFS_META_BUCKET, EXPORT_ADMISSION_LOCK).await?;
+    let admission_guard = lock.get_write_lock(crate::set_disk::get_lock_acquire_timeout()).await?;
+
+    let proof = acquire_ilm_recovery_export_fleet_proof()
+        .await
+        .ok_or_else(|| Error::other("ILM recovery export fleet proof is unavailable"))?;
+    if ilm_recovery_export_topology_generation(&proof) != observation.topology_generation
+        || ilm_recovery_export_member_epochs_sha256(&proof) != observation.member_epochs_sha256
+    {
+        return Err(Error::PreconditionFailed);
+    }
+    let control_object = recovery_control_record_object_name(IlmRecoveryProtocol::TierDeleteJournal, &observation.control_id)
+        .map_err(Error::other)?;
+    let control_lock = api.new_ns_lock(RUSTFS_META_BUCKET, &control_object).await?;
+    let control_guard = control_lock
+        .get_read_lock(crate::set_disk::get_lock_acquire_timeout())
+        .await?;
+    let source_lock = api
+        .new_ns_lock(RUSTFS_META_BUCKET, &observation.canonical_source_path)
+        .await?;
+    let source_guard = source_lock.get_read_lock(crate::set_disk::get_lock_acquire_timeout()).await?;
+    let locks_current = || !admission_guard.is_lock_lost() && !control_guard.is_lock_lost() && !source_guard.is_lock_lost();
+    let (current, current_source_bytes) = current_observation_under_proof_no_lock(api.clone(), observation, &proof).await?;
+    if &current != observation || !locks_current() {
+        return Err(Error::PreconditionFailed);
+    }
+    let current_source_base64 = base64_simd::STANDARD.encode_to_string(current_source_bytes);
+    let candidate_export_id = recovery_export_id(&current.control_id, &current.source_generation)?;
+    let object = recovery_export_record_object_name(current.protocol, &candidate_export_id)?;
+    match load_recovery_export_decoded(api.clone(), &candidate_export_id).await {
+        Ok((existing, export)) if export_matches_observation(&export, observation) => {
+            if !locks_current() || !ilm_recovery_export_fleet_proof_matches(&proof).await {
+                return Err(Error::PreconditionFailed);
+            }
+            api.record_durable_ilm_decommission_progress(&object, &existing.encoded)
+                .await?;
+            if !locks_current() {
+                return Err(Error::PreconditionFailed);
+            }
+            return Ok(existing.with_replayed());
+        }
+        Ok(_) => return Err(Error::PreconditionFailed),
+        Err(Error::ConfigNotFound) => {}
+        Err(err) => return Err(err),
+    }
+    let inventory = collect_export_inventory(api.clone()).await?;
+    if !locks_current() || !ilm_recovery_export_fleet_proof_matches(&proof).await {
+        return Err(Error::PreconditionFailed);
+    }
+    let created_at_unix_nanos = now_unix_nanos()?;
+    let export = build_export_from_source(&current, creator_sha256, created_at_unix_nanos, &current_source_base64)?;
+    let encoded = export.encode()?;
+    inventory.check(creator_sha256, encoded.len(), created_at_unix_nanos)?;
+
+    let mut write_options = ObjectOptions {
+        max_parity: true,
+        write_completion: WriteCompletion::TailDrained,
+        http_preconditions: Some(HTTPPreconditions {
+            if_none_match: Some("*".to_string()),
+            ..Default::default()
+        }),
+        ..Default::default()
+    };
+    write_options.add_namespace_lock_guard(&admission_guard);
+    write_options.add_namespace_lock_guard(&control_guard);
+    write_options.add_namespace_lock_guard(&source_guard);
+    if !locks_current() {
+        return Err(Error::PreconditionFailed);
+    }
+    let write_result = config_boundary::save_config_with_opts(api.clone(), &object, encoded.clone(), &write_options).await;
+    let stored = match load_recovery_export(api.clone(), &export.export_id).await {
+        Ok(stored) if stored.encoded == encoded => stored,
+        Ok(_) => return Err(Error::PreconditionFailed),
+        Err(read_err) => return Err(write_result.err().unwrap_or(read_err)),
+    };
+    if !locks_current() || !ilm_recovery_export_fleet_proof_matches(&proof).await {
+        return Err(Error::PreconditionFailed);
+    }
+    api.record_durable_ilm_decommission_progress(&object, &encoded).await?;
+    if !locks_current() {
+        return Err(Error::PreconditionFailed);
+    }
+    Ok(stored)
+}
+
+pub async fn load_recovery_export(api: Arc<ECStore>, export_id: &str) -> Result<IlmRecoveryExportCreated> {
+    let (created, _) = load_recovery_export_decoded(api, export_id).await?;
+    Ok(created)
+}
+
+async fn load_recovery_export_decoded(
+    api: Arc<ECStore>,
+    export_id: &str,
+) -> Result<(IlmRecoveryExportCreated, IlmRecoveryExport)> {
+    let object = recovery_export_record_object_name(IlmRecoveryProtocol::TierDeleteJournal, export_id)?;
+    let encoded = config_boundary::read_config_limited_preserve_empty(api, &object, MAX_ILM_RECOVERY_EXPORT_SIZE).await?;
+    let export = IlmRecoveryExport::decode(export_id, &encoded)?;
+    let content_sha256 = hex_sha256(&encoded, ToOwned::to_owned);
+    Ok((
+        IlmRecoveryExportCreated {
+            export_id: export.export_id.clone(),
+            content_sha256,
+            encoded,
+            replayed: false,
+        },
+        export,
+    ))
+}
+
+impl IlmRecoveryExportCreated {
+    fn with_replayed(mut self) -> Self {
+        self.replayed = true;
+        self
+    }
+}
+
+async fn load_exportable_control(api: Arc<ECStore>, control_id: &str) -> Result<ObservedIlmRecoveryControl> {
+    load_exportable_control_with_options(api, control_id, &ObjectOptions::default()).await
+}
+
+async fn load_exportable_control_no_lock(api: Arc<ECStore>, control_id: &str) -> Result<ObservedIlmRecoveryControl> {
+    load_exportable_control_with_options(
+        api,
+        control_id,
+        &ObjectOptions {
+            no_lock: true,
+            ..Default::default()
+        },
+    )
+    .await
+}
+
+async fn load_exportable_control_with_options(
+    api: Arc<ECStore>,
+    control_id: &str,
+    options: &ObjectOptions,
+) -> Result<ObservedIlmRecoveryControl> {
+    let object = recovery_control_record_object_name(IlmRecoveryProtocol::TierDeleteJournal, control_id).map_err(Error::other)?;
+    let (data, metadata) =
+        config_boundary::read_config_limited_preserve_empty_with_metadata(api, &object, options, MAX_ILM_RECOVERY_CONTROL_SIZE)
+            .await?;
+    let etag = metadata
+        .etag
+        .filter(|etag| !etag.trim().is_empty())
+        .ok_or_else(|| Error::other("ILM recovery control is missing an ETag"))?;
+    let control = IlmRecoveryControl::decode(control_id, &data).map_err(Error::other)?;
+    if control.identity.protocol != IlmRecoveryProtocol::TierDeleteJournal
+        || control.classification != IlmRecoveryClassification::RetainedAmbiguous
+        || !is_legacy_export_schema(&control.observed_source_generation.source_schema)
+    {
+        return Err(Error::other("ILM recovery control is not exportable"));
+    }
+    Ok(ObservedIlmRecoveryControl { control, etag })
+}
+
+async fn current_observation_under_proof_no_lock(
+    api: Arc<ECStore>,
+    expected: &IlmRecoveryExportObservation,
+    proof: &crate::services::notification_sys::IlmRecoveryExportFleetProofToken,
+) -> Result<(IlmRecoveryExportObservation, Vec<u8>)> {
+    let observed_control = load_exportable_control_no_lock(api.clone(), &expected.control_id).await?;
+    let observed_source = observe_export_source_no_lock(
+        api,
+        &observed_control.control.identity.canonical_source_path,
+        &observed_control.control.observed_source_generation.source_schema,
+    )
+    .await?;
+    let source_bytes = observed_source
+        .canonical_data
+        .clone()
+        .ok_or_else(|| Error::other("ILM recovery export source copies diverge"))?;
+    if !observed_source.is_consistent()
+        || observed_source.generation != observed_control.control.observed_source_generation
+        || !ilm_recovery_export_fleet_proof_matches(proof).await
+    {
+        return Err(Error::PreconditionFailed);
+    }
+    Ok((
+        IlmRecoveryExportObservation {
+            control_id: expected.control_id.clone(),
+            protocol: observed_control.control.identity.protocol,
+            control_etag: observed_control.etag,
+            control_revision: observed_control.control.revision,
+            classification: observed_control.control.classification,
+            canonical_source_path: observed_control.control.identity.canonical_source_path,
+            source_generation: observed_source.generation,
+            topology_generation: ilm_recovery_export_topology_generation(proof),
+            member_epochs_sha256: ilm_recovery_export_member_epochs_sha256(proof),
+        },
+        source_bytes,
+    ))
+}
+
+async fn observe_export_source(
+    api: Arc<ECStore>,
+    canonical_path: &str,
+    source_schema: &str,
+) -> Result<ObservedIlmRecoverySource> {
+    if canonical_path.is_empty()
+        || canonical_path.starts_with('/')
+        || canonical_path.ends_with('/')
+        || canonical_path.split('/').any(str::is_empty)
+        || !is_legacy_export_schema(source_schema)
+    {
+        return Err(Error::other("ILM recovery export source identity is invalid"));
+    }
+    let lock = api.new_ns_lock(RUSTFS_META_BUCKET, canonical_path).await?;
+    let _guard = lock.get_read_lock(crate::set_disk::get_lock_acquire_timeout()).await?;
+    observe_export_source_no_lock(api, canonical_path, source_schema).await
+}
+
+async fn observe_export_source_no_lock(
+    api: Arc<ECStore>,
+    canonical_path: &str,
+    source_schema: &str,
+) -> Result<ObservedIlmRecoverySource> {
+    let mut copies = Vec::new();
+    let mut canonical: Option<(String, String, Vec<u8>)> = None;
+    let mut consistent = true;
+    for set in api.all_set_disks() {
+        let authority = format!("pool-{}/set-{}", set.pool_index, set.set_index);
+        let result = config_boundary::read_config_limited_preserve_empty_with_metadata(
+            set,
+            canonical_path,
+            &ObjectOptions {
+                no_lock: true,
+                ..Default::default()
+            },
+            MAX_LEGACY_TIER_DELETE_SOURCE_SIZE,
+        )
+        .await;
+        match result {
+            Ok((data, metadata)) => {
+                if data.is_empty() || data.len() > MAX_LEGACY_TIER_DELETE_SOURCE_SIZE {
+                    return Err(Error::other("ILM recovery export source exceeds its protocol size limit"));
+                }
+                validate_legacy_tier_delete_recovery_source(canonical_path, source_schema, &data)?;
+                let etag = metadata
+                    .etag
+                    .filter(|etag| !etag.trim().is_empty())
+                    .ok_or_else(|| Error::other("ILM recovery export source copy is missing an ETag"))?;
+                let content_sha256 = hex_sha256(&data, ToOwned::to_owned);
+                let encoded_len =
+                    u64::try_from(data.len()).map_err(|_| Error::other("ILM recovery export source length does not fit u64"))?;
+                copies.push(IlmRecoverySourceCopy {
+                    authority,
+                    canonical_path: canonical_path.to_string(),
+                    etag: etag.clone(),
+                    encoded_len,
+                    content_sha256: content_sha256.clone(),
+                });
+                match canonical.as_ref() {
+                    Some((first_etag, first_digest, first_data)) => {
+                        consistent &= first_etag == &etag && first_digest == &content_sha256 && first_data == &data;
+                    }
+                    None => canonical = Some((etag, content_sha256, data)),
+                }
+            }
+            Err(err) if export_source_is_missing(&err) => {}
+            Err(err) => return Err(err),
+        }
+    }
+    let Some((source_etag, content_sha256, source_bytes)) = canonical else {
+        return Err(Error::ConfigNotFound);
+    };
+    let generation =
+        IlmRecoverySourceGeneration::new(source_schema, source_etag, content_sha256, copies).map_err(Error::other)?;
+    Ok(ObservedIlmRecoverySource {
+        generation,
+        canonical_data: consistent.then_some(source_bytes),
+    })
+}
+
+fn export_source_is_missing(err: &Error) -> bool {
+    matches!(
+        err,
+        Error::ConfigNotFound | Error::FileNotFound | Error::ObjectNotFound(_, _) | Error::VersionNotFound(_, _, _)
+    )
+}
+
+fn build_export_from_source(
+    observation: &IlmRecoveryExportObservation,
+    creator_sha256: &str,
+    created_at_unix_nanos: i64,
+    source_bytes_base64: &str,
+) -> Result<IlmRecoveryExport> {
+    let retain_until_unix_nanos = created_at_unix_nanos
+        .checked_add(EXPORT_RETENTION_NANOS)
+        .ok_or_else(|| Error::other("ILM recovery export retention timestamp overflow"))?;
+    let export = IlmRecoveryExport {
+        export_id: recovery_export_id(&observation.control_id, &observation.source_generation)?,
+        control_id: observation.control_id.clone(),
+        protocol: observation.protocol,
+        control_etag: observation.control_etag.clone(),
+        control_revision: observation.control_revision,
+        classification: observation.classification,
+        canonical_source_path: observation.canonical_source_path.clone(),
+        source_generation: observation.source_generation.clone(),
+        topology_generation: observation.topology_generation.clone(),
+        member_epochs_sha256: observation.member_epochs_sha256.clone(),
+        creator_sha256: creator_sha256.to_string(),
+        created_at_unix_nanos,
+        retain_until_unix_nanos,
+        source_bytes_base64: source_bytes_base64.to_string(),
+    };
+    export.validate()?;
+    Ok(export)
+}
+
+pub(crate) fn recovery_export_id(control_id: &str, generation: &IlmRecoverySourceGeneration) -> Result<String> {
+    validate_sha256(control_id, "ILM recovery export control ID is invalid")?;
+    validate_sha256(&generation.content_sha256, "ILM recovery export source checksum is invalid")?;
+    validate_sha256(&generation.copy_set_sha256, "ILM recovery export copy-set checksum is invalid")?;
+    let mut data = Vec::new();
+    for part in [control_id, &generation.content_sha256, &generation.copy_set_sha256] {
+        data.extend_from_slice(&(part.len() as u64).to_be_bytes());
+        data.extend_from_slice(part.as_bytes());
+    }
+    Ok(hex_sha256(&data, ToOwned::to_owned))
+}
+
+fn export_matches_observation(export: &IlmRecoveryExport, observation: &IlmRecoveryExportObservation) -> bool {
+    export.control_id == observation.control_id
+        && export.protocol == observation.protocol
+        && export.classification == observation.classification
+        && export.canonical_source_path == observation.canonical_source_path
+        && export.source_generation == observation.source_generation
+}
+
+#[derive(Debug, Default)]
+struct IlmRecoveryExportInventory {
+    count: usize,
+    bytes: u64,
+    creations: Vec<(i64, String)>,
+}
+
+impl IlmRecoveryExportInventory {
+    fn check(&self, creator_sha256: &str, candidate_len: usize, now: i64) -> Result<()> {
+        let recent_after = now.saturating_sub(60 * 1_000_000_000);
+        let cluster_recent = self
+            .creations
+            .iter()
+            .filter(|(created_at, _)| *created_at > recent_after)
+            .count();
+        let actor_recent = self
+            .creations
+            .iter()
+            .filter(|(created_at, creator)| *created_at > recent_after && creator == creator_sha256)
+            .count();
+        check_export_admission(self.count, self.bytes, actor_recent, cluster_recent, candidate_len)
+    }
+}
+
+async fn collect_export_inventory(api: Arc<ECStore>) -> Result<IlmRecoveryExportInventory> {
+    let mut marker = None;
+    let mut seen_markers = HashSet::new();
+    let mut inventory = IlmRecoveryExportInventory::default();
+    loop {
+        let page = api
+            .clone()
+            .list_objects_v2(
+                RUSTFS_META_BUCKET,
+                &format!("{ILM_RECOVERY_EXPORT_PREFIX}/"),
+                marker.clone(),
+                None,
+                1_000,
+                false,
+                None,
+                false,
+            )
+            .await?;
+        for object in page.objects {
+            let (_, export_id) = recovery_export_id_from_record_object_name(&object.name)?;
+            let (stored, export) = load_recovery_export_decoded(api.clone(), &export_id).await?;
+            inventory.count = inventory
+                .count
+                .checked_add(1)
+                .ok_or_else(|| Error::other("ILM recovery export count overflow"))?;
+            inventory.bytes = inventory
+                .bytes
+                .checked_add(u64::try_from(stored.encoded.len()).map_err(|_| Error::other("ILM recovery export size overflow"))?)
+                .ok_or_else(|| Error::other("ILM recovery export byte total overflow"))?;
+            inventory
+                .creations
+                .push((export.created_at_unix_nanos, export.creator_sha256));
+        }
+        if !page.is_truncated {
+            break;
+        }
+        let next = page
+            .next_continuation_token
+            .ok_or_else(|| Error::other("ILM recovery export inventory omitted its continuation marker"))?;
+        marker = Some(record_export_inventory_marker(&mut seen_markers, next)?);
+    }
+    Ok(inventory)
+}
+
+fn record_export_inventory_marker(seen_markers: &mut HashSet<String>, next: String) -> Result<String> {
+    if !seen_markers.insert(next.clone()) {
+        return Err(Error::other("ILM recovery export inventory repeated its continuation marker"));
+    }
+    Ok(next)
+}
+
+fn check_export_admission(
+    count: usize,
+    bytes: u64,
+    actor_recent: usize,
+    cluster_recent: usize,
+    candidate_len: usize,
+) -> Result<()> {
+    let candidate_len = u64::try_from(candidate_len).map_err(|_| Error::other("ILM recovery export size does not fit u64"))?;
+    if count >= MAX_ILM_RECOVERY_EXPORTS
+        || bytes
+            .checked_add(candidate_len)
+            .is_none_or(|total| total > MAX_ILM_RECOVERY_EXPORT_BYTES)
+        || actor_recent >= MAX_ACTOR_EXPORTS_PER_MINUTE
+        || cluster_recent >= MAX_CLUSTER_EXPORTS_PER_MINUTE
+    {
+        return Err(Error::SlowDown);
+    }
+    Ok(())
+}
+
+fn is_legacy_export_schema(schema: &str) -> bool {
+    matches!(schema, TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA | TIER_DELETE_JOURNAL_V2_RECOVERY_SCHEMA)
+}
+
+fn validate_sha256(value: &str, message: &'static str) -> Result<()> {
+    if !is_sha256_checksum(value)
+        || value
+            .bytes()
+            .any(|byte| byte.is_ascii_hexdigit() && byte.is_ascii_uppercase())
+    {
+        return Err(Error::other(message));
+    }
+    Ok(())
+}
+
+fn now_unix_nanos() -> Result<i64> {
+    i64::try_from(time::OffsetDateTime::now_utc().unix_timestamp_nanos())
+        .map_err(|_| Error::other("ILM recovery export timestamp does not fit i64"))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::bucket::lifecycle::recovery_control::IlmRecoverySourceCopy;
+
+    const PINNED_V1_EXPORT: &[u8] = br#"{"schema":"rustfs-ilm-recovery-export-v1","content_sha256":"3dfb3ec3892256e909de1211c1a963ca7008963ff32b3a869f7161a7b9b44028","export":{"export_id":"2b78e7a825bfc2edbf7f773d0b6ed3bf93e360ff1702d73a449109c11bfaa105","control_id":"0fcd568a5cb9bdb4677b69354b11ee415af8f784519cff3da49a26f84eaee7f2","protocol":"tier_delete_journal","control_etag":"control-etag","control_revision":1,"classification":"retained_ambiguous","canonical_source_path":"ilm/tier-delete-journal/872072554f66ab326f10ce7adbae11422b7a4b0663aa7112d6061a8f6ed41b94.json","source_generation":{"source_schema":"rustfs-tier-delete-journal-v1","source_etag":"etag-a","content_sha256":"0e0b010ebdeeb7b41473fe8575e989d6bb1303c0ca551dd984e9400f0ae306bd","copy_set_sha256":"5a7406115b6c3923ffe79dcd1f43ccae7beed786e557163f019dd10ec409a653","copies":[{"authority":"pool-0/set-0","canonical_path":"ilm/tier-delete-journal/872072554f66ab326f10ce7adbae11422b7a4b0663aa7112d6061a8f6ed41b94.json","etag":"etag-a","encoded_len":81,"content_sha256":"0e0b010ebdeeb7b41473fe8575e989d6bb1303c0ca551dd984e9400f0ae306bd"}]},"topology_generation":"e6e2b826e31fca5c36125c48f130dcb6f961e698ff8a8776a1f290cf0892e8e6","member_epochs_sha256":"612dd8a861161819a4ad8f6f3e2a0567602877c043a2353ca933a13c78dc0ed4","creator_sha256":"50c9c4aeb40b5b206b6d98f516f8b8c0efd29ce2e56a76b345fb9240c225a1b7","created_at_unix_nanos":1000000000,"retain_until_unix_nanos":7776001000000000,"source_bytes_base64":"eyJ2ZXJzaW9uIjoxLCJvYmpfbmFtZSI6ImxlZ2FjeS9yZW1vdGUiLCJ2ZXJzaW9uX2lkIjoib3BhcXVlIiwidGllcl9uYW1lIjoiV0FSTSJ9"}}"#;
+
+    fn legacy_source() -> Vec<u8> {
+        br#"{"version":1,"obj_name":"legacy/remote","version_id":"opaque","tier_name":"WARM"}"#.to_vec()
+    }
+
+    fn observation() -> IlmRecoveryExportObservation {
+        let source = legacy_source();
+        let source_path = super::super::tier_delete_journal::tier_delete_journal_object_name(
+            &super::super::tier_delete_journal::decode_tier_delete_journal_entry(&source).expect("legacy fixture should decode"),
+        );
+        let source_sha256 = hex_sha256(&source, ToOwned::to_owned);
+        let generation = IlmRecoverySourceGeneration::new(
+            TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA,
+            "etag-a",
+            source_sha256.clone(),
+            vec![IlmRecoverySourceCopy {
+                authority: "pool-0/set-0".to_string(),
+                canonical_path: source_path.clone(),
+                etag: "etag-a".to_string(),
+                encoded_len: source.len() as u64,
+                content_sha256: source_sha256,
+            }],
+        )
+        .expect("generation should be valid");
+        IlmRecoveryExportObservation {
+            control_id: hex_sha256(b"control", ToOwned::to_owned),
+            protocol: IlmRecoveryProtocol::TierDeleteJournal,
+            control_etag: "control-etag".to_string(),
+            control_revision: 1,
+            classification: IlmRecoveryClassification::RetainedAmbiguous,
+            canonical_source_path: source_path,
+            source_generation: generation,
+            topology_generation: hex_sha256(b"topology", ToOwned::to_owned),
+            member_epochs_sha256: hex_sha256(b"epochs", ToOwned::to_owned),
+        }
+    }
+
+    #[test]
+    fn recovery_export_round_trip_is_strict_and_deterministic() {
+        let observed = observation();
+        let creator = hex_sha256(b"actor", ToOwned::to_owned);
+        let export = build_export_from_source(
+            &observed,
+            &creator,
+            1_000_000_000,
+            &base64_simd::STANDARD.encode_to_string(legacy_source()),
+        )
+        .expect("export should be valid");
+        assert_eq!(
+            export.export_id,
+            recovery_export_id(&observed.control_id, &observed.source_generation).unwrap()
+        );
+        let encoded = export.encode().expect("export should encode");
+        assert_eq!(encoded, PINNED_V1_EXPORT, "v1 export wire format must remain pinned");
+        assert_eq!(IlmRecoveryExport::decode(&export.export_id, &encoded).unwrap(), export);
+        assert_eq!(
+            IlmRecoveryExport::decode("2b78e7a825bfc2edbf7f773d0b6ed3bf93e360ff1702d73a449109c11bfaa105", PINNED_V1_EXPORT)
+                .unwrap(),
+            export,
+        );
+
+        let path = recovery_export_record_object_name(export.protocol, &export.export_id).unwrap();
+        let durable = super::super::durable_namespace::validate_durable_ilm_record(&path, &encoded)
+            .expect("export should be registered as a durable ILM record");
+        assert_eq!(durable.namespace, "recovery-export");
+        assert_eq!(durable.id_kind, "export_id");
+        assert_eq!(durable.id, export.export_id);
+
+        let mut wrong_source = export.clone();
+        wrong_source.source_bytes_base64 = base64_simd::STANDARD.encode_to_string(b"changed");
+        assert!(wrong_source.encode().is_err());
+
+        let mut persisted: serde_json::Value = serde_json::from_slice(&encoded).unwrap();
+        persisted["unknown"] = serde_json::json!(true);
+        assert!(IlmRecoveryExport::decode(&export.export_id, &serde_json::to_vec(&persisted).unwrap()).is_err());
+    }
+
+    #[test]
+    fn export_inventory_rejects_non_adjacent_continuation_cycles() {
+        let mut seen = HashSet::new();
+        assert_eq!(record_export_inventory_marker(&mut seen, "a".to_string()).unwrap(), "a");
+        assert_eq!(record_export_inventory_marker(&mut seen, "b".to_string()).unwrap(), "b");
+        record_export_inventory_marker(&mut seen, "a".to_string())
+            .expect_err("a non-adjacent continuation marker cycle must fail closed");
+    }
+
+    #[test]
+    fn recovery_export_path_rejects_noncanonical_shards() {
+        let id = hex_sha256(b"export", ToOwned::to_owned);
+        let path = recovery_export_record_object_name(IlmRecoveryProtocol::TierDeleteJournal, &id).unwrap();
+        assert_eq!(recovery_export_id_from_record_object_name(&path).unwrap().1, id);
+        let wrong_shard = path.replacen(&format!("/{}/", &id[..2]), "/zz/", 1);
+        assert!(recovery_export_id_from_record_object_name(&wrong_shard).is_err());
+    }
+
+    #[test]
+    fn canonical_replay_survives_fleet_rotation_but_not_source_change() {
+        let observed = observation();
+        let creator = hex_sha256(b"actor", ToOwned::to_owned);
+        let export = build_export_from_source(
+            &observed,
+            &creator,
+            1_000_000_000,
+            &base64_simd::STANDARD.encode_to_string(legacy_source()),
+        )
+        .unwrap();
+        let mut rotated = observed;
+        rotated.control_etag = "new-control-etag".to_string();
+        rotated.control_revision += 1;
+        rotated.topology_generation = hex_sha256(b"new-topology", ToOwned::to_owned);
+        rotated.member_epochs_sha256 = hex_sha256(b"new-members", ToOwned::to_owned);
+        assert!(export_matches_observation(&export, &rotated));
+
+        rotated.source_generation.content_sha256 = hex_sha256(b"changed", ToOwned::to_owned);
+        assert!(!export_matches_observation(&export, &rotated));
+    }
+
+    #[test]
+    fn export_admission_enforces_exact_count_byte_and_rate_boundaries() {
+        assert!(check_export_admission(9_999, MAX_ILM_RECOVERY_EXPORT_BYTES - 1, 9, 99, 1).is_ok());
+        assert!(check_export_admission(10_000, 0, 0, 0, 1).is_err());
+        assert!(check_export_admission(0, MAX_ILM_RECOVERY_EXPORT_BYTES, 0, 0, 1).is_err());
+        assert!(check_export_admission(0, 0, 10, 0, 1).is_err());
+        assert!(check_export_admission(0, 0, 0, 100, 1).is_err());
+    }
+}

crates/ecstore/src/bucket/lifecycle/tier_delete_journal.rs

@@ -82,8 +82,8 @@ const TIER_DELETE_DISPATCH_MEMBER_DELETE_CONCURRENCY: usize = 32;
 const TIER_DELETE_DISPATCH_PREPARE_CONCURRENCY: usize = 16;
 const TIER_DELETE_DISPATCH_CAS_CONCURRENCY: usize = 32;
 const TIER_DELETE_JOURNAL_VERSION: u8 = 2;
-const TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA: &str = "rustfs-tier-delete-journal-v1";
+pub(crate) const TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA: &str = "rustfs-tier-delete-journal-v1";
-const TIER_DELETE_JOURNAL_V2_RECOVERY_SCHEMA: &str = "rustfs-tier-delete-journal-v2";
+pub(crate) const TIER_DELETE_JOURNAL_V2_RECOVERY_SCHEMA: &str = "rustfs-tier-delete-journal-v2";
 const TIER_DELETE_JOURNAL_UNKNOWN_RECOVERY_SCHEMA: &str = "rustfs-tier-delete-journal-unknown";
 const TIER_DELETE_JOURNAL_V1_RECOVERY_CLASS: &str = "tier_delete_journal_v1";
 const TIER_DELETE_JOURNAL_V2_RECOVERY_CLASS: &str = "tier_delete_journal_v2";
@@ -884,6 +884,23 @@ struct PersistedTierDeleteJournalEntry {
 }
 
 impl PersistedTierDeleteJournalEntry {
+    fn validate_legacy_recovery_shape(&self) -> Result<()> {
+        let has_later_version_fields = self.version_id_exact.is_some()
+            || self.version_state.is_some()
+            || self.state.is_some()
+            || self.source.is_some()
+            || self.dispatch.is_some();
+        match self.version {
+            1 if self.backend_identity.is_none() && !has_later_version_fields => Ok(()),
+            TIER_DELETE_JOURNAL_VERSION if self.backend_identity.is_some() && !has_later_version_fields => Ok(()),
+            1 => Err(Error::other("tier delete journal v1 entry contains fields from a later version")),
+            TIER_DELETE_JOURNAL_VERSION => Err(Error::other(
+                "tier delete journal v2 entry is missing its identity or contains fields from a later version",
+            )),
+            _ => Err(Error::other("tier delete journal is not an exportable legacy version")),
+        }
+    }
+
     fn from_jentry(je: &Jentry) -> Result<Self> {
         validate_version_state(je.version_state, &je.version_id, je.version_id_exact)?;
         let legacy_unknown = je.version_state == rustfs_filemeta::TransitionVersionState::Unknown;
@@ -5531,6 +5548,12 @@ fn canonical_legacy_tier_delete_journal_identity(object_name: &str) -> Option<&s
     .then_some(identity)
 }
 
+pub(crate) fn validate_legacy_tier_delete_recovery_path(object_name: &str) -> Result<()> {
+    canonical_legacy_tier_delete_journal_identity(object_name)
+        .map(|_| ())
+        .ok_or_else(|| Error::other("legacy tier delete journal path is not canonical"))
+}
+
 fn legacy_tier_delete_recovery_descriptor(entry: &Jentry) -> Option<(&'static str, &'static str)> {
     match entry.persisted_version {
         1 => Some((TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA, TIER_DELETE_JOURNAL_V1_RECOVERY_CLASS)),
@@ -5539,6 +5562,21 @@ fn legacy_tier_delete_recovery_descriptor(entry: &Jentry) -> Option<(&'static st
     }
 }
 
+pub(crate) fn validate_legacy_tier_delete_recovery_source(object_name: &str, source_schema: &str, data: &[u8]) -> Result<()> {
+    validate_legacy_tier_delete_recovery_path(object_name)?;
+    let persisted: PersistedTierDeleteJournalEntry =
+        serde_json::from_slice(data).map_err(|err| Error::other_with_context("decode tier delete journal failed", err))?;
+    persisted.validate_legacy_recovery_shape()?;
+    let entry = persisted.into_jentry()?;
+    let Some((decoded_schema, _)) = legacy_tier_delete_recovery_descriptor(&entry) else {
+        return Err(Error::other("tier delete journal is not an exportable legacy version"));
+    };
+    if decoded_schema != source_schema || tier_delete_journal_object_name(&entry) != object_name {
+        return Err(Error::other("legacy tier delete journal identity does not match its recovery source"));
+    }
+    Ok(())
+}
+
 fn legacy_tier_delete_control_matches(
     control: &IlmRecoveryControl,
     identity: &IlmRecoveryControlIdentity,
@@ -6140,17 +6178,18 @@ where
 #[cfg(test)]
 mod tests {
     use super::{
-        TIER_DELETE_DISPATCH_MANIFEST_VERSION, TIER_DELETE_DISPATCH_PARENT_RECORD_TYPE, TIER_DELETE_DISPATCH_PARENT_VERSION,
+        PersistedTierDeleteJournalEntry, TIER_DELETE_DISPATCH_MANIFEST_VERSION, TIER_DELETE_DISPATCH_PARENT_RECORD_TYPE,
-        TIER_DELETE_JOURNAL_EXACT_VERSION, TIER_DELETE_JOURNAL_LEGACY_PREFIX, TIER_DELETE_JOURNAL_SOLE_OWNER_VERSION,
+        TIER_DELETE_DISPATCH_PARENT_VERSION, TIER_DELETE_JOURNAL_EXACT_VERSION, TIER_DELETE_JOURNAL_LEGACY_PREFIX,
-        TIER_DELETE_JOURNAL_STATE_VERSION, TIER_DELETE_JOURNAL_TRANSACTION_VERSION, TIER_DELETE_JOURNAL_V6_PREFIX,
+        TIER_DELETE_JOURNAL_SOLE_OWNER_VERSION, TIER_DELETE_JOURNAL_STATE_VERSION, TIER_DELETE_JOURNAL_TRANSACTION_VERSION,
-        TierDeleteDispatchChunkBinding, TierDeleteDispatchManifest, TierDeleteDispatchManifestState, TierDeleteDispatchParent,
+        TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA, TIER_DELETE_JOURNAL_V2_RECOVERY_SCHEMA, TIER_DELETE_JOURNAL_V6_PREFIX,
-        TierDeleteDispatchParentState, TierDeleteDispatchRecord, await_tier_delete_journal_recovery,
+        TIER_DELETE_JOURNAL_VERSION, TierDeleteDispatchChunkBinding, TierDeleteDispatchManifest, TierDeleteDispatchManifestState,
+        TierDeleteDispatchParent, TierDeleteDispatchParentState, TierDeleteDispatchRecord, await_tier_delete_journal_recovery,
         decode_tier_delete_dispatch_record, decode_tier_delete_journal_entry, encode_tier_delete_dispatch_manifest,
         encode_tier_delete_dispatch_parent, encode_tier_delete_journal_entry, object_info_references_tier_delete,
         record_tier_delete_journal_backend_identity, same_tier_delete_authorization_identity, same_tier_delete_journal_identity,
         tier_delete_dispatch_child_matches_parent, tier_delete_dispatch_chunk_manifest_object_name,
         tier_delete_dispatch_journal_set_digest, tier_delete_dispatch_manifest_object_name, tier_delete_journal_object_name,
-        tier_delete_source_matches_dispatch_scope,
+        tier_delete_source_matches_dispatch_scope, validate_legacy_tier_delete_recovery_source,
     };
     use crate::bucket::lifecycle::tier_sweeper::{
         Jentry, TierDeleteDispatchBinding, TierDeleteJournalState, TierDeleteSourceIdentity,
@@ -6609,6 +6648,72 @@ mod tests {
         }
     }
 
+    #[test]
+    fn legacy_recovery_export_rejects_fields_from_later_journal_versions() {
+        let later = bound_v6_journal_entry(TierDeleteJournalState::Prepared);
+        let v1 = PersistedTierDeleteJournalEntry {
+            version: 1,
+            obj_name: "remote/object".to_string(),
+            version_id: "opaque".to_string(),
+            tier_name: "WARM".to_string(),
+            backend_identity: None,
+            version_id_exact: None,
+            version_state: None,
+            state: None,
+            source: None,
+            dispatch: None,
+        };
+        let mut v2 = v1.clone();
+        v2.version = TIER_DELETE_JOURNAL_VERSION;
+        v2.backend_identity = Some([7; 32]);
+
+        let assert_rejected = |persisted: PersistedTierDeleteJournalEntry, schema: &str| {
+            let normalized = persisted
+                .clone()
+                .into_jentry()
+                .expect("the generic compatibility decoder should demonstrate the discarded field");
+            let object_name = tier_delete_journal_object_name(&normalized);
+            let encoded = serde_json::to_vec(&persisted).expect("mixed-version journal fixture should encode");
+            let err = validate_legacy_tier_delete_recovery_source(&object_name, schema, &encoded)
+                .expect_err("legacy recovery export must reject fields from later versions");
+            assert!(err.to_string().contains("later version"));
+        };
+
+        let mut invalid_v1 = Vec::new();
+        let mut with_backend = v1.clone();
+        with_backend.backend_identity = Some([7; 32]);
+        invalid_v1.push(with_backend);
+        for persisted in [&v1, &v2] {
+            let schema = if persisted.version == 1 {
+                TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA
+            } else {
+                TIER_DELETE_JOURNAL_V2_RECOVERY_SCHEMA
+            };
+            let mut invalid = Vec::new();
+            let mut with_exact = persisted.clone();
+            with_exact.version_id_exact = Some(false);
+            invalid.push(with_exact);
+            let mut with_version_state = persisted.clone();
+            with_version_state.version_state = Some(rustfs_filemeta::TransitionVersionState::Unknown);
+            invalid.push(with_version_state);
+            let mut with_state = persisted.clone();
+            with_state.state = Some(TierDeleteJournalState::Committed);
+            invalid.push(with_state);
+            let mut with_source = persisted.clone();
+            with_source.source = later.source.clone();
+            invalid.push(with_source);
+            let mut with_dispatch = persisted.clone();
+            with_dispatch.dispatch = later.dispatch.clone();
+            invalid.push(with_dispatch);
+            for record in invalid {
+                assert_rejected(record, schema);
+            }
+        }
+        for record in invalid_v1 {
+            assert_rejected(record, TIER_DELETE_JOURNAL_V1_RECOVERY_SCHEMA);
+        }
+    }
+
     #[test]
     fn tier_delete_journal_path_is_stable_and_sanitized() {
         let je = journal_entry();

crates/ecstore/src/bucket/lifecycle/transition_transaction.rs

@@ -34,7 +34,7 @@ use crate::bucket::lifecycle::tier_sweeper::{
 };
 use crate::disk::RUSTFS_META_BUCKET;
 use crate::error::{Error, Result as EcstoreResult};
-use crate::object_api::ObjectOptions;
+use crate::object_api::{ObjectInfo, ObjectOptions};
 use crate::services::tier::{tier::TierConfigMgr, warm_backend::TransitionCandidateProbe};
 use crate::storage_api_contracts::{
     list::ListOperations as _,
@@ -273,6 +273,14 @@ pub struct TransitionTransactionInit {
 
 impl TransitionTransaction {
     pub fn new(init: TransitionTransactionInit) -> Result<Self> {
+        Self::new_with_initial_state(init, TransitionTransactionState::UploadStarted)
+    }
+
+    pub(crate) fn new_compact(init: TransitionTransactionInit) -> Result<Self> {
+        Self::new_with_initial_state(init, TransitionTransactionState::UploadOutcomeUnknown)
+    }
+
+    fn new_with_initial_state(init: TransitionTransactionInit, state: TransitionTransactionState) -> Result<Self> {
         let remote_object =
             canonical_transition_remote_object(init.deployment_id, &init.source.bucket, init.transaction_id, init.write_id)?;
         let transaction = Self {
@@ -286,7 +294,7 @@ impl TransitionTransaction {
             backend_fingerprint: init.backend_fingerprint,
             remote_object,
             remote_version: TransitionRemoteVersion::unknown(),
-            state: TransitionTransactionState::UploadStarted,
+            state,
             not_after_unix_nanos: init.not_after_unix_nanos,
         };
         transaction.validate()?;
@@ -356,7 +364,7 @@ impl TransitionTransaction {
         remote_version: Option<TransitionRemoteVersion>,
     ) -> Result<TransitionTransactionFence> {
         self.check_fence(fence)?;
-        if !state_change_allowed(self.state, next) {
+        if !state_change_allowed_at(self.state, next, self.revision) {
             return Err(TransitionTransactionError::InvalidStateChange {
                 from: self.state,
                 to: next,
@@ -388,6 +396,14 @@ impl TransitionTransaction {
                 }
                 self.remote_version = TransitionRemoteVersion::unknown();
             }
+            TransitionTransactionState::LocalCommitStarted if self.state == TransitionTransactionState::UploadOutcomeUnknown => {
+                let remote_version =
+                    remote_version.ok_or(TransitionTransactionError::Corrupt("compact local commit requires remote version"))?;
+                if remote_version.is_unknown() {
+                    return Err(TransitionTransactionError::Corrupt("compact local commit requires known remote version"));
+                }
+                self.remote_version = remote_version;
+            }
             TransitionTransactionState::LocalCommitStarted | TransitionTransactionState::Committed => {
                 if let Some(remote_version) = remote_version
                     && remote_version != self.remote_version
@@ -640,7 +656,7 @@ pub(crate) async fn save_transition_transaction_record_if_current(
 ) -> EcstoreResult<()> {
     let object = transition_transaction_record_object_name(next.transaction_id).map_err(transition_transaction_store_error)?;
     let revision_is_next = expected.revision.checked_add(1) == Some(next.revision);
-    let state_is_next = state_change_allowed(expected.state, next.state)
+    let state_is_next = state_change_allowed_at(expected.state, next.state, expected.revision)
         || matches!(
             (expected.state, next.state),
             (
@@ -954,6 +970,10 @@ pub enum TransitionOperatorError {
         expected: String,
         actual: TransitionOperatorProbe,
     },
+    #[error("transition recovery control is stale")]
+    StaleRecoveryControl,
+    #[error("transition recovery control is not eligible for operator retry")]
+    RetryNotAllowed,
     #[error("transition transaction store failed: {0}")]
     Store(#[source] Error),
     #[error("remote tier reconciliation failed: {0}")]
@@ -962,6 +982,179 @@ pub enum TransitionOperatorError {
 
 type TransitionOperatorResult<T> = std::result::Result<T, TransitionOperatorError>;
 
+#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
+pub struct TransitionRecoveryRetryStatus {
+    pub control_id: String,
+    pub transaction_id: Uuid,
+    pub state: TransitionTransactionState,
+    pub classification: IlmRecoveryClassification,
+    pub control_revision: u64,
+    pub attempt_count: u64,
+    pub consecutive_failure_count: u32,
+    pub last_error_code: IlmRecoveryErrorCode,
+    pub source_generation_sha256: String,
+    pub copy_set_sha256: String,
+    pub retry_ready: bool,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub retry_not_ready_reason: Option<&'static str>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
+pub struct TransitionRecoveryRetryResult {
+    pub control_id: String,
+    pub transaction_id: Uuid,
+    pub previous_revision: u64,
+    pub revision: u64,
+    pub classification: IlmRecoveryClassification,
+    pub attempt_count: u64,
+    pub source_generation_sha256: String,
+}
+
+struct TransitionRecoveryRetryContext {
+    observed: ObservedIlmRecoveryControl,
+    transaction: TransitionTransaction,
+    source_generation_sha256: String,
+}
+
+fn transition_recovery_retry_readiness(control: &IlmRecoveryControl) -> (bool, Option<&'static str>) {
+    if control.owner.is_some() {
+        return (false, Some("attempt_owned"));
+    }
+    match control.classification {
+        IlmRecoveryClassification::RetainedAmbiguous | IlmRecoveryClassification::OperatorRequired => (true, None),
+        IlmRecoveryClassification::Retrying => (false, Some("already_retrying")),
+        IlmRecoveryClassification::Corrupt => (false, Some("source_corrupt")),
+        IlmRecoveryClassification::Abandoned => (false, Some("source_abandoned")),
+        IlmRecoveryClassification::Terminal => (false, Some("source_terminal")),
+    }
+}
+
+async fn load_transition_recovery_retry_context(
+    api: Arc<ECStore>,
+    control_id: &str,
+) -> TransitionOperatorResult<TransitionRecoveryRetryContext> {
+    let observed = match load_recovery_control(api.clone(), IlmRecoveryProtocol::TransitionTransaction, control_id).await {
+        Ok(observed) => observed,
+        Err(Error::ConfigNotFound) => return Err(TransitionOperatorError::NotFound),
+        Err(err) => return Err(TransitionOperatorError::Store(err)),
+    };
+    let transaction_id = Uuid::parse_str(&observed.control.identity.stable_operation_identity)
+        .ok()
+        .filter(|transaction_id| !transaction_id.is_nil())
+        .ok_or(TransitionOperatorError::StaleRecoveryControl)?;
+    let canonical_path = transition_transaction_record_object_name(transaction_id)
+        .map_err(|err| TransitionOperatorError::Store(Error::other(err)))?;
+    if observed.control.identity.canonical_source_path != canonical_path
+        || observed.control.identity.record_class != "transition_transaction_v1"
+    {
+        return Err(TransitionOperatorError::StaleRecoveryControl);
+    }
+    let transaction = match load_transition_transaction_record(api.clone(), transaction_id).await {
+        Ok(transaction) => transaction,
+        Err(Error::ConfigNotFound) => return Err(TransitionOperatorError::NotFound),
+        Err(err) => return Err(TransitionOperatorError::Store(err)),
+    };
+    let source = observe_recovery_source(api, &canonical_path, TRANSITION_TRANSACTION_SCHEMA)
+        .await
+        .map_err(TransitionOperatorError::Store)?;
+    let exact_source = source.is_consistent()
+        && source.generation == observed.control.observed_source_generation
+        && source
+            .canonical_data
+            .as_deref()
+            .is_some_and(|data| TransitionTransaction::decode(transaction_id, data).is_ok_and(|decoded| decoded == transaction));
+    if !exact_source {
+        return Err(TransitionOperatorError::StaleRecoveryControl);
+    }
+    let generation = serde_json::to_vec(&observed.control.observed_source_generation)
+        .map_err(|err| TransitionOperatorError::Store(Error::other(err)))?;
+    Ok(TransitionRecoveryRetryContext {
+        observed,
+        transaction,
+        source_generation_sha256: hex_sha256(&generation, ToOwned::to_owned),
+    })
+}
+
+pub async fn inspect_transition_recovery_retry_for_operator(
+    api: Arc<ECStore>,
+    control_id: &str,
+) -> TransitionOperatorResult<TransitionRecoveryRetryStatus> {
+    let context = load_transition_recovery_retry_context(api, control_id).await?;
+    let (retry_ready, retry_not_ready_reason) = transition_recovery_retry_readiness(&context.observed.control);
+    Ok(TransitionRecoveryRetryStatus {
+        control_id: control_id.to_string(),
+        transaction_id: context.transaction.transaction_id,
+        state: context.transaction.state,
+        classification: context.observed.control.classification,
+        control_revision: context.observed.control.revision,
+        attempt_count: context.observed.control.attempt_count,
+        consecutive_failure_count: context.observed.control.consecutive_failure_count,
+        last_error_code: context.observed.control.last_error_code,
+        source_generation_sha256: context.source_generation_sha256,
+        copy_set_sha256: context.observed.control.observed_source_generation.copy_set_sha256.clone(),
+        retry_ready,
+        retry_not_ready_reason,
+    })
+}
+
+pub async fn retry_transition_recovery_for_operator(
+    api: Arc<ECStore>,
+    control_id: &str,
+    expected_control_revision: u64,
+    expected_source_generation_sha256: &str,
+) -> TransitionOperatorResult<TransitionRecoveryRetryResult> {
+    let control_object = recovery_control_record_object_name(IlmRecoveryProtocol::TransitionTransaction, control_id)
+        .map_err(|err| TransitionOperatorError::Store(Error::other(err)))?;
+    let retry_lock = api
+        .new_ns_lock(RUSTFS_META_BUCKET, &format!("{control_object}.recovery-lock"))
+        .await
+        .map_err(TransitionOperatorError::Store)?;
+    let retry_guard = retry_lock
+        .get_write_lock(crate::set_disk::get_lock_acquire_timeout())
+        .await
+        .map_err(|err| TransitionOperatorError::Store(Error::other(err)))?;
+    let context = load_transition_recovery_retry_context(api.clone(), control_id).await?;
+    let (retry_ready, _) = transition_recovery_retry_readiness(&context.observed.control);
+    if !retry_ready {
+        return Err(TransitionOperatorError::RetryNotAllowed);
+    }
+    if retry_guard.is_lock_lost()
+        || expected_control_revision == 0
+        || context.observed.control.revision != expected_control_revision
+        || context.source_generation_sha256 != expected_source_generation_sha256
+    {
+        return Err(TransitionOperatorError::StaleRecoveryControl);
+    }
+    let previous_revision = context.observed.control.revision;
+    let mut next = context.observed.control.clone();
+    next.retry_for_operator(&context.observed.control.observed_source_generation)
+        .map_err(|_| TransitionOperatorError::RetryNotAllowed)?;
+    if retry_guard.is_lock_lost() {
+        return Err(TransitionOperatorError::StaleRecoveryControl);
+    }
+    save_recovery_control_if_current(api.clone(), &context.observed, &next)
+        .await
+        .map_err(|err| match err {
+            Error::PreconditionFailed => TransitionOperatorError::StaleRecoveryControl,
+            err => TransitionOperatorError::Store(err),
+        })?;
+    let persisted = load_recovery_control(api, IlmRecoveryProtocol::TransitionTransaction, control_id)
+        .await
+        .map_err(TransitionOperatorError::Store)?;
+    if retry_guard.is_lock_lost() || persisted.control != next {
+        return Err(TransitionOperatorError::StaleRecoveryControl);
+    }
+    Ok(TransitionRecoveryRetryResult {
+        control_id: control_id.to_string(),
+        transaction_id: context.transaction.transaction_id,
+        previous_revision,
+        revision: persisted.control.revision,
+        classification: persisted.control.classification,
+        attempt_count: persisted.control.attempt_count,
+        source_generation_sha256: context.source_generation_sha256,
+    })
+}
+
 fn validate_operator_reconcile_transaction(
     transaction: &TransitionTransaction,
     now_unix_nanos: i128,
@@ -1706,12 +1899,27 @@ async fn local_commit_matches_transaction(api: Arc<ECStore>, transaction: &Trans
         .get_object_info(&transaction.source.bucket, &transaction.source.object, &opts)
         .await?;
     let transitioned = &object.transitioned_object;
-    Ok(transitioned.status == TRANSITION_COMPLETE
+    Ok(local_object_matches_transition_source(&object, &transaction.source)
+        && transitioned.status == TRANSITION_COMPLETE
         && transitioned.name == transaction.remote_object
         && transitioned.tier == transaction.tier_name
         && transitioned.version_id == transaction.remote_version.tier_delete_version_id().unwrap_or_default())
 }
 
+fn local_object_matches_transition_source(object: &ObjectInfo, source: &TransitionSourceIdentity) -> bool {
+    let observed_version_id = object.version_id.filter(|version_id| !version_id.is_nil());
+    let observed_mod_time = object
+        .mod_time
+        .and_then(|mod_time| i64::try_from(mod_time.unix_timestamp_nanos()).ok());
+    object.bucket == source.bucket
+        && object.name == source.object
+        && observed_version_id == source.version_id
+        && object.data_dir == Some(source.data_dir)
+        && observed_mod_time == Some(source.mod_time_unix_nanos)
+        && object.size == source.size
+        && object.etag.as_deref() == Some(source.etag.as_str())
+}
+
 fn transition_source_lookup_options(transaction: &TransitionTransaction) -> ObjectOptions {
     ObjectOptions {
         version_id: match transaction.source.version_mode {
@@ -1954,7 +2162,7 @@ where
     }
 }
 
-fn state_change_allowed(from: TransitionTransactionState, to: TransitionTransactionState) -> bool {
+fn state_change_allowed_at(from: TransitionTransactionState, to: TransitionTransactionState, revision: u64) -> bool {
     matches!(
         (from, to),
         (TransitionTransactionState::UploadStarted, TransitionTransactionState::Uploaded)
@@ -1966,7 +2174,9 @@ fn state_change_allowed(from: TransitionTransactionState, to: TransitionTransact
             | (TransitionTransactionState::UploadOutcomeUnknown, TransitionTransactionState::Uploaded)
             | (TransitionTransactionState::Uploaded, TransitionTransactionState::LocalCommitStarted)
             | (TransitionTransactionState::LocalCommitStarted, TransitionTransactionState::Committed)
-    )
+    ) || (revision == 1
+        && from == TransitionTransactionState::UploadOutcomeUnknown
+        && to == TransitionTransactionState::LocalCommitStarted)
 }
 
 fn state_requires_known_remote_version(state: TransitionTransactionState) -> bool {
@@ -2183,6 +2393,41 @@ mod tests {
         }
     }
 
+    #[test]
+    fn local_commit_proof_requires_the_complete_source_identity() {
+        let source = source_identity(TransitionSourceVersionMode::Versioned);
+        let exact = ObjectInfo {
+            bucket: source.bucket.clone(),
+            name: source.object.clone(),
+            version_id: source.version_id,
+            data_dir: Some(source.data_dir),
+            mod_time: Some(
+                time::OffsetDateTime::from_unix_timestamp_nanos(i128::from(source.mod_time_unix_nanos))
+                    .expect("source timestamp should be valid"),
+            ),
+            size: source.size,
+            etag: Some(source.etag.clone()),
+            ..Default::default()
+        };
+        assert!(local_object_matches_transition_source(&exact, &source));
+
+        let mut changed = exact.clone();
+        changed.version_id = Some(Uuid::new_v4());
+        assert!(!local_object_matches_transition_source(&changed, &source));
+        changed = exact.clone();
+        changed.data_dir = Some(Uuid::new_v4());
+        assert!(!local_object_matches_transition_source(&changed, &source));
+        changed = exact.clone();
+        changed.mod_time = changed.mod_time.map(|value| value + Duration::from_nanos(1));
+        assert!(!local_object_matches_transition_source(&changed, &source));
+        changed = exact.clone();
+        changed.size += 1;
+        assert!(!local_object_matches_transition_source(&changed, &source));
+        changed = exact;
+        changed.etag = Some("different-etag".to_string());
+        assert!(!local_object_matches_transition_source(&changed, &source));
+    }
+
     fn cleanup_proof(transaction: &TransitionTransaction, decision: TransitionCleanupDecision) -> TransitionCleanupProof {
         TransitionCleanupProof {
             transaction_id: transaction.transaction_id,
@@ -2384,6 +2629,57 @@ mod tests {
         assert_eq!(transaction.state, TransitionTransactionState::Committed);
     }
 
+    #[test]
+    fn compact_state_sequence_is_distinguishable_and_keeps_legacy_edges_strict() {
+        let init = TransitionTransactionInit {
+            deployment_id: Uuid::new_v4(),
+            transaction_id: Uuid::new_v4(),
+            owner_epoch: Uuid::new_v4(),
+            write_id: Uuid::new_v4(),
+            source: source_identity(TransitionSourceVersionMode::Versioned),
+            tier_name: "warm-tier".to_string(),
+            backend_fingerprint: BACKEND_FINGERPRINT,
+            not_after_unix_nanos: 1_780_000_000_000_000_000,
+        };
+        let mut compact = TransitionTransaction::new_compact(init).expect("compact transaction should be created");
+        assert_eq!(compact.state, TransitionTransactionState::UploadOutcomeUnknown);
+        assert_eq!(compact.revision, 1);
+        let remote_version = TransitionRemoteVersion::versioned(Uuid::new_v4().to_string());
+        let fence = compact
+            .advance(
+                compact.fence(),
+                TransitionTransactionState::LocalCommitStarted,
+                Some(remote_version.clone()),
+            )
+            .expect("compact upload should persist its exact candidate at the local commit fence");
+        assert_eq!(fence.revision, 2);
+        assert_eq!(compact.remote_version, remote_version);
+        assert_eq!(compact.state, TransitionTransactionState::LocalCommitStarted);
+        let encoded = compact
+            .encode()
+            .expect("compact transaction should encode as v1-compatible bytes");
+        assert_eq!(
+            TransitionTransaction::decode(compact.transaction_id, &encoded).expect("compact transaction should decode"),
+            compact
+        );
+
+        let mut legacy_unknown = new_transaction();
+        legacy_unknown
+            .advance(legacy_unknown.fence(), TransitionTransactionState::UploadOutcomeUnknown, None)
+            .expect("legacy transaction should persist its pre-upload fence");
+        assert!(matches!(
+            legacy_unknown.advance(
+                legacy_unknown.fence(),
+                TransitionTransactionState::LocalCommitStarted,
+                Some(TransitionRemoteVersion::unversioned()),
+            ),
+            Err(TransitionTransactionError::InvalidStateChange {
+                from: TransitionTransactionState::UploadOutcomeUnknown,
+                to: TransitionTransactionState::LocalCommitStarted,
+            })
+        ));
+    }
+
     #[test]
     fn cleanup_pending_requires_exact_proof_and_state_specific_decision() {
         let mut transaction = new_transaction();

crates/ecstore/src/bucket/metadata_sys.rs

@@ -54,6 +54,9 @@ pub type BucketConfigPublishHook = Box<dyn Fn(&str, &str, Option<(&[u8], OffsetD
 pub static BUCKET_CONFIG_PUBLISH_HOOK: std::sync::OnceLock<BucketConfigPublishHook> = std::sync::OnceLock::new();
 
 const BUCKET_METADATA_REFRESH_INTERVAL: Duration = Duration::from_secs(15 * 60);
+const LOG_COMPONENT_ECSTORE: &str = "ecstore";
+const LOG_SUBSYSTEM_BUCKET_METADATA: &str = "bucket_metadata";
+const EVENT_BUCKET_METADATA_LOAD_FAILED: &str = "bucket_metadata_load_failed";
 
 #[cfg(any(test, feature = "test-util"))]
 struct ConfigWriteLockProbeState {
@@ -1614,13 +1617,20 @@ impl BucketMetadataSys {
 
         let results = join_all(futures).await;
 
-        for (idx, res) in results.into_iter().enumerate() {
+        for (bucket, res) in buckets.iter().zip(results) {
             match res {
                 Ok(()) => {}
                 Err(e) => {
-                    error!("Unable to load bucket metadata, will be retried: {:?}", e);
+                    if failed_buckets.insert(bucket.clone()) {
-                    if let Some(bucket) = buckets.get(idx) {
+                        error!(
-                        failed_buckets.insert(bucket.clone());
+                            event = EVENT_BUCKET_METADATA_LOAD_FAILED,
+                            component = LOG_COMPONENT_ECSTORE,
+                            subsystem = LOG_SUBSYSTEM_BUCKET_METADATA,
+                            result = "retry_pending",
+                            bucket = %bucket,
+                            error_code = ?e.code(),
+                            "Unable to load bucket metadata; retry scheduled"
+                        );
                     }
                 }
             }
@@ -1647,12 +1657,19 @@ impl BucketMetadataSys {
             });
         }
         let results = join_all(futures).await;
-        for (idx, result) in results.into_iter().enumerate() {
+        for (bucket, result) in buckets.iter().zip(results) {
-            if let Err(err) = result {
+            if let Err(err) = result
-                error!("Unable to load bucket metadata, will be retried: {:?}", err);
+                && failed_buckets.insert(bucket.clone())
-                if let Some(bucket) = buckets.get(idx) {
+            {
-                    failed_buckets.insert(bucket.clone());
+                error!(
-                }
+                    event = EVENT_BUCKET_METADATA_LOAD_FAILED,
+                    component = LOG_COMPONENT_ECSTORE,
+                    subsystem = LOG_SUBSYSTEM_BUCKET_METADATA,
+                    result = "retry_pending",
+                    bucket = %bucket,
+                    error_code = ?err.code(),
+                    "Unable to load bucket metadata; retry scheduled"
+                );
             }
         }
     }

crates/ecstore/src/bucket/replication/mod.rs

@@ -66,6 +66,7 @@ pub(crate) use replication_lifecycle_bridge::ReplicationLifecycleBridge;
 pub(crate) use replication_migration_bridge::ReplicationMigrationBridge;
 pub use replication_object_bridge::ReplicationObjectBridge;
 pub use replication_object_config::{DeleteReplicationConfigSnapshot, ReplicationConfig};
+pub(crate) use replication_object_decision_boundary::replication_etags_match;
 pub use replication_object_decision_boundary::{
     MustReplicateOptions, ReplicationDeleteScheduleInput, ReplicationDeleteStateSource, delete_replication_state_from_config,
     delete_replication_version_id, should_schedule_delete_replication, should_use_existing_delete_replication_info,
@@ -88,5 +89,6 @@ pub use replication_state::{ReplicationStats, RuntimeReplicationTargetBacklog};
 pub use replication_stats_boundary::{BucketReplicationStat, BucketReplicationStats, BucketStats, InQueueMetric, XferStats};
 pub use replication_storage_boundary::{ReplicationObjectIO, ReplicationStorage};
 pub use replication_target_boundary::SsecPassthroughCapability;
+pub use replication_target_boundary::VersionIdentityCapability;
 pub use replication_target_boundary::{ObjectLockIntegrity, object_lock_put_integrity};
 pub(crate) use replication_target_config_bridge::ReplicationTargetConfigBridge;

crates/ecstore/src/bucket/replication/replication_filemeta_boundary.rs

@@ -12,6 +12,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#[cfg(test)]
+pub(crate) use rustfs_filemeta::ObjectPartInfo;
 pub use rustfs_replication::{MrfOpKind, MrfReplicateEntry};
 pub(crate) use rustfs_replication::{
     REPLICATE_EXISTING, REPLICATE_HEAL_DELETE, ReplicateTargetDecision, ReplicatedInfos, ReplicatedTargetInfo, ReplicationAction,

crates/ecstore/src/bucket/replication/replication_object_decision_boundary.rs

@@ -12,6 +12,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#[cfg(test)]
+pub(crate) use rustfs_replication::ReplicationMultipartPlanError;
 pub use rustfs_replication::{
     MustReplicateOptions, ReplicationDeleteScheduleInput, ReplicationDeleteStateSource, delete_replication_state_from_config,
     delete_replication_version_id, should_schedule_delete_replication, should_use_existing_delete_replication_info,

crates/ecstore/src/bucket/replication/replication_pool.rs

@@ -75,6 +75,7 @@ use tracing::{debug, info, instrument, warn};
 const EVENT_REPLICATION_WORKER_RESIZE_SKIPPED: &str = "replication_worker_resize_skipped";
 const EVENT_REPLICATION_WORKER_RESIZED: &str = "replication_worker_resized";
 const EVENT_REPLICATION_BACKPRESSURE: &str = "replication_backpressure";
+const EVENT_REPLICATION_IN_FLIGHT_SKIPPED: &str = "replication_in_flight_skipped";
 const EVENT_REPLICATION_RESYNC_LOAD_SKIPPED: &str = "replication_resync_load_skipped";
 const EVENT_REPLICATION_RESYNC_RECOVERED: &str = "replication_resync_recovered";
 const EVENT_REPLICATION_MRF_QUEUE_UNAVAILABLE: &str = "replication_mrf_queue_unavailable";
@@ -1089,6 +1090,9 @@ pub struct ReplicationPool<S: ReplicationStorage> {
     workers: RwLock<Vec<Sender<ReplicationOperation>>>,
     lrg_workers: RwLock<Vec<Sender<ReplicationOperation>>>,
 
+    /// Object versions queued or being replicated right now (backlog#2362).
+    in_flight: Arc<ReplicationInFlight>,
+
     // MRF (Most Recent Failures) channels
     mrf_replica_tx: Sender<ReplicationOperation>,
     // Shared among N MRF workers; Arc allows spawning more than one worker.
@@ -1147,6 +1151,7 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
             storage,
             workers: RwLock::new(Vec::new()),
             lrg_workers: RwLock::new(Vec::new()),
+            in_flight: Arc::new(ReplicationInFlight::default()),
             mrf_replica_tx,
             mrf_replica_rx: Arc::new(Mutex::new(mrf_replica_rx)),
             mrf_save_tx,
@@ -1202,12 +1207,13 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
             let active_counter = self.active_lrg_workers.clone();
             let storage = self.storage.clone();
             let stats = self.stats.clone();
+            let in_flight = self.in_flight.clone();
 
             let handle = tokio::spawn(async move {
                 let mut rx = rx;
                 while let Some(operation) = rx.recv().await {
                     let _active = ActiveWorkerGuard::new(active_counter.clone());
-                    process_replication_operation(operation, stats.clone(), storage.clone()).await;
+                    process_replication_operation(operation, stats.clone(), storage.clone(), in_flight.clone()).await;
                 }
             });
 
@@ -1261,12 +1267,13 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
             let active_counter = self.active_workers.clone();
             let stats = self.stats.clone();
             let storage = self.storage.clone();
+            let in_flight = self.in_flight.clone();
 
             let handle = tokio::spawn(async move {
                 let mut rx = rx;
                 while let Some(operation) = rx.recv().await {
                     let _active = ActiveWorkerGuard::new(active_counter.clone());
-                    process_replication_operation(operation, stats.clone(), storage.clone()).await;
+                    process_replication_operation(operation, stats.clone(), storage.clone(), in_flight.clone()).await;
                 }
             });
 
@@ -1305,6 +1312,7 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
             let active_counter = self.active_mrf_workers.clone();
             let stats = self.stats.clone();
             let storage = self.storage.clone();
+            let in_flight = self.in_flight.clone();
             let mrf_rx = Arc::clone(&self.mrf_replica_rx);
 
             let handle = tokio::spawn(async move {
@@ -1324,7 +1332,7 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
                     let Some(operation) = operation else { break };
 
                     let _active = ActiveWorkerGuard::new(active_counter.clone());
-                    process_replication_operation(operation, stats.clone(), storage.clone()).await;
+                    process_replication_operation(operation, stats.clone(), storage.clone(), in_flight.clone()).await;
                 }
             });
             self.task_handles.lock().await.push(handle);
@@ -1454,6 +1462,24 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
 
     /// Queues a replica task
     pub async fn queue_replica_task(&self, ri: ReplicateObjectInfo) -> ReplicationQueueAdmission {
+        // A version that is already queued or being uploaded is not driven a
+        // second time: the scanner heal pass sees it as PENDING until the
+        // first upload lands and would otherwise re-queue it every cycle
+        // (backlog#2362). The key is released when the worker finishes, or
+        // below when no worker accepts the task.
+        if !self.in_flight.try_begin(&ri) {
+            debug!(
+                event = EVENT_REPLICATION_IN_FLIGHT_SKIPPED,
+                component = LOG_COMPONENT_ECSTORE,
+                subsystem = LOG_SUBSYSTEM_REPLICATION,
+                bucket = %ri.bucket,
+                object = %ri.name,
+                version_id = ?ri.version_id,
+                op_type = ?ri.op_type,
+                "Replication task already in flight; not queued again"
+            );
+            return ReplicationQueueAdmission::Skipped;
+        }
         let target_arns = ri.dsc.replicate_target_arns();
         // If object is large, queue it to a static set of large workers
         if should_queue_large_object(ri.size) {
@@ -1484,7 +1510,9 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
                     let resize = large_worker_backpressure_resize(existing, self.active_lrg_workers(), max_l_workers);
                     drop(lrg_workers);
 
-                    // Queue to MRF if worker is busy.
+                    // Queue to MRF if worker is busy. The MRF replay re-enters
+                    // this function, so the version is no longer in flight.
+                    self.in_flight.finish(&ri);
                     let admission = self.queue_mrf_save_admission(ri.to_mrf_entry(), "large_object").await;
 
                     if let Some(resize) = resize {
@@ -1493,6 +1521,7 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
                     return admission;
                 }
             }
+            self.in_flight.finish(&ri);
             return ReplicationQueueAdmission::Missed;
         }
 
@@ -1501,6 +1530,7 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
         let ch = self.worker_queue_channel(&ri.op_type, &ri.bucket, &ri.name, ri.size).await;
 
         let Some(channel) = ch else {
+            self.in_flight.finish(&ri);
             return ReplicationQueueAdmission::Missed;
         };
 
@@ -1512,7 +1542,9 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
         self.stats.dec_q(&ri.bucket, ri.size, ri.delete_marker, ri.op_type);
         self.stats.dec_target_q(&ri.bucket, &target_arns, ri.size);
 
-        // Queue to MRF if all workers are busy.
+        // Queue to MRF if all workers are busy. The MRF replay re-enters this
+        // function, so the version is no longer in flight.
+        self.in_flight.finish(&ri);
         let admission = self.queue_mrf_save_admission(ri.to_mrf_entry(), "object").await;
 
         // Try to scale up workers based on priority
@@ -1811,7 +1843,7 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
     ) {
         while let Some(operation) = rx.recv().await {
             let _active = ActiveWorkerGuard::new(active_counter.clone());
-            process_replication_operation(operation, stats.clone(), self.storage.clone()).await;
+            process_replication_operation(operation, stats.clone(), self.storage.clone(), self.in_flight.clone()).await;
         }
     }
 
@@ -1829,7 +1861,7 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
     ) {
         while let Some(operation) = rx.recv().await {
             let _active = ActiveWorkerGuard::new(active_counter.clone());
-            process_replication_operation(operation, stats.clone(), storage.clone()).await;
+            process_replication_operation(operation, stats.clone(), storage.clone(), self.in_flight.clone()).await;
         }
     }
 
@@ -1846,7 +1878,7 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
     ) {
         while let Some(operation) = rx.recv().await {
             let _active = ActiveWorkerGuard::new(active_counter.clone());
-            process_replication_operation(operation, stats.clone(), self.storage.clone()).await;
+            process_replication_operation(operation, stats.clone(), self.storage.clone(), self.in_flight.clone()).await;
         }
     }
 
@@ -2281,6 +2313,64 @@ impl<S: ReplicationStorage> ReplicationPool<S> {
     }
 }
 
+/// Object versions currently queued or being uploaded, keyed by bucket,
+/// object name and version. `queue_replica_task` admits a version only once
+/// while it is in flight; the scanner heal pass and MRF replays that arrive
+/// in the meantime are `Skipped` instead of driving a second complete upload
+/// (backlog#2362). Entries are removed when the worker finishes the task or
+/// when no worker accepted it.
+#[derive(Debug, Default)]
+pub(crate) struct ReplicationInFlight {
+    keys: std::sync::Mutex<std::collections::HashSet<(String, String, Option<uuid::Uuid>)>>,
+}
+
+impl ReplicationInFlight {
+    fn lock(&self) -> std::sync::MutexGuard<'_, std::collections::HashSet<(String, String, Option<uuid::Uuid>)>> {
+        self.keys.lock().unwrap_or_else(|poisoned| poisoned.into_inner())
+    }
+
+    /// Claim `ri`; `false` when the same version is already in flight.
+    fn try_begin(&self, ri: &ReplicateObjectInfo) -> bool {
+        self.lock().insert((ri.bucket.clone(), ri.name.clone(), ri.version_id))
+    }
+
+    fn finish(&self, ri: &ReplicateObjectInfo) {
+        self.lock().remove(&(ri.bucket.clone(), ri.name.clone(), ri.version_id));
+    }
+
+    #[cfg(test)]
+    fn len(&self) -> usize {
+        self.lock().len()
+    }
+}
+
+/// Releases the in-flight claim when the worker is done with the task,
+/// including when replication panics.
+struct ReplicationInFlightGuard {
+    in_flight: Arc<ReplicationInFlight>,
+    key: ReplicateObjectInfo,
+}
+
+impl ReplicationInFlightGuard {
+    fn new(in_flight: Arc<ReplicationInFlight>, ri: &ReplicateObjectInfo) -> Self {
+        Self {
+            in_flight,
+            key: ReplicateObjectInfo {
+                bucket: ri.bucket.clone(),
+                name: ri.name.clone(),
+                version_id: ri.version_id,
+                ..Default::default()
+            },
+        }
+    }
+}
+
+impl Drop for ReplicationInFlightGuard {
+    fn drop(&mut self) {
+        self.in_flight.finish(&self.key);
+    }
+}
+
 struct ActiveWorkerGuard {
     counter: Arc<AtomicI32>,
 }
@@ -2342,10 +2432,12 @@ async fn process_replication_operation<S: ReplicationStorage>(
     operation: ReplicationOperation,
     stats: Arc<ReplicationStats>,
     storage: Arc<S>,
+    in_flight: Arc<ReplicationInFlight>,
 ) {
     match operation {
         ReplicationOperation::Object(obj_info) => {
             let _backlog = ReplicationBacklogGuard::for_object(stats, obj_info.as_ref());
+            let _in_flight = ReplicationInFlightGuard::new(in_flight, obj_info.as_ref());
             replicate_object(*obj_info, storage).await;
         }
         ReplicationOperation::Delete(del_info) => {
@@ -3079,7 +3171,11 @@ pub async fn queue_replication_heal(bucket: &str, oi: ObjectInfo, retry_count: u
     }
 
     let rcfg = match ReplicationMetadataStore::optional_replication_config(bucket).await {
-        Ok(Some(config)) => config,
+        Ok(Some(config)) => Some(config),
+        // A bucket without a configuration still owes its pending purges an
+        // answer: the delete worker finishes them locally as abandoned, which
+        // is what makes the bucket deletable again (rustfs/backlog#2340).
+        Ok(None) if owes_version_purge(&oi) => None,
         Ok(None) => return ReplicationQueueAdmission::Skipped,
         Err(err) => {
             debug!(
@@ -3129,7 +3225,7 @@ pub async fn queue_replication_heal(bucket: &str, oi: ObjectInfo, retry_count: u
         }
     };
 
-    let rcfg_wrapper = ReplicationConfig::new(Some(rcfg), tgts);
+    let rcfg_wrapper = ReplicationConfig::new(rcfg, tgts);
     queue_replication_heal_internal(bucket, oi, rcfg_wrapper, retry_count)
         .await
         .admission
@@ -3157,6 +3253,17 @@ pub async fn queue_replication_metadata(bucket: &str, oi: ObjectInfo, retry_coun
     }
 }
 
+/// A version purge the persisted state still owes to named targets. Without
+/// the target list nothing can be settled, so such a version keeps the
+/// ordinary "no configuration, nothing to heal" skip.
+fn owes_version_purge(oi: &ObjectInfo) -> bool {
+    !oi.version_purge_status.is_empty()
+        && oi
+            .version_purge_status_internal
+            .as_deref()
+            .is_some_and(|statuses| !statuses.trim().is_empty())
+}
+
 /// queue_replication_heal_internal enqueues objects that failed replication OR eligible for resyncing through
 /// an ongoing resync operation or via existing objects replication configuration setting.
 pub(crate) async fn queue_replication_heal_internal(
@@ -3175,7 +3282,11 @@ pub(crate) async fn queue_replication_heal_internal(
         };
     }
 
-    if rcfg.config.is_none() || rcfg.remotes.is_none() {
+    // Without a configuration or targets there is nothing to replicate —
+    // except a version purge the bucket still owes: its stored decision names
+    // the targets, and the delete worker settles the ones no longer
+    // configured as abandoned (rustfs/backlog#2340).
+    if (rcfg.config.is_none() || rcfg.remotes.is_none()) && !owes_version_purge(&oi) {
         return ReplicationHealQueueResult {
             object_info: roi,
             admission: ReplicationQueueAdmission::Skipped,
@@ -3220,12 +3331,15 @@ pub(crate) async fn queue_replication_heal_internal(
         }
         ReplicationHealQueueAction::QueueDelete(dv) => {
             // A purge the peer denied under object lock cannot succeed until
-            // the lock lapses (#6850); requeuing it every heal cycle only
+            // the lock lapses (#6850), and one whose replica cannot be told
+            // apart on a target that mints its own version ids cannot
+            // succeed until the ledger or an operator resolves it
+            // (rustfs/backlog#2340); requeuing either every heal cycle only
             // burns bandwidth and failure counters. The backoff expires on
             // its own, so the purge is probed again — and converges — once
-            // the retention window has a chance of being over.
+            // the condition has a chance of being over.
             if super::replication_object_decision_boundary::is_version_delete_replication(&dv.delete_object)
-                && super::replication_resyncer::object_lock_denied_purge_backoff_active(&dv)
+                && super::replication_resyncer::purge_backoff_active(&dv)
             {
                 return ReplicationHealQueueResult {
                     object_info: roi,
@@ -3707,6 +3821,7 @@ mod tests {
             stats: Arc::new(ReplicationStats::new()),
             workers: RwLock::new(Vec::new()),
             lrg_workers: RwLock::new(Vec::new()),
+            in_flight: Arc::new(ReplicationInFlight::default()),
             mrf_replica_tx,
             mrf_replica_rx: Arc::new(Mutex::new(mrf_replica_rx)),
             mrf_save_tx,
@@ -3773,6 +3888,90 @@ mod tests {
         assert_eq!(current_queue(&pool, "admission-bucket").await, (1, 4096));
     }
 
+    #[tokio::test]
+    async fn queue_replica_task_admits_a_version_once_while_it_is_in_flight() {
+        let pool = new_test_replication_pool(Arc::new(LoadResyncNodeStore::new("node-a", empty_resync_shared_state()))).await;
+        let (tx, _rx) = mpsc::channel(4);
+        pool.workers.write().await.push(tx);
+        let ri = ReplicateObjectInfo {
+            bucket: "in-flight-bucket".to_string(),
+            name: "object".to_string(),
+            version_id: Some(uuid::Uuid::new_v4()),
+            size: 4096,
+            op_type: ReplicationType::Object,
+            ..Default::default()
+        };
+
+        assert_eq!(pool.queue_replica_task(ri.clone()).await, ReplicationQueueAdmission::Queued);
+        // backlog#2362: the scanner heal pass sees the version as PENDING
+        // until the worker lands it; a second request must not drive it again.
+        assert_eq!(pool.queue_replica_task(ri.clone()).await, ReplicationQueueAdmission::Skipped);
+        assert_eq!(current_queue(&pool, "in-flight-bucket").await, (1, 4096));
+
+        // Another version of the same key is independent work.
+        let newer = ReplicateObjectInfo {
+            version_id: Some(uuid::Uuid::new_v4()),
+            ..ri.clone()
+        };
+        assert_eq!(pool.queue_replica_task(newer).await, ReplicationQueueAdmission::Queued);
+        assert_eq!(pool.in_flight.len(), 2);
+
+        // Once the worker finishes, the same version may be queued again
+        // (for example after a FAILED status).
+        pool.in_flight.finish(&ri);
+        assert_eq!(pool.queue_replica_task(ri).await, ReplicationQueueAdmission::Queued);
+        assert_eq!(current_queue(&pool, "in-flight-bucket").await, (3, 3 * 4096));
+    }
+
+    #[tokio::test]
+    async fn queue_replica_task_releases_the_version_when_no_worker_accepts_it() {
+        let pool = new_test_replication_pool(Arc::new(LoadResyncNodeStore::new("node-a", empty_resync_shared_state()))).await;
+        let ri = ReplicateObjectInfo {
+            bucket: "no-worker-bucket".to_string(),
+            name: "object".to_string(),
+            version_id: Some(uuid::Uuid::new_v4()),
+            size: 4096,
+            op_type: ReplicationType::Object,
+            ..Default::default()
+        };
+
+        // No worker channel: the task is missed and must not stay claimed.
+        assert_eq!(pool.queue_replica_task(ri.clone()).await, ReplicationQueueAdmission::Missed);
+        assert_eq!(pool.in_flight.len(), 0);
+        assert_eq!(pool.queue_replica_task(ri.clone()).await, ReplicationQueueAdmission::Missed);
+
+        // A full worker channel hands the task to the MRF save path; the MRF
+        // replay re-enters the queue, so the claim is released here too.
+        let (tx, _rx) = mpsc::channel(1);
+        pool.workers.write().await.push(tx);
+        assert_eq!(pool.queue_replica_task(ri.clone()).await, ReplicationQueueAdmission::Queued);
+        let overflow = ReplicateObjectInfo {
+            version_id: Some(uuid::Uuid::new_v4()),
+            ..ri
+        };
+        assert_eq!(pool.queue_replica_task(overflow).await, ReplicationQueueAdmission::Queued);
+        assert_eq!(pool.in_flight.len(), 1, "only the version held by the worker channel stays in flight");
+    }
+
+    #[test]
+    fn in_flight_guard_releases_the_version_on_drop() {
+        let in_flight = Arc::new(ReplicationInFlight::default());
+        let ri = ReplicateObjectInfo {
+            bucket: "guard-bucket".to_string(),
+            name: "object".to_string(),
+            version_id: Some(uuid::Uuid::new_v4()),
+            ..Default::default()
+        };
+        assert!(in_flight.try_begin(&ri));
+        assert!(!in_flight.try_begin(&ri));
+        {
+            let _guard = ReplicationInFlightGuard::new(in_flight.clone(), &ri);
+            assert_eq!(in_flight.len(), 1);
+        }
+        assert_eq!(in_flight.len(), 0);
+        assert!(in_flight.try_begin(&ri));
+    }
+
     #[tokio::test]
     async fn regular_worker_admission_counts_target_backlog_before_receive() {
         let pool = new_test_replication_pool(Arc::new(LoadResyncNodeStore::new("node-a", empty_resync_shared_state()))).await;

crates/ecstore/src/bucket/replication/replication_target_boundary.rs

@@ -38,7 +38,7 @@ use time::format_description::well_known::Rfc3339;
 
 pub(crate) use crate::bucket::bucket_target_sys::{
     AdvancedPutOptions, HeadObjectSdkError, PutObjectOptions, PutObjectPartOptions, RemotePutObjectResponse, RemoveObjectOptions,
-    S3ClientError, TargetClient, resolve_read_api_version_id,
+    ReplicaLocation, S3ClientError, TargetClient, resolve_read_api_version_id,
 };
 #[cfg(test)]
 pub(crate) use crate::bucket::target::BucketTarget;
@@ -48,6 +48,7 @@ pub use rustfs_replication::{ObjectLockIntegrity, object_lock_put_integrity};
 pub(crate) use rustfs_replication::{
     SsecPassthroughGate, is_replication_target_offline_error, ssec_passthrough_gate, version_identity_drifted,
 };
+pub use rustfs_replication::{VersionIdentityCapability, version_identity_capability_from_put};
 
 use super::replication_config_store::ReplicationConfigStore;
 use super::replication_error_boundary::{Error, Result};
@@ -192,6 +193,14 @@ impl ReplicationTargetStore {
             .await
     }
 
+    pub(crate) fn version_identity_capability(arn: &str) -> VersionIdentityCapability {
+        BucketTargetSys::get().version_identity_capability(arn)
+    }
+
+    pub(crate) fn record_version_identity_capability(arn: &str, capability: VersionIdentityCapability) {
+        BucketTargetSys::get().record_version_identity_capability(arn, capability)
+    }
+
     #[cfg(test)]
     pub(crate) async fn register_test_target(target_client: &Arc<TargetClient>) {
         BucketTargetSys::get().arn_remotes_map.write().await.insert(
@@ -238,6 +247,23 @@ pub(crate) fn replication_put_object_options(sc: &str, object_info: &ObjectInfo)
         meta.insert(key.to_string(), value.to_string());
     }
 
+    // A compressed SSE-C object passes through as its stored bytes. The target
+    // cannot infer the compression layout from ciphertext, so the scheme and
+    // the plaintext size travel as transport headers; each UploadPart carries
+    // its own plaintext length (backlog#2363).
+    if is_ssec && let Some(scheme) = get_str(&object_info.user_defined, rustfs_utils::http::SUFFIX_COMPRESSION) {
+        insert_header_map(&mut meta, rustfs_utils::http::SUFFIX_REPLICATION_COMPRESSION, scheme);
+        if let Ok(actual_size) = object_info.get_actual_size()
+            && actual_size >= 0
+        {
+            insert_header_map(
+                &mut meta,
+                rustfs_utils::http::SUFFIX_REPLICATION_COMPRESSION_ACTUAL_SIZE,
+                actual_size.to_string(),
+            );
+        }
+    }
+
     // Managed SSE replicates as plaintext (the replication reader decrypts via
     // the object-encryption resolver) and re-encrypts on the target with the
     // target's own KMS. Send only the encryption intent — never the source
@@ -248,7 +274,16 @@ pub(crate) fn replication_put_object_options(sc: &str, object_info: &ObjectInfo)
         meta.insert(AMZ_SERVER_SIDE_ENCRYPTION.to_string(), "aws:kms".to_string());
     }
 
-    let mut is_multipart = object_info.is_multipart();
+    // Older transformed objects can have physical parts without logical part
+    // lengths. Keep their existing whole-object transport: physical sizes are
+    // not plaintext boundaries for a multipart replication read.
+    let legacy_single_put = object_info.etag.as_deref().is_none_or(|etag| etag.len() == 32);
+    let base_is_multipart = object_info.is_multipart()
+        && !(legacy_single_put
+            && object_info.parts.len() > 1
+            && (object_info.is_compressed() || object_info.is_encrypted())
+            && object_info.parts.iter().any(|part| part.actual_size <= 0));
+    let mut is_multipart = base_is_multipart;
 
     if let Some(checksum_data) = &object_info.checksum
         && !checksum_data.is_empty()
@@ -259,8 +294,8 @@ pub(crate) fn replication_put_object_options(sc: &str, object_info: &ObjectInfo)
         } else if object_info.is_encrypted() {
             // Encrypted checksums cannot be exposed as plaintext headers, and
             // decrypt_checksums reports is_multipart=false for them (a value
-            // the response path relies on). Keep the object's own multipart
+            // the response path relies on). Keep the transport selected from
-            // flag so encrypted objects stay on the multipart route.
+            // the object's layout and readable part boundaries.
         } else {
             let (checksum_meta, checksum_record_is_multipart) = object_info.decrypt_checksums(0, &HeaderMap::new())?;
             // The checksum record describes how the *checksum* is composed,
@@ -268,23 +303,37 @@ pub(crate) fn replication_put_object_options(sc: &str, object_info: &ObjectInfo)
             // MULTIPART flag even on a multipart upload, so trusting it here
             // routed a 768-part object through a single PutObject and the
             // target rejected the 6 GiB body with EntityTooLarge
-            // (rustfs#6825). The object's own shape is the authority: the
+            // (rustfs#6825). The usable part layout is the authority: the
             // record may only add multipart-ness, never take it away.
-            is_multipart = object_info.is_multipart() || checksum_record_is_multipart;
+            is_multipart = base_is_multipart || checksum_record_is_multipart;
 
-            for (key, value) in checksum_meta.iter() {
+            if !base_is_multipart
-                if key != AMZ_CHECKSUM_TYPE {
-                    meta.insert(key.clone(), value.clone());
-                }
-            }
-
-            if !object_info.is_multipart()
                 && checksum_meta
                     .get(AMZ_CHECKSUM_TYPE)
                     .is_some_and(|value| value == AMZ_CHECKSUM_TYPE_FULL_OBJECT)
             {
                 is_multipart = false;
             }
+
+            // The record keys each checksum by algorithm name ("CRC32"); the
+            // target only reads `x-amz-checksum-<algorithm>`. Inserting the bare
+            // name here made `PutObjectOptions::header()` send it as user
+            // metadata (`x-amz-meta-crc32`), so no replica ever carried the
+            // source checksum (rustfs/backlog#2340). The object-level record
+            // describes one PUT body: a multipart replica is rebuilt part by
+            // part, and its CreateMultipartUpload must not announce a checksum
+            // the parts do not carry, so the record is forwarded on the
+            // single-PUT route only (MinIO `getCRCMeta` parity).
+            if !is_multipart {
+                for (key, value) in checksum_meta.iter() {
+                    if key == AMZ_CHECKSUM_TYPE {
+                        continue;
+                    }
+                    if let Some(header) = rustfs_rio::ChecksumType::from_string(key).key() {
+                        meta.insert(header.to_string(), value.clone());
+                    }
+                }
+            }
         }
     }
 
@@ -516,6 +565,7 @@ fn is_standard_header(key: &str) -> bool {
 
 #[cfg(test)]
 mod tests {
+    use super::super::replication_filemeta_boundary::ObjectPartInfo;
     use super::*;
     use aws_smithy_types::DateTime;
     use rustfs_replication::content_matches_by_etag;
@@ -550,6 +600,162 @@ mod tests {
         checksum.to_bytes(&combined)
     }
 
+    fn replication_route_metadata() -> [(&'static str, Arc<HashMap<String, String>>); 4] {
+        let mut compressed = HashMap::new();
+        rustfs_utils::http::insert_str(&mut compressed, rustfs_utils::http::SUFFIX_COMPRESSION, "zstd".to_string());
+        [
+            ("plain", Arc::new(HashMap::new())),
+            ("compressed", Arc::new(compressed)),
+            (
+                "encrypted",
+                Arc::new(HashMap::from([(AMZ_SERVER_SIDE_ENCRYPTION.to_string(), "AES256".to_string())])),
+            ),
+            (
+                "ssec",
+                Arc::new(HashMap::from([(SSEC_ALGORITHM_HEADER.to_string(), "AES256".to_string())])),
+            ),
+        ]
+    }
+
+    fn replication_route_object(
+        etag: Option<&str>,
+        actual_sizes: [i64; 3],
+        metadata: Arc<HashMap<String, String>>,
+    ) -> ObjectInfo {
+        ObjectInfo {
+            etag: etag.map(str::to_string),
+            size: 48,
+            actual_size: 12,
+            user_defined: metadata,
+            parts: Arc::new(
+                actual_sizes
+                    .into_iter()
+                    .enumerate()
+                    .map(|(index, actual_size)| ObjectPartInfo {
+                        number: index + 1,
+                        size: 16,
+                        actual_size,
+                        ..Default::default()
+                    })
+                    .collect(),
+            ),
+            ..Default::default()
+        }
+    }
+
+    #[test]
+    fn compressed_ssec_objects_declare_their_compression_layout_on_the_wire() {
+        use rustfs_utils::http::{
+            SUFFIX_ACTUAL_SIZE, SUFFIX_COMPRESSION, SUFFIX_REPLICATION_COMPRESSION, SUFFIX_REPLICATION_COMPRESSION_ACTUAL_SIZE,
+            insert_str,
+        };
+
+        let mut ssec_compressed = HashMap::from([(SSEC_ALGORITHM_HEADER.to_string(), "AES256".to_string())]);
+        insert_str(&mut ssec_compressed, SUFFIX_COMPRESSION, "klauspost/compress/s2".to_string());
+        insert_str(&mut ssec_compressed, SUFFIX_ACTUAL_SIZE, "6295552".to_string());
+        let object_info = ObjectInfo {
+            etag: Some("0123456789abcdef0123456789abcdef-2".to_string()),
+            size: 4321,
+            actual_size: 6295552,
+            user_defined: Arc::new(ssec_compressed),
+            ..Default::default()
+        };
+
+        // SSE-C passthrough sends stored bytes: the scheme and the plaintext
+        // size travel as transport headers, never as the internal key
+        // (backlog#2363).
+        let (options, _) = replication_put_object_options("STANDARD", &object_info).expect("ssec put options");
+        assert_eq!(
+            get_header_map(&options.user_metadata, SUFFIX_REPLICATION_COMPRESSION).as_deref(),
+            Some("klauspost/compress/s2")
+        );
+        assert_eq!(
+            get_header_map(&options.user_metadata, SUFFIX_REPLICATION_COMPRESSION_ACTUAL_SIZE).as_deref(),
+            Some("6295552")
+        );
+        assert!(
+            !options
+                .user_metadata
+                .keys()
+                .any(|key| rustfs_utils::http::is_internal_key(key)),
+            "internal metadata never leaves the source as plain metadata: {:?}",
+            options.user_metadata
+        );
+
+        // A compressed object that is not SSE-C is decompressed by the
+        // replication reader and travels as plaintext: no layout headers.
+        let mut plain_compressed = HashMap::new();
+        insert_str(&mut plain_compressed, SUFFIX_COMPRESSION, "klauspost/compress/s2".to_string());
+        insert_str(&mut plain_compressed, SUFFIX_ACTUAL_SIZE, "6295552".to_string());
+        let plain = ObjectInfo {
+            user_defined: Arc::new(plain_compressed),
+            ..object_info
+        };
+        let (options, _) = replication_put_object_options("STANDARD", &plain).expect("plain put options");
+        assert!(get_header_map(&options.user_metadata, SUFFIX_REPLICATION_COMPRESSION).is_none());
+        assert!(get_header_map(&options.user_metadata, SUFFIX_REPLICATION_COMPRESSION_ACTUAL_SIZE).is_none());
+    }
+
+    #[test]
+    fn legacy_transformed_single_put_parts_keep_the_previous_replication_route() {
+        let [_, (_, compressed), (_, encrypted), (_, ssec)] = replication_route_metadata();
+        let cases = [
+            (
+                "compressed middle zero",
+                compressed.clone(),
+                Some("0123456789abcdef0123456789abcdef"),
+                [4, 0, 4],
+            ),
+            ("compressed tail unknown", compressed, None, [4, 4, -1]),
+            (
+                "encrypted middle unknown",
+                encrypted,
+                Some("gggggggggggggggggggggggggggggggg"),
+                [4, -1, 4],
+            ),
+            ("ssec tail zero", ssec.clone(), None, [4, 4, 0]),
+            ("ssec middle unknown", ssec, Some("gggggggggggggggggggggggggggggggg"), [4, -1, 4]),
+        ];
+        for (name, metadata, etag, actual_sizes) in cases {
+            for checksum in [None, Some(full_object_multipart_checksum_record())] {
+                let mut object_info = replication_route_object(etag, actual_sizes, metadata.clone());
+                object_info.checksum = checksum;
+                assert!(object_info.is_multipart(), "{name}: physical parts remain visible to metadata APIs");
+                assert!(object_info.is_compressed() || object_info.is_encrypted());
+
+                let (options, is_multipart) =
+                    replication_put_object_options("STANDARD", &object_info).expect("legacy transformed put options");
+                assert!(
+                    !is_multipart,
+                    "{name}: unknown logical part sizes must preserve the old whole-object route"
+                );
+                assert_eq!(options.internal.source_etag, etag.unwrap_or_default());
+                if metadata.contains_key(SSEC_ALGORITHM_HEADER) {
+                    assert_eq!(
+                        get_header_map(&options.user_metadata, SUFFIX_REPLICATION_SSEC_CRC).is_some(),
+                        object_info.checksum.is_some(),
+                        "SSE-C checksums retain their raw passthrough transport"
+                    );
+                }
+            }
+        }
+    }
+
+    #[test]
+    fn positive_part_sizes_and_legacy_multipart_etags_keep_the_replication_route() {
+        for (name, metadata) in replication_route_metadata() {
+            for (etag, actual_sizes) in [
+                ("0123456789abcdef0123456789abcdef", [4, 4, 4]),
+                ("0123456789abcdef0123456789abcdef-3", [4, 0, -1]),
+            ] {
+                let mut object_info = replication_route_object(Some(etag), actual_sizes, metadata.clone());
+                object_info.checksum = Some(full_object_multipart_checksum_record());
+                let (_, is_multipart) = replication_put_object_options("STANDARD", &object_info).expect("multipart put options");
+                assert!(is_multipart, "{name}/{etag}: usable sizes and old multipart ETags must retain MPU");
+            }
+        }
+    }
+
     #[test]
     fn multipart_object_with_full_object_checksum_keeps_the_multipart_route() {
         // rustfs#6825: a 768-part upload was replicated with a single
@@ -582,6 +788,36 @@ mod tests {
         );
     }
 
+    #[test]
+    fn stored_multipart_parts_keep_the_replication_route_without_a_multipart_etag() {
+        for etag in [Some("0123456789abcdef0123456789abcdef"), None] {
+            for checksum in [None, Some(full_object_multipart_checksum_record())] {
+                let object_info = ObjectInfo {
+                    etag: etag.map(str::to_string),
+                    checksum,
+                    parts: Arc::new(
+                        (1..=2)
+                            .map(|number| ObjectPartInfo {
+                                number,
+                                ..Default::default()
+                            })
+                            .collect(),
+                    ),
+                    ..Default::default()
+                };
+                let (options, is_multipart) =
+                    replication_put_object_options("STANDARD", &object_info).expect("build put options");
+
+                assert!(
+                    is_multipart,
+                    "stored parts must retain multipart routing: etag={etag:?}, checksum={:?}",
+                    object_info.checksum
+                );
+                assert_eq!(options.internal.source_etag, etag.unwrap_or_default());
+            }
+        }
+    }
+
     #[test]
     fn checksum_record_never_changes_the_transport_a_single_part_object_needs() {
         // The mirror of the rustfs#6825 guard: an object stored as one PUT
@@ -592,6 +828,10 @@ mod tests {
         let object_info = ObjectInfo {
             etag: Some("0123456789abcdef0123456789abcdef".to_string()),
             checksum: Some(checksum.to_bytes(&[])),
+            parts: Arc::new(vec![ObjectPartInfo {
+                number: 1,
+                ..Default::default()
+            }]),
             ..Default::default()
         };
 
@@ -628,6 +868,19 @@ mod tests {
         let (_, is_multipart) = replication_put_object_options("STANDARD", &object_info).expect("build put options");
 
         assert!(is_multipart, "a composite-checksum multipart object must stay on the multipart transport");
+
+        for (name, metadata) in replication_route_metadata() {
+            let mut legacy = replication_route_object(Some("0123456789abcdef0123456789abcdef"), [4, 0, 4], metadata);
+            legacy.checksum = Some(checksum.to_bytes(&combined));
+            let (_, record_is_multipart) = legacy.decrypt_checksums(0, &HeaderMap::new()).expect("decode checksum");
+            let (_, is_multipart) = replication_put_object_options("STANDARD", &legacy).expect("legacy checksum put options");
+            if legacy.is_encrypted() {
+                assert!(!is_multipart, "{name}: encrypted checksum records must not change the old transport");
+            } else {
+                assert!(record_is_multipart, "the composite checksum must carry its own multipart signal");
+                assert!(is_multipart, "{name}: a composite record can still promote the legacy route to MPU");
+            }
+        }
     }
 
     #[test]
@@ -1308,12 +1561,63 @@ mod tests {
                 ..Default::default()
             };
 
-            let (opts, _is_multipart) = replication_put_object_options("", &object_info).expect("build replication put options");
+            let (opts, is_multipart) = replication_put_object_options("", &object_info).expect("build replication put options");
 
+            assert!(!is_multipart, "{name}: a single-part checksum record must keep the single-PUT route");
+            let header = ty.key().expect("every forwarded algorithm has an x-amz-checksum header");
             assert_eq!(
-                opts.user_metadata.get(name),
+                opts.user_metadata.get(header),
                 Some(&checksum.encoded),
-                "replication must forward the {name} checksum into user_metadata identically to the classic algorithms"
+                "replication must forward the {name} checksum as the {header} header"
+            );
+            assert!(
+                !opts.user_metadata.contains_key(name),
+                "{name}: the bare algorithm name would leave as x-amz-meta user metadata"
+            );
+        }
+    }
+
+    /// The object-level record of a multipart upload (composite or full-object)
+    /// must not become a PutObject checksum header: the replica is rebuilt
+    /// through CreateMultipartUpload/UploadPart, and a checksum announced there
+    /// that the parts do not carry would be rejected by the target.
+    #[test]
+    fn replication_put_object_options_keeps_multipart_checksum_records_off_the_wire() {
+        let mut composite_type = rustfs_rio::ChecksumType::from_string("crc32");
+        composite_type
+            .merge(rustfs_rio::ChecksumType::MULTIPART)
+            .merge(rustfs_rio::ChecksumType::INCLUDES_MULTIPART);
+        let mut combined = Vec::new();
+        for part in [b"part-one".as_slice(), b"part-two".as_slice()] {
+            let part_checksum =
+                rustfs_rio::Checksum::new_from_data(rustfs_rio::ChecksumType::from_string("crc32"), part).expect("part checksum");
+            combined.extend_from_slice(part_checksum.raw.as_slice());
+        }
+        let composite = rustfs_rio::Checksum::new_from_data(composite_type, &combined)
+            .expect("composite checksum")
+            .to_bytes(&combined);
+
+        for (label, checksum, etag) in [
+            ("composite", composite, "0123456789abcdef0123456789abcdef-2"),
+            (
+                "full-object",
+                full_object_multipart_checksum_record(),
+                "0123456789abcdef0123456789abcdef-3",
+            ),
+        ] {
+            let object_info = ObjectInfo {
+                etag: Some(etag.to_string()),
+                checksum: Some(checksum),
+                ..Default::default()
+            };
+            let (opts, is_multipart) = replication_put_object_options("", &object_info).expect("build replication put options");
+            assert!(is_multipart, "{label}: a multipart object must keep the multipart route");
+            assert!(
+                opts.user_metadata
+                    .keys()
+                    .all(|key| !key.starts_with("x-amz-checksum-") && key != "CRC32"),
+                "{label}: no object-level checksum may reach the target's CreateMultipartUpload: {:?}",
+                opts.user_metadata
             );
         }
     }

crates/ecstore/src/cluster/rpc/mod.rs

@@ -48,7 +48,8 @@ pub use internode_data_transport::build_internode_data_transport_from_env;
 pub(crate) use peer_rest_client::TierConfigReloadOutcome;
 pub use peer_rest_client::{
     KMS_SIGNAL_SUBSYSTEM, PEER_RESTDRY_RUN, PEER_RESTSIGNAL, PEER_RESTSUB_SYS, PeerRestClient, SERVICE_SIGNAL_REFRESH_CONFIG,
-    SERVICE_SIGNAL_RELOAD_DYNAMIC, ScannerPeerActivity, ScannerPeerDirtyUsageSnapshot, ScannerPublicationLease,
+    SERVICE_SIGNAL_RELOAD_DYNAMIC, ScannerDirtyUsageAcknowledgement, ScannerPeerActivity, ScannerPeerDirtyUsageBucket,
+    ScannerPeerDirtyUsageSnapshot, ScannerPublicationLease, ScannerScopedDirtyUsageAckEntry,
 };
 pub(crate) use peer_s3_client::heal_bucket_local_on_disks;
 pub use peer_s3_client::{

crates/ecstore/src/cluster/rpc/peer_rest_client.rs

@@ -49,10 +49,11 @@ use rustfs_protos::proto_gen::node_service::{
     LoadRebalanceMetaRequest, LoadServiceAccountRequest, LoadTransitionTierConfigRequest, LoadUserRequest,
     LocalStorageInfoRequest, Mss, ReloadPoolMetaRequest, ReloadSiteReplicationConfigRequest, ReplacementRecoveryStatusRequest,
     ScannerActivityRequest, ScannerActivityResponse, ScannerDirtyUsageSnapshotRequest, ScannerDirtyUsageSnapshotResponse,
-    ScannerPublicationLeaseReleaseRequest, ScannerPublicationLeaseRequest, ScannerPublicationLeaseResponse, ServerInfoRequest,
+    ScannerPublicationLeaseReleaseRequest, ScannerPublicationLeaseRequest, ScannerPublicationLeaseResponse,
-    SignalServiceRequest, SignalServiceResponse, StartDecommissionRequest, StartProfilingRequest, StopRebalanceRequest,
+    ScannerScopedDirtyUsageAckRequest, ScannerScopedDirtyUsageEntry, ServerInfoRequest, SignalServiceRequest,
-    TierDailyStatsRequest, TierMutationAbortRequest, TierMutationCommitRequest, TierMutationControlResponse,
+    SignalServiceResponse, StartDecommissionRequest, StartProfilingRequest, StopRebalanceRequest, TierDailyStatsRequest,
-    TierMutationFailureClass, TierMutationPeerState, TierMutationPrepareRequest, node_service_client::NodeServiceClient,
+    TierMutationAbortRequest, TierMutationCommitRequest, TierMutationControlResponse, TierMutationFailureClass,
+    TierMutationPeerState, TierMutationPrepareRequest, node_service_client::NodeServiceClient,
     tier_mutation_control_service_client::TierMutationControlServiceClient,
 };
 pub use rustfs_protos::{PEER_RESTDRY_RUN, PEER_RESTSIGNAL, PEER_RESTSUB_SYS};
@@ -92,6 +93,7 @@ const HEAL_CONTROL_PAYLOAD_MAX_SIZE: usize = 64 * 1024;
 const PEER_REST_RECOVERY_MAX_ATTEMPTS: u32 = 60;
 const PEER_REST_RECOVERY_MAX_BACKOFF: Duration = Duration::from_secs(30);
 const SCANNER_ACTIVITY_MAX_MESSAGE_SIZE: usize = 1024;
+const SCANNER_SCOPED_DIRTY_USAGE_STAGE_TIMEOUT: Duration = Duration::from_secs(5);
 /// Reserve time for the acquire response's network/clock uncertainty.  The
 /// server owns the real expiry; this local deadline is intentionally earlier
 /// so a coordinator never starts a bounded persistence operation at the edge
@@ -192,12 +194,102 @@ pub struct ScannerPeerActivity {
 
 #[derive(Clone, Debug, PartialEq, Eq)]
 pub struct ScannerPeerDirtyUsageSnapshot {
+    pub owner_id: String,
     pub instance_id: String,
     pub generation: u64,
     pub pending_bucket_count: u64,
     pub protocol_version: u32,
     pub complete: bool,
-    pub buckets: BTreeMap<String, u64>,
+    pub buckets: BTreeMap<String, ScannerPeerDirtyUsageBucket>,
+}
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+pub struct ScannerPeerDirtyUsageBucket {
+    pub bucket_incarnation: Uuid,
+    pub generation: u64,
+}
+
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub struct ScannerScopedDirtyUsageAckEntry {
+    pub bucket: String,
+    pub bucket_incarnation: Uuid,
+    pub generation: u64,
+}
+
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub enum ScannerDirtyUsageAcknowledgement {
+    Generation {
+        host: String,
+        instance_id: String,
+        generation: u64,
+    },
+    Scoped {
+        host: String,
+        owner_id: String,
+        instance_id: String,
+        entries: Vec<ScannerScopedDirtyUsageAckEntry>,
+    },
+}
+
+fn scanner_scoped_dirty_usage_ack_payloads(
+    owner_id: String,
+    instance_id: String,
+    probe_only: bool,
+    entries: Vec<ScannerScopedDirtyUsageAckEntry>,
+) -> Result<Vec<ScannerScopedDirtyUsageAckRequest>> {
+    use rustfs_protos::scoped_dirty_usage::*;
+
+    if entries.is_empty() {
+        return Err(Error::other("scoped dirty usage acknowledgement entries must be nonempty"));
+    }
+
+    let mut payloads = Vec::with_capacity(entries.len().div_ceil(SCOPED_DIRTY_USAGE_MAX_ENTRIES as usize));
+    let mut batch = Vec::with_capacity(SCOPED_DIRTY_USAGE_MAX_ENTRIES as usize);
+    for entry in entries {
+        batch.push(ScannerScopedDirtyUsageEntry {
+            bucket: entry.bucket,
+            bucket_incarnation: entry.bucket_incarnation.as_bytes().to_vec().into(),
+            generation: entry.generation,
+        });
+        if batch.len() == SCOPED_DIRTY_USAGE_MAX_ENTRIES as usize {
+            payloads.push(scanner_scoped_dirty_usage_ack_payload(
+                &owner_id,
+                &instance_id,
+                probe_only,
+                std::mem::take(&mut batch),
+            )?);
+        }
+    }
+    if !batch.is_empty() {
+        payloads.push(scanner_scoped_dirty_usage_ack_payload(&owner_id, &instance_id, probe_only, batch)?);
+    }
+
+    Ok(payloads)
+}
+
+fn scanner_scoped_dirty_usage_ack_payload(
+    owner_id: &str,
+    instance_id: &str,
+    probe_only: bool,
+    entries: Vec<ScannerScopedDirtyUsageEntry>,
+) -> Result<ScannerScopedDirtyUsageAckRequest> {
+    use rustfs_protos::scoped_dirty_usage::*;
+
+    let payload = ScannerScopedDirtyUsageAckRequest {
+        challenge: Uuid::new_v4().as_bytes().to_vec().into(),
+        protocol_version: SCOPED_DIRTY_USAGE_PROTOCOL_VERSION,
+        owner_id: owner_id.to_string(),
+        instance_id: instance_id.to_string(),
+        scope: SCOPED_DIRTY_USAGE_BUCKET_SCOPE,
+        probe_only,
+        entries,
+    };
+    canonical_scoped_dirty_usage_request(&payload).map_err(|err| Error::other(err.to_string()))?;
+    Ok(payload)
+}
+
+fn scanner_scoped_dirty_usage_ack_reconciled(activity: &ScannerPeerActivity, expected_instance_id: &str) -> bool {
+    activity.instance_id == expected_instance_id && activity.dirty_usage_pending == Some(false)
 }
 
 fn scanner_instance_id_is_valid(instance_id: &str) -> bool {
@@ -351,6 +443,11 @@ fn decode_scanner_dirty_usage_snapshot_with_verifier(
     if !scanner_instance_id_is_valid(&response.instance_id) {
         return Err(Error::other("peer returned an invalid scanner dirty usage snapshot instance ID"));
     }
+    let owner_id = Uuid::parse_str(&response.owner_id)
+        .ok()
+        .filter(|owner_id| !owner_id.is_nil())
+        .map(|owner_id| owner_id.to_string())
+        .ok_or_else(|| Error::other("peer returned an invalid scanner dirty usage snapshot owner"))?;
     if response.generation == u64::MAX {
         return Err(Error::other("peer scanner dirty usage snapshot exhausted its generation"));
     }
@@ -386,9 +483,14 @@ fn decode_scanner_dirty_usage_snapshot_with_verifier(
         if bucket.generation == 0 || bucket.generation > response.generation {
             return Err(Error::other("peer scanner dirty usage snapshot contains an invalid bucket generation"));
         }
+        Uuid::from_slice(bucket.bucket_incarnation.as_ref())
+            .ok()
+            .filter(|bucket_incarnation| !bucket_incarnation.is_nil())
+            .ok_or_else(|| Error::other("peer scanner dirty usage snapshot contains an invalid bucket incarnation"))?;
     }
 
     Ok(ScannerPeerDirtyUsageSnapshot {
+        owner_id,
         instance_id: response.instance_id,
         generation: response.generation,
         pending_bucket_count: response.pending_bucket_count,
@@ -397,7 +499,16 @@ fn decode_scanner_dirty_usage_snapshot_with_verifier(
         buckets: response
             .buckets
             .into_iter()
-            .map(|bucket| (bucket.bucket, bucket.generation))
+            .map(|bucket| {
+                (
+                    bucket.bucket,
+                    ScannerPeerDirtyUsageBucket {
+                        bucket_incarnation: Uuid::from_slice(bucket.bucket_incarnation.as_ref())
+                            .expect("bucket incarnation was validated"),
+                        generation: bucket.generation,
+                    },
+                )
+            })
             .collect(),
     })
 }
@@ -1688,6 +1799,24 @@ impl PeerRestClient {
         Ok((self.topology_member.clone(), supported_version, epoch))
     }
 
+    pub async fn probe_ilm_recovery_export(&self, topology_fingerprint: String) -> Result<(String, Uuid)> {
+        let probe = rustfs_protos::ilm_recovery_export_capability_probe(Uuid::new_v4().as_bytes());
+        let result = self
+            .heal_control(rustfs_protos::HEAL_CONTROL_PROTOCOL_VERSION, topology_fingerprint, probe)
+            .await?;
+        let epoch = decode_remote_version_state_capability(&self.topology_member, &result)?;
+        Ok((self.topology_member.clone(), epoch))
+    }
+
+    pub async fn probe_transition_transaction_compaction(&self, topology_fingerprint: String) -> Result<(String, Uuid)> {
+        let probe = rustfs_protos::transition_transaction_compaction_capability_probe(Uuid::new_v4().as_bytes());
+        let result = self
+            .heal_control(rustfs_protos::HEAL_CONTROL_PROTOCOL_VERSION, topology_fingerprint, probe)
+            .await?;
+        let epoch = decode_remote_version_state_capability(&self.topology_member, &result)?;
+        Ok((self.topology_member.clone(), epoch))
+    }
+
     pub async fn load_bucket_metadata(&self, bucket: &str, scanner_maintenance_change: bool) -> Result<()> {
         let result = tokio::time::timeout(BUCKET_METADATA_RELOAD_TIMEOUT, async {
             let result = self.load_bucket_metadata_once(bucket, scanner_maintenance_change).await;
@@ -2068,19 +2197,10 @@ impl PeerRestClient {
         &self,
         owner_id: String,
         instance_id: String,
-        entries: Vec<rustfs_protos::proto_gen::node_service::ScannerScopedDirtyUsageEntry>,
+        entries: Vec<ScannerScopedDirtyUsageAckEntry>,
     ) -> Result<bool> {
         use rustfs_protos::scoped_dirty_usage::*;
-        let payload = rustfs_protos::proto_gen::node_service::ScannerScopedDirtyUsageAckRequest {
+        let payloads = scanner_scoped_dirty_usage_ack_payloads(owner_id, instance_id, true, entries)?;
-            challenge: Uuid::new_v4().as_bytes().to_vec().into(),
-            protocol_version: SCOPED_DIRTY_USAGE_PROTOCOL_VERSION,
-            owner_id,
-            instance_id,
-            scope: SCOPED_DIRTY_USAGE_BUCKET_SCOPE,
-            probe_only: true,
-            entries,
-        };
-        let canonical = canonical_scoped_dirty_usage_request(&payload).map_err(|err| Error::other(err.to_string()))?;
         self.finalize_result(
             async {
                 let mut client = super::client::scanner_control_time_out_client(
@@ -2088,26 +2208,106 @@ impl PeerRestClient {
                     TonicInterceptor::Signature(gen_tonic_signature_interceptor()),
                 )
                 .await?;
+                for payload in payloads {
+                    let canonical =
+                        canonical_scoped_dirty_usage_request(&payload).map_err(|err| Error::other(err.to_string()))?;
+                    let mut request = Request::new(payload.clone());
+                    set_tonic_canonical_body_digest(&mut request, &canonical)?;
+                    let response = client.scanner_scoped_dirty_usage_ack(request).await?.into_inner();
+                    let body = canonical_scoped_dirty_usage_response(&canonical, &response)
+                        .map_err(|_| Error::other("scoped dirty usage capability response is too large"))?;
+                    verify_tonic_rpc_response_proof(&body, response.response_proof.as_ref())?;
+                    if response.protocol_version != SCOPED_DIRTY_USAGE_PROTOCOL_VERSION
+                        || response.owner_id != payload.owner_id
+                        || response.instance_id != payload.instance_id
+                        || response.max_entries != SCOPED_DIRTY_USAGE_MAX_ENTRIES
+                        || response.max_request_bytes != SCOPED_DIRTY_USAGE_MAX_REQUEST_BYTES
+                        || response.cleared != 0
+                    {
+                        return Err(Error::other("scoped dirty usage capability response does not match request"));
+                    }
+                    if !response.supported {
+                        return Ok(false);
+                    }
+                }
+                Ok(true)
+            }
+            .await,
+        )
+        .await
+    }
+
+    pub async fn acknowledge_scanner_scoped_dirty_usage(
+        &self,
+        owner_id: String,
+        instance_id: String,
+        entries: Vec<ScannerScopedDirtyUsageAckEntry>,
+    ) -> Result<ScannerPeerActivity> {
+        use rustfs_protos::scoped_dirty_usage::*;
+        let payloads = scanner_scoped_dirty_usage_ack_payloads(owner_id, instance_id.clone(), false, entries)?;
+        let ack_attempt = async {
+            let mut client = super::client::scanner_control_time_out_client(
+                &self.grid_host,
+                TonicInterceptor::Signature(gen_tonic_signature_interceptor()),
+            )
+            .await?;
+            for payload in payloads {
+                let canonical = canonical_scoped_dirty_usage_request(&payload).map_err(|err| Error::other(err.to_string()))?;
                 let mut request = Request::new(payload.clone());
                 set_tonic_canonical_body_digest(&mut request, &canonical)?;
                 let response = client.scanner_scoped_dirty_usage_ack(request).await?.into_inner();
                 let body = canonical_scoped_dirty_usage_response(&canonical, &response)
-                    .map_err(|_| Error::other("scoped dirty usage capability response is too large"))?;
+                    .map_err(|_| Error::other("scoped dirty usage acknowledgement response is too large"))?;
                 verify_tonic_rpc_response_proof(&body, response.response_proof.as_ref())?;
                 if response.protocol_version != SCOPED_DIRTY_USAGE_PROTOCOL_VERSION
                     || response.owner_id != payload.owner_id
                     || response.instance_id != payload.instance_id
                     || response.max_entries != SCOPED_DIRTY_USAGE_MAX_ENTRIES
                     || response.max_request_bytes != SCOPED_DIRTY_USAGE_MAX_REQUEST_BYTES
-                    || response.cleared != 0
+                    || !response.supported
                 {
-                    return Err(Error::other("scoped dirty usage capability response does not match request"));
+                    return Err(Error::other("scoped dirty usage acknowledgement response does not match request"));
                 }
-                Ok(response.supported)
             }
-            .await,
+            Ok(())
-        )
+        };
-        .await
+        let result = match timeout(SCANNER_SCOPED_DIRTY_USAGE_STAGE_TIMEOUT, ack_attempt).await {
+            Ok(result) => self.finalize_result(result).await,
+            Err(_) => {
+                self.prepare_retry_with_timeout(SCANNER_SCOPED_DIRTY_USAGE_STAGE_TIMEOUT)
+                    .await;
+                Err(Error::other("scoped dirty usage acknowledgement deadline elapsed"))
+            }
+        };
+
+        match result {
+            Ok(()) => {
+                let activity = self.scanner_scoped_dirty_usage_activity_confirmation().await?;
+                if activity.instance_id == instance_id {
+                    Ok(activity)
+                } else {
+                    Err(Error::other(
+                        "scoped dirty usage acknowledgement peer restarted before activity confirmation",
+                    ))
+                }
+            }
+            Err(err) => {
+                if Self::is_network_like_error(&err) {
+                    self.prepare_retry_with_timeout(SCANNER_SCOPED_DIRTY_USAGE_STAGE_TIMEOUT)
+                        .await;
+                }
+                match self.scanner_scoped_dirty_usage_activity_confirmation().await {
+                    Ok(activity) if scanner_scoped_dirty_usage_ack_reconciled(&activity, &instance_id) => Ok(activity),
+                    _ => Err(err),
+                }
+            }
+        }
+    }
+
+    async fn scanner_scoped_dirty_usage_activity_confirmation(&self) -> Result<ScannerPeerActivity> {
+        timeout(SCANNER_SCOPED_DIRTY_USAGE_STAGE_TIMEOUT, self.scanner_activity())
+            .await
+            .map_err(|_| Error::other("scoped dirty usage activity confirmation timed out"))?
     }
 
     pub async fn acknowledge_scanner_dirty_usage(&self, instance_id: String, generation: u64) -> Result<ScannerPeerActivity> {
@@ -2836,16 +3036,79 @@ mod tests {
                 rustfs_protos::proto_gen::node_service::ScannerDirtyUsageBucket {
                     bucket: "archive".to_string(),
                     generation: 3,
+                    bucket_incarnation: Uuid::from_u128(0x11111111111111111111111111111111).as_bytes().to_vec().into(),
                 },
                 rustfs_protos::proto_gen::node_service::ScannerDirtyUsageBucket {
                     bucket: "photos".to_string(),
                     generation: 7,
+                    bucket_incarnation: Uuid::from_u128(0x22222222222222222222222222222222).as_bytes().to_vec().into(),
                 },
             ],
             response_proof: b"proof".to_vec().into(),
+            owner_id: "33333333-3333-3333-3333-333333333333".to_string(),
         }
     }
 
+    #[test]
+    fn scanner_scoped_dirty_usage_ack_payloads_split_at_protocol_limit() {
+        use rustfs_protos::scoped_dirty_usage::{SCOPED_DIRTY_USAGE_MAX_ENTRIES, canonical_scoped_dirty_usage_request};
+
+        let entries = (0..=SCOPED_DIRTY_USAGE_MAX_ENTRIES)
+            .map(|index| ScannerScopedDirtyUsageAckEntry {
+                bucket: format!("bucket-{index:02}"),
+                bucket_incarnation: Uuid::from_u128(0x11111111111111111111111111111111),
+                generation: 9,
+            })
+            .collect::<Vec<_>>();
+
+        let payloads = scanner_scoped_dirty_usage_ack_payloads(
+            "33333333-3333-3333-3333-333333333333".to_string(),
+            "0123456789abcdef0123456789abcdef".to_string(),
+            false,
+            entries,
+        )
+        .expect("33 entries should split into valid scoped dirty usage requests");
+
+        assert_eq!(payloads.len(), 2);
+        assert_eq!(payloads[0].entries.len(), SCOPED_DIRTY_USAGE_MAX_ENTRIES as usize);
+        assert_eq!(payloads[1].entries.len(), 1);
+        assert_eq!(payloads[0].entries.first().map(|entry| entry.bucket.as_str()), Some("bucket-00"));
+        assert_eq!(payloads[0].entries.last().map(|entry| entry.bucket.as_str()), Some("bucket-31"));
+        assert_eq!(payloads[1].entries.first().map(|entry| entry.bucket.as_str()), Some("bucket-32"));
+        for payload in payloads {
+            canonical_scoped_dirty_usage_request(&payload).expect("each split scoped ACK payload should be canonical");
+        }
+    }
+
+    #[test]
+    fn scanner_scoped_dirty_usage_ack_reconciliation_requires_same_clean_instance() {
+        let activity = |instance_id: &str, pending| ScannerPeerActivity {
+            instance_id: instance_id.to_string(),
+            namespace_generation: 1,
+            maintenance_generation: 1,
+            protocol_version: SCANNER_ACTIVITY_PROTOCOL_VERSION,
+            topology_digest: Some([1; 32]),
+            data_movement_active: Some(false),
+            dirty_usage_generation: Some(9),
+            dirty_usage_pending: pending,
+            movement_generation: Some(1),
+            publication_blocked: Some(false),
+        };
+
+        assert!(scanner_scoped_dirty_usage_ack_reconciled(
+            &activity("0123456789abcdef0123456789abcdef", Some(false)),
+            "0123456789abcdef0123456789abcdef"
+        ));
+        assert!(!scanner_scoped_dirty_usage_ack_reconciled(
+            &activity("0123456789abcdef0123456789abcdef", Some(true)),
+            "0123456789abcdef0123456789abcdef"
+        ));
+        assert!(!scanner_scoped_dirty_usage_ack_reconciled(
+            &activity("fedcba9876543210fedcba9876543210", Some(false)),
+            "0123456789abcdef0123456789abcdef"
+        ));
+    }
+
     #[test]
     fn scanner_dirty_usage_snapshot_requires_a_complete_authenticated_ordered_view() {
         let decoded = decode_test_scanner_dirty_usage_snapshot(test_scanner_dirty_usage_snapshot_response())
@@ -2854,9 +3117,18 @@ mod tests {
         assert_eq!(decoded.generation, 7);
         assert_eq!(decoded.pending_bucket_count, 2);
         assert_eq!(decoded.protocol_version, SCANNER_DIRTY_USAGE_SNAPSHOT_PROTOCOL_VERSION);
+        assert_eq!(decoded.owner_id, "33333333-3333-3333-3333-333333333333");
         assert!(decoded.complete);
-        assert_eq!(decoded.buckets.get("archive"), Some(&3));
+        assert_eq!(
-        assert_eq!(decoded.buckets.get("photos"), Some(&7));
+            decoded.buckets.get("archive").map(|bucket| bucket.bucket_incarnation),
+            Some(Uuid::from_u128(0x11111111111111111111111111111111))
+        );
+        assert_eq!(decoded.buckets.get("archive").map(|bucket| bucket.generation), Some(3));
+        assert_eq!(
+            decoded.buckets.get("photos").map(|bucket| bucket.bucket_incarnation),
+            Some(Uuid::from_u128(0x22222222222222222222222222222222))
+        );
+        assert_eq!(decoded.buckets.get("photos").map(|bucket| bucket.generation), Some(7));
 
         let overflow_count =
             u64::try_from(SCANNER_DIRTY_USAGE_SNAPSHOT_MAX_ENTRIES + 1).expect("the test snapshot entry limit should fit in u64");
@@ -2907,6 +3179,14 @@ mod tests {
         empty_bucket.buckets[0].bucket.clear();
         cases.push((empty_bucket, "empty bucket name"));
 
+        let mut invalid_owner = test_scanner_dirty_usage_snapshot_response();
+        invalid_owner.owner_id.clear();
+        cases.push((invalid_owner, "snapshot owner"));
+
+        let mut invalid_incarnation = test_scanner_dirty_usage_snapshot_response();
+        invalid_incarnation.buckets[0].bucket_incarnation = Uuid::nil().as_bytes().to_vec().into();
+        cases.push((invalid_incarnation, "bucket incarnation"));
+
         let mut partial = test_scanner_dirty_usage_snapshot_response();
         partial.complete = false;
         cases.push((partial, "entry-limit overflow"));
@@ -2919,6 +3199,7 @@ mod tests {
                 .map(|index| rustfs_protos::proto_gen::node_service::ScannerDirtyUsageBucket {
                     bucket: format!("bucket-{index:04}"),
                     generation: 1,
+                    bucket_incarnation: Uuid::from_u128(0x11111111111111111111111111111111).as_bytes().to_vec().into(),
                 })
                 .collect(),
             ..test_scanner_dirty_usage_snapshot_response()

crates/ecstore/src/cluster/rpc/remote_locker.rs

@@ -22,21 +22,152 @@ use rustfs_lock::{
     LockClient, LockError, LockInfo, LockRequest, LockResponse, LockStats, LockStatus, LockType, Result,
     types::{LockId, LockMetadata, LockPriority},
 };
-use rustfs_protos::proto_gen::node_service::{BatchGenerallyLockRequest, GenerallyLockRequest, PingRequest};
+use rustfs_protos::proto_gen::node_service::{
+    BatchGenerallyLockRequest, BatchGenerallyLockResponse, GenerallyLockRequest, GenerallyLockResponse, GenerallyLockResult,
+    PingRequest,
+};
 use rustfs_protos::{
     ConnectionEvictionLogLevel, evict_failed_connection_with_log_level, models::PingBodyBuilder,
     proto_gen::node_service::node_service_client::NodeServiceClient,
 };
-use std::{sync::OnceLock, time::Duration};
+use std::collections::HashMap;
-use tokio::time::timeout;
+use std::future::Future;
-use tonic::Request;
+use std::pin::Pin;
+use std::sync::{Mutex, OnceLock};
+use std::time::Duration;
+use tokio::task::JoinHandle;
+use tokio::time::{Instant, timeout};
 use tonic::service::interceptor::InterceptedService;
+use tonic::{Request, Response};
 use tracing::{debug, info, warn};
 
 fn attach_lock_mutation_body_digest<T: rustfs_protos::CanonicalMutationBody>(request: &mut Request<T>) -> std::io::Result<()> {
     set_tonic_rolling_mutation_body_digest(request)
 }
 
+/// Work to run if an RPC that already timed out for its caller completes later.
+type LateCompletion<T> = Option<Box<dyn FnOnce(T) -> Pin<Box<dyn Future<Output = ()> + Send>> + Send>>;
+
+/// The liveness window is this many RPC deadlines: a peer that completed a
+/// lock RPC within it is slow, not gone, and keeps its channel on a timeout.
+const LOCK_RPC_LIVENESS_WINDOW_DEADLINES: u32 = 2;
+
+/// Recent history of the shared lock channel to one peer (issue #7363).
+///
+/// A single request deadline says nothing about the HTTP/2 connection it ran
+/// on: a peer whose lock service is merely slow keeps answering other streams.
+/// Evicting the cached channel on every timeout turned that slowness into a
+/// `RST_STREAM`/`GOAWAY too_many_resets`/re-dial loop across the cluster, so
+/// eviction now requires the peer to have gone quiet and is rate limited.
+#[derive(Debug, Clone, Copy, Default)]
+struct LockPeerChannelHealth {
+    last_success: Option<Instant>,
+    last_eviction: Option<Instant>,
+    consecutive_timeouts: u32,
+    /// Timed-out RPCs still running in the background for this peer.
+    detached_rpcs: usize,
+}
+
+fn lock_peer_channel_health() -> &'static Mutex<HashMap<String, LockPeerChannelHealth>> {
+    static HEALTH: OnceLock<Mutex<HashMap<String, LockPeerChannelHealth>>> = OnceLock::new();
+    HEALTH.get_or_init(Mutex::default)
+}
+
+fn with_lock_peer_health<R>(addr: &str, update: impl FnOnce(&mut LockPeerChannelHealth) -> R) -> R {
+    let mut peers = lock_peer_channel_health()
+        .lock()
+        .unwrap_or_else(std::sync::PoisonError::into_inner);
+    update(peers.entry(addr.to_string()).or_default())
+}
+
+#[cfg(test)]
+fn lock_peer_health_for_test(addr: &str) -> LockPeerChannelHealth {
+    lock_peer_channel_health()
+        .lock()
+        .unwrap_or_else(std::sync::PoisonError::into_inner)
+        .get(addr)
+        .copied()
+        .unwrap_or_default()
+}
+
+#[cfg(test)]
+fn reset_lock_peer_health_for_test(addr: &str) {
+    lock_peer_channel_health()
+        .lock()
+        .unwrap_or_else(std::sync::PoisonError::into_inner)
+        .remove(addr);
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+enum EvictionTrigger {
+    /// The caller's deadline expired while the stream was still open.
+    Timeout,
+    /// The transport itself reported the failure (refused, reset, GOAWAY, ...).
+    Transport,
+}
+
+impl EvictionTrigger {
+    fn as_str(self) -> &'static str {
+        match self {
+            Self::Timeout => "timeout",
+            Self::Transport => "transport",
+        }
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+enum EvictionVerdict {
+    Evict,
+    /// The peer completed a lock RPC within the liveness window: slow, not gone.
+    PeerRecentlyServed,
+    /// The channel was re-dialed within the cooldown; let it prove itself first.
+    CoolingDown,
+}
+
+impl EvictionVerdict {
+    fn as_str(self) -> &'static str {
+        match self {
+            Self::Evict => "evict",
+            Self::PeerRecentlyServed => "peer_recently_served",
+            Self::CoolingDown => "cooling_down",
+        }
+    }
+}
+
+/// Decide whether a failed lock RPC may evict the shared channel to its peer.
+fn eviction_verdict(
+    health: &LockPeerChannelHealth,
+    now: Instant,
+    trigger: EvictionTrigger,
+    liveness_window: Duration,
+    cooldown: Duration,
+) -> EvictionVerdict {
+    if trigger == EvictionTrigger::Timeout
+        && health
+            .last_success
+            .is_some_and(|at| now.saturating_duration_since(at) < liveness_window)
+    {
+        return EvictionVerdict::PeerRecentlyServed;
+    }
+    if health
+        .last_eviction
+        .is_some_and(|at| now.saturating_duration_since(at) < cooldown)
+    {
+        return EvictionVerdict::CoolingDown;
+    }
+    EvictionVerdict::Evict
+}
+
+/// Lock ids whose batch entry the server reports as granted.
+fn acquired_lock_ids(lock_ids: &[LockId], results: &[GenerallyLockResult]) -> Vec<LockId> {
+    results
+        .iter()
+        .zip(lock_ids)
+        .filter(|(result, _)| result.success)
+        .map(|(_, lock_id)| lock_id.clone())
+        .collect()
+}
+
 /// Remote lock client implementation
 #[derive(Debug, Clone)]
 pub struct RemoteClient {
@@ -198,14 +329,202 @@ impl RemoteClient {
         )
     }
 
-    async fn execute_rpc<T, F>(&self, op: &'static str, resource_summary: &str, future: F) -> std::result::Result<T, LockError>
+    fn eviction_cooldown() -> Duration {
+        Duration::from_millis(rustfs_utils::get_env_u64(
+            rustfs_config::ENV_OBJECT_LOCK_RPC_EVICTION_COOLDOWN_MS,
+            rustfs_config::DEFAULT_OBJECT_LOCK_RPC_EVICTION_COOLDOWN_MS,
+        ))
+    }
+
+    fn detached_rpc_limit() -> usize {
+        rustfs_utils::get_env_usize(
+            rustfs_config::ENV_OBJECT_LOCK_RPC_DETACHED_LIMIT,
+            rustfs_config::DEFAULT_OBJECT_LOCK_RPC_DETACHED_LIMIT,
+        )
+    }
+
+    fn liveness_window(deadline: Duration) -> Duration {
+        deadline.saturating_mul(LOCK_RPC_LIVENESS_WINDOW_DEADLINES)
+    }
+
+    fn record_rpc_success(&self) {
+        with_lock_peer_health(&self.addr, |health| {
+            health.last_success = Some(Instant::now());
+            health.consecutive_timeouts = 0;
+        });
+    }
+
+    /// Apply the per-peer eviction policy after a failed RPC.
+    async fn maybe_evict_connection(
+        &self,
+        op: &'static str,
+        reason: &str,
+        resource_summary: &str,
+        trigger: EvictionTrigger,
+        deadline: Duration,
+    ) {
+        let now = Instant::now();
+        let cooldown = Self::eviction_cooldown();
+        let liveness_window = Self::liveness_window(deadline);
+        let (verdict, consecutive_timeouts) = with_lock_peer_health(&self.addr, |health| {
+            if trigger == EvictionTrigger::Timeout {
+                health.consecutive_timeouts = health.consecutive_timeouts.saturating_add(1);
+            }
+            let verdict = eviction_verdict(health, now, trigger, liveness_window, cooldown);
+            if verdict == EvictionVerdict::Evict {
+                health.last_eviction = Some(now);
+            }
+            (verdict, health.consecutive_timeouts)
+        });
+        if verdict == EvictionVerdict::Evict {
+            rustfs_io_metrics::lock_metrics::record_remote_lock_channel_eviction(&self.addr, trigger.as_str());
+            self.evict_connection(op, reason, resource_summary).await;
+            return;
+        }
+        rustfs_io_metrics::lock_metrics::record_remote_lock_channel_eviction_suppressed(&self.addr, verdict.as_str());
+        debug!(
+            addr = %self.addr,
+            op,
+            resource_summary,
+            trigger = trigger.as_str(),
+            verdict = verdict.as_str(),
+            consecutive_timeouts,
+            "Keeping cached remote lock connection after RPC failure"
+        );
+    }
+
+    /// Keep a timed-out RPC running instead of cancelling its stream.
+    ///
+    /// Dropping the future sends `RST_STREAM`; under load those resets pile up
+    /// in the server's pending-accept queue until it answers `GOAWAY
+    /// too_many_resets` and kills every stream on the connection. A detached
+    /// stream ends on its own within the internode RPC timeout, the number per
+    /// peer is bounded, and a lock granted after its caller gave up is released.
+    fn detach_timed_out_rpc<T: Send + 'static>(
+        &self,
+        op: &'static str,
+        resource_summary: &str,
+        handle: JoinHandle<std::result::Result<T, tonic::Status>>,
+        late: LateCompletion<T>,
+    ) {
+        let limit = Self::detached_rpc_limit();
+        let admitted = with_lock_peer_health(&self.addr, |health| {
+            if health.detached_rpcs >= limit {
+                false
+            } else {
+                health.detached_rpcs += 1;
+                true
+            }
+        });
+        if !admitted {
+            handle.abort();
+            rustfs_io_metrics::lock_metrics::record_remote_lock_rpc_detached(op, "aborted");
+            debug!(
+                addr = %self.addr,
+                op,
+                resource_summary,
+                limit,
+                "Cancelled timed-out remote lock RPC because the detached stream budget is exhausted"
+            );
+            return;
+        }
+        rustfs_io_metrics::lock_metrics::record_remote_lock_rpc_detached(op, "detached");
+        let addr = self.addr.clone();
+        tokio::spawn(async move {
+            let outcome = handle.await;
+            with_lock_peer_health(&addr, |health| health.detached_rpcs = health.detached_rpcs.saturating_sub(1));
+            match outcome {
+                Ok(Ok(response)) => {
+                    with_lock_peer_health(&addr, |health| {
+                        health.last_success = Some(Instant::now());
+                        health.consecutive_timeouts = 0;
+                    });
+                    rustfs_io_metrics::lock_metrics::record_remote_lock_rpc_late_completion(op, "success");
+                    if let Some(late) = late {
+                        late(response).await;
+                    }
+                }
+                Ok(Err(status)) => {
+                    rustfs_io_metrics::lock_metrics::record_remote_lock_rpc_late_completion(op, "error");
+                    debug!(
+                        addr = %addr,
+                        op,
+                        tonic_code = ?status.code(),
+                        tonic_message = status.message(),
+                        "Detached remote lock RPC failed after its caller timed out"
+                    );
+                }
+                Err(join_error) => {
+                    rustfs_io_metrics::lock_metrics::record_remote_lock_rpc_late_completion(op, "join_error");
+                    debug!(addr = %addr, op, error = %join_error, "Detached remote lock RPC task ended abnormally");
+                }
+            }
+        });
+    }
+
+    fn late_release_hook(&self, lock_id: LockId) -> LateCompletion<Response<GenerallyLockResponse>> {
+        let client = self.clone();
+        Some(Box::new(move |response: Response<GenerallyLockResponse>| {
+            Box::pin(async move {
+                if response.get_ref().success {
+                    client.release_late_acquisitions(vec![lock_id]).await;
+                }
+            })
+        }))
+    }
+
+    fn late_release_batch_hook(&self, lock_ids: Vec<LockId>) -> LateCompletion<Response<BatchGenerallyLockResponse>> {
+        let client = self.clone();
+        Some(Box::new(move |response: Response<BatchGenerallyLockResponse>| {
+            Box::pin(async move {
+                let acquired = acquired_lock_ids(&lock_ids, &response.get_ref().results);
+                if !acquired.is_empty() {
+                    client.release_late_acquisitions(acquired).await;
+                }
+            })
+        }))
+    }
+
+    /// A lock granted after its caller stopped waiting is an orphan until its
+    /// lease expires; hand it back right away, best effort.
+    async fn release_late_acquisitions(&self, lock_ids: Vec<LockId>) {
+        let outcome = match self.release_locks_batch(&lock_ids).await {
+            Ok(released) if released.iter().all(|released| *released) => "released",
+            Ok(_) => "partial",
+            Err(_) => "failed",
+        };
+        rustfs_io_metrics::lock_metrics::record_remote_lock_late_release(outcome);
+        if outcome == "released" {
+            debug!(addr = %self.addr, count = lock_ids.len(), "Released remote locks granted after their caller timed out");
+        } else {
+            warn!(
+                addr = %self.addr,
+                count = lock_ids.len(),
+                outcome,
+                "Could not release every remote lock granted after its caller timed out; the server lease will expire it"
+            );
+        }
+    }
+
+    async fn execute_rpc<T, Fut>(
+        &self,
+        op: &'static str,
+        resource_summary: &str,
+        deadline: Duration,
+        future: Fut,
+        late: LateCompletion<T>,
+    ) -> std::result::Result<T, LockError>
     where
-        F: std::future::Future<Output = std::result::Result<T, tonic::Status>>,
+        Fut: Future<Output = std::result::Result<T, tonic::Status>> + Send + 'static,
+        T: Send + 'static,
     {
-        let lock_timeout = Self::rpc_timeout();
+        let mut handle = tokio::spawn(future);
-        match timeout(lock_timeout, future).await {
+        match timeout(deadline, &mut handle).await {
-            Ok(Ok(response)) => Ok(response),
+            Ok(Ok(Ok(response))) => {
-            Ok(Err(err)) => {
+                self.record_rpc_success();
+                Ok(response)
+            }
+            Ok(Ok(Err(err))) => {
                 let reason = err.to_string();
                 // Only evict (and re-dial) the cached channel when the failure is a genuine
                 // transport problem. A server-produced application status (auth denied, peer
@@ -217,7 +536,7 @@ impl RemoteClient {
                     debug!(
                         addr = %self.addr,
                         op,
-                        timeout_ms = lock_timeout.as_millis(),
+                        timeout_ms = deadline.as_millis(),
                         resource_summary,
                         tonic_code = ?err.code(),
                         tonic_message = err.message(),
@@ -228,7 +547,7 @@ impl RemoteClient {
                     warn!(
                         addr = %self.addr,
                         op,
-                        timeout_ms = lock_timeout.as_millis(),
+                        timeout_ms = deadline.as_millis(),
                         resource_summary,
                         tonic_code = ?err.code(),
                         tonic_message = err.message(),
@@ -237,17 +556,29 @@ impl RemoteClient {
                     );
                 }
                 if transport_failure {
-                    self.evict_connection(op, &reason, resource_summary).await;
+                    self.maybe_evict_connection(op, &reason, resource_summary, EvictionTrigger::Transport, deadline)
+                        .await;
                 }
                 Err(LockError::internal(format!("{op} RPC failed: {reason}")))
             }
+            Ok(Err(join_error)) => {
+                warn!(
+                    addr = %self.addr,
+                    op,
+                    resource_summary,
+                    error = %join_error,
+                    "Remote lock RPC task ended abnormally"
+                );
+                Err(LockError::internal(format!("{op} RPC task failed: {join_error}")))
+            }
             Err(_) => {
-                let reason = format!("RPC timed out after {:?}", lock_timeout);
+                let reason = format!("RPC timed out after {deadline:?}");
+                rustfs_io_metrics::lock_metrics::record_remote_lock_rpc_timeout(&self.addr, op);
                 if Self::is_scanner_leader_lock(resource_summary) {
                     debug!(
                         addr = %self.addr,
                         op,
-                        timeout_ms = lock_timeout.as_millis(),
+                        timeout_ms = deadline.as_millis(),
                         resource_summary,
                         "Remote lock RPC timed out for scanner leader lock"
                     );
@@ -255,13 +586,15 @@ impl RemoteClient {
                     warn!(
                         addr = %self.addr,
                         op,
-                        timeout_ms = lock_timeout.as_millis(),
+                        timeout_ms = deadline.as_millis(),
                         resource_summary,
                         "Remote lock RPC timed out"
                     );
                 }
-                self.evict_connection(op, &reason, resource_summary).await;
+                self.maybe_evict_connection(op, &reason, resource_summary, EvictionTrigger::Timeout, deadline)
-                Err(LockError::timeout(format!("remote lock RPC {op} on {}", self.addr), lock_timeout))
+                    .await;
+                self.detach_timed_out_rpc(op, resource_summary, handle, late);
+                Err(LockError::timeout(format!("remote lock RPC {op} on {}", self.addr), deadline))
             }
         }
     }
@@ -354,8 +687,18 @@ impl LockClient for RemoteClient {
                 .map_err(|e| LockError::internal(format!("Failed to serialize request: {e}")))?,
         });
         attach_lock_mutation_body_digest(&mut req)?;
+        let late = self.late_release_hook(request.lock_id.clone());
 
-        let resp = match self.execute_rpc("lock", &resource_summary, client.lock(req)).await {
+        let resp = match self
+            .execute_rpc(
+                "lock",
+                &resource_summary,
+                Self::rpc_timeout(),
+                async move { client.lock(req).await },
+                late,
+            )
+            .await
+        {
             Ok(resp) => resp.into_inner(),
             Err(err @ LockError::Timeout { .. }) => return Ok(Self::rpc_timeout_failure_response(request, &err)),
             Err(err) => return Ok(Self::rpc_failure_response(request, &err)),
@@ -393,9 +736,16 @@ impl LockClient for RemoteClient {
                 .collect::<Result<Vec<_>>>()?,
         });
         attach_lock_mutation_body_digest(&mut req)?;
+        let late = self.late_release_batch_hook(requests.iter().map(|request| request.lock_id.clone()).collect());
 
         let resp = match self
-            .execute_rpc("lock_batch", &resource_summary, client.lock_batch(req))
+            .execute_rpc(
+                "lock_batch",
+                &resource_summary,
+                Self::rpc_timeout(),
+                async move { client.lock_batch(req).await },
+                late,
+            )
             .await
         {
             Ok(resp) => resp.into_inner(),
@@ -436,7 +786,13 @@ impl LockClient for RemoteClient {
         let mut req = Request::new(GenerallyLockRequest { args: request_string });
         attach_lock_mutation_body_digest(&mut req)?;
         let resp = self
-            .execute_rpc("release", &resource_summary, client.un_lock(req))
+            .execute_rpc(
+                "release",
+                &resource_summary,
+                Self::rpc_timeout(),
+                async move { client.un_lock(req).await },
+                None,
+            )
             .await?
             .into_inner();
         if let Some(error_info) = resp.error_info {
@@ -464,7 +820,13 @@ impl LockClient for RemoteClient {
         attach_lock_mutation_body_digest(&mut req)?;
 
         let resp = self
-            .execute_rpc("release_batch", &resource_summary, client.un_lock_batch(req))
+            .execute_rpc(
+                "release_batch",
+                &resource_summary,
+                Self::rpc_timeout(),
+                async move { client.un_lock_batch(req).await },
+                None,
+            )
             .await?
             .into_inner();
 
@@ -486,7 +848,13 @@ impl LockClient for RemoteClient {
         });
         attach_lock_mutation_body_digest(&mut req)?;
         let resp = self
-            .execute_rpc("refresh", &resource_summary, client.refresh(req))
+            .execute_rpc(
+                "refresh",
+                &resource_summary,
+                Self::rpc_timeout(),
+                async move { client.refresh(req).await },
+                None,
+            )
             .await?
             .into_inner();
         if let Some(error_info) = resp.error_info {
@@ -506,7 +874,13 @@ impl LockClient for RemoteClient {
         });
         attach_lock_mutation_body_digest(&mut req)?;
         let resp = self
-            .execute_rpc("force_release", &resource_summary, client.force_un_lock(req))
+            .execute_rpc(
+                "force_release",
+                &resource_summary,
+                Self::rpc_timeout(),
+                async move { client.force_un_lock(req).await },
+                None,
+            )
             .await?
             .into_inner();
         if let Some(error_info) = resp.error_info {
@@ -523,16 +897,26 @@ impl LockClient for RemoteClient {
         let status_request = Self::create_unlock_request(lock_id);
         let resource_summary = status_request.resource.to_string();
         let mut client = self.get_client().await?;
+        let args = serde_json::to_string(&status_request)
+            .map_err(|e| LockError::internal(format!("Failed to serialize request: {e}")))?;
 
         // Try to acquire a very short-lived lock to test availability
-        let mut req = Request::new(GenerallyLockRequest {
+        let mut req = Request::new(GenerallyLockRequest { args: args.clone() });
-            args: serde_json::to_string(&status_request)
-                .map_err(|e| LockError::internal(format!("Failed to serialize request: {e}")))?,
-        });
         attach_lock_mutation_body_digest(&mut req)?;
+        // A probe lock granted after the deadline must not linger on the peer.
+        let late = self.late_release_hook(lock_id.clone());
 
         // Try exclusive lock first with very short timeout
-        let resp = match self.execute_rpc("check_status", &resource_summary, client.lock(req)).await {
+        let resp = match self
+            .execute_rpc(
+                "check_status",
+                &resource_summary,
+                Self::rpc_timeout(),
+                async move { client.lock(req).await },
+                late,
+            )
+            .await
+        {
             Ok(response) => response.into_inner(),
             Err(_) => return Ok(Some(Self::unknown_lock_info(lock_id))),
         };
@@ -540,14 +924,19 @@ impl LockClient for RemoteClient {
         if resp.success {
             // If we successfully acquired the lock, the resource was free.
             // Immediately release it on a best-effort basis.
-            let mut release_req = Request::new(GenerallyLockRequest {
+            let mut release_req = Request::new(GenerallyLockRequest { args });
-                args: serde_json::to_string(&status_request)
-                    .map_err(|e| LockError::internal(format!("Failed to serialize request: {e}")))?,
-            });
             attach_lock_mutation_body_digest(&mut release_req)?;
-            let _ = self
+            if let Ok(mut client) = self.get_client().await {
-                .execute_rpc("check_status_release", &resource_summary, client.un_lock(release_req))
+                let _ = self
-                .await;
+                    .execute_rpc(
+                        "check_status_release",
+                        &resource_summary,
+                        Self::rpc_timeout(),
+                        async move { client.un_lock(release_req).await },
+                        None,
+                    )
+                    .await;
+            }
 
             Ok(None)
         } else {
@@ -582,19 +971,8 @@ impl LockClient for RemoteClient {
 
     async fn is_online(&self) -> bool {
         let online_timeout = Self::online_check_timeout();
-        match timeout(online_timeout, async {
+        let mut client = match timeout(online_timeout, self.get_client()).await {
-            let mut client = self.get_client().await?;
+            Ok(Ok(client)) => client,
-            let ping_req = Request::new(Self::build_ping_request());
-            self.execute_rpc("ping", Self::ONLINE_CHECK_RESOURCE, client.ping(ping_req))
-                .await?;
-            Ok::<(), LockError>(())
-        })
-        .await
-        {
-            Ok(Ok(())) => {
-                debug!(addr = %self.addr, timeout_ms = online_timeout.as_millis(), "remote lock client is online");
-                true
-            }
             Ok(Err(err)) => {
                 debug!(
                     addr = %self.addr,
@@ -602,16 +980,39 @@ impl LockClient for RemoteClient {
                     error = %err,
                     "remote lock client online check failed"
                 );
-                false
+                return false;
             }
             Err(_) => {
-                let reason = format!("online check timed out after {:?}", online_timeout);
                 warn!(
                     addr = %self.addr,
                     timeout_ms = online_timeout.as_millis(),
-                    "remote lock client online check timed out"
+                    "remote lock client online check timed out while dialing"
+                );
+                return false;
+            }
+        };
+        let ping_req = Request::new(Self::build_ping_request());
+        match self
+            .execute_rpc(
+                "ping",
+                Self::ONLINE_CHECK_RESOURCE,
+                online_timeout,
+                async move { client.ping(ping_req).await },
+                None,
+            )
+            .await
+        {
+            Ok(_) => {
+                debug!(addr = %self.addr, timeout_ms = online_timeout.as_millis(), "remote lock client is online");
+                true
+            }
+            Err(err) => {
+                debug!(
+                    addr = %self.addr,
+                    timeout_ms = online_timeout.as_millis(),
+                    error = %err,
+                    "remote lock client online check failed"
                 );
-                self.evict_connection("ping", &reason, Self::ONLINE_CHECK_RESOURCE).await;
                 false
             }
         }
@@ -673,6 +1074,232 @@ mod tests {
             .with_priority(LockPriority::Normal)
     }
 
+    #[test]
+    fn eviction_verdict_distinguishes_slow_peers_from_dead_channels() {
+        let now = Instant::now() + Duration::from_secs(3600);
+        let window = Duration::from_secs(6);
+        let cooldown = Duration::from_secs(5);
+
+        let idle = LockPeerChannelHealth::default();
+        assert_eq!(
+            eviction_verdict(&idle, now, EvictionTrigger::Timeout, window, cooldown),
+            EvictionVerdict::Evict
+        );
+
+        let serving = LockPeerChannelHealth {
+            last_success: Some(now - Duration::from_secs(1)),
+            ..Default::default()
+        };
+        assert_eq!(
+            eviction_verdict(&serving, now, EvictionTrigger::Timeout, window, cooldown),
+            EvictionVerdict::PeerRecentlyServed,
+            "a timeout on a peer that just answered is load, not a dead channel"
+        );
+        assert_eq!(
+            eviction_verdict(&serving, now, EvictionTrigger::Transport, window, cooldown),
+            EvictionVerdict::Evict,
+            "a transport failure is reported by the channel itself and still evicts"
+        );
+
+        let quiet = LockPeerChannelHealth {
+            last_success: Some(now - Duration::from_secs(30)),
+            ..Default::default()
+        };
+        assert_eq!(
+            eviction_verdict(&quiet, now, EvictionTrigger::Timeout, window, cooldown),
+            EvictionVerdict::Evict
+        );
+
+        let just_evicted = LockPeerChannelHealth {
+            last_eviction: Some(now - Duration::from_secs(1)),
+            ..Default::default()
+        };
+        assert_eq!(
+            eviction_verdict(&just_evicted, now, EvictionTrigger::Timeout, window, cooldown),
+            EvictionVerdict::CoolingDown
+        );
+        assert_eq!(
+            eviction_verdict(&just_evicted, now, EvictionTrigger::Transport, window, cooldown),
+            EvictionVerdict::CoolingDown
+        );
+
+        let cooled = LockPeerChannelHealth {
+            last_eviction: Some(now - Duration::from_secs(10)),
+            ..Default::default()
+        };
+        assert_eq!(
+            eviction_verdict(&cooled, now, EvictionTrigger::Timeout, window, cooldown),
+            EvictionVerdict::Evict
+        );
+    }
+
+    #[test]
+    fn acquired_lock_ids_picks_only_granted_batch_entries() {
+        let lock_ids = vec![
+            LockId::new_unique(&ObjectKey::new("bucket", "a")),
+            LockId::new_unique(&ObjectKey::new("bucket", "b")),
+            LockId::new_unique(&ObjectKey::new("bucket", "c")),
+        ];
+        let results = vec![
+            GenerallyLockResult {
+                success: true,
+                ..Default::default()
+            },
+            GenerallyLockResult {
+                success: false,
+                ..Default::default()
+            },
+        ];
+        let acquired = acquired_lock_ids(&lock_ids, &results);
+        assert_eq!(
+            acquired,
+            vec![lock_ids[0].clone()],
+            "only granted entries with a matching id are released"
+        );
+        assert!(acquired_lock_ids(&lock_ids, &[]).is_empty());
+    }
+
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn test_remote_client_timeout_keeps_channel_of_recently_serving_peer() {
+        ensure_test_rpc_secret();
+        let Some((addr, accept_task)) = spawn_hanging_listener().await else {
+            return;
+        };
+        reset_lock_peer_health_for_test(&addr);
+        cache_lazy_channel(&addr).await;
+        with_lock_peer_health(&addr, |health| health.last_success = Some(Instant::now()));
+
+        temp_env::async_with_vars([(rustfs_config::ENV_OBJECT_LOCK_RPC_TIMEOUT_MS, Some("50"))], async {
+            let client = RemoteClient::new(addr.clone());
+            let response = client
+                .acquire_lock(&test_lock_request(Duration::from_millis(5)))
+                .await
+                .unwrap();
+            assert!(!response.success, "timed out lock acquisition should fail");
+            assert!(
+                runtime_sources::test_node_channel_is_cached(&addr).await,
+                "a peer that served a lock RPC within the liveness window is slow, not gone"
+            );
+            assert_eq!(lock_peer_health_for_test(&addr).consecutive_timeouts, 1);
+        })
+        .await;
+
+        accept_task.abort();
+        reset_lock_peer_health_for_test(&addr);
+    }
+
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn test_remote_client_repeated_timeouts_evict_at_most_once_per_cooldown() {
+        ensure_test_rpc_secret();
+        let Some((addr, accept_task)) = spawn_hanging_listener().await else {
+            return;
+        };
+        reset_lock_peer_health_for_test(&addr);
+        cache_lazy_channel(&addr).await;
+
+        temp_env::async_with_vars(
+            [
+                (rustfs_config::ENV_OBJECT_LOCK_RPC_TIMEOUT_MS, Some("50")),
+                (rustfs_config::ENV_OBJECT_LOCK_RPC_EVICTION_COOLDOWN_MS, Some("60000")),
+            ],
+            async {
+                let client = RemoteClient::new(addr.clone());
+                let request = test_lock_request(Duration::from_millis(5));
+
+                let _ = client.acquire_lock(&request).await.unwrap();
+                assert!(
+                    !runtime_sources::test_node_channel_is_cached(&addr).await,
+                    "the first timeout on a quiet peer evicts the cached channel"
+                );
+
+                cache_lazy_channel(&addr).await;
+                let _ = client.acquire_lock(&request).await.unwrap();
+                assert!(
+                    runtime_sources::test_node_channel_is_cached(&addr).await,
+                    "a second timeout inside the cooldown must not tear the fresh channel down again"
+                );
+                assert_eq!(lock_peer_health_for_test(&addr).consecutive_timeouts, 2);
+            },
+        )
+        .await;
+
+        accept_task.abort();
+        reset_lock_peer_health_for_test(&addr);
+    }
+
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn test_remote_client_detaches_timed_out_rpc_and_reclaims_its_slot() {
+        ensure_test_rpc_secret();
+        let Some((addr, accept_task)) = spawn_hanging_listener().await else {
+            return;
+        };
+        reset_lock_peer_health_for_test(&addr);
+        cache_lazy_channel(&addr).await;
+
+        temp_env::async_with_vars([(rustfs_config::ENV_OBJECT_LOCK_RPC_TIMEOUT_MS, Some("50"))], async {
+            let client = RemoteClient::new(addr.clone());
+            let _ = client
+                .acquire_lock(&test_lock_request(Duration::from_millis(5)))
+                .await
+                .unwrap();
+            assert_eq!(
+                lock_peer_health_for_test(&addr).detached_rpcs,
+                1,
+                "the timed-out stream keeps running instead of being reset"
+            );
+
+            // The hanging listener drops its socket after two seconds; the detached
+            // task then observes the transport failure and frees its slot.
+            let deadline = Instant::now() + Duration::from_secs(10);
+            while lock_peer_health_for_test(&addr).detached_rpcs != 0 {
+                assert!(Instant::now() < deadline, "detached RPC slot must be reclaimed once the stream ends");
+                tokio::time::sleep(Duration::from_millis(50)).await;
+            }
+        })
+        .await;
+
+        accept_task.abort();
+        reset_lock_peer_health_for_test(&addr);
+    }
+
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn test_remote_client_cancels_timed_out_rpc_when_detached_budget_is_exhausted() {
+        ensure_test_rpc_secret();
+        let Some((addr, accept_task)) = spawn_hanging_listener().await else {
+            return;
+        };
+        reset_lock_peer_health_for_test(&addr);
+        cache_lazy_channel(&addr).await;
+
+        temp_env::async_with_vars(
+            [
+                (rustfs_config::ENV_OBJECT_LOCK_RPC_TIMEOUT_MS, Some("50")),
+                (rustfs_config::ENV_OBJECT_LOCK_RPC_DETACHED_LIMIT, Some("0")),
+            ],
+            async {
+                let client = RemoteClient::new(addr.clone());
+                let response = client
+                    .acquire_lock(&test_lock_request(Duration::from_millis(5)))
+                    .await
+                    .unwrap();
+                assert!(!response.success);
+                assert_eq!(
+                    lock_peer_health_for_test(&addr).detached_rpcs,
+                    0,
+                    "an exhausted detached budget falls back to cancelling the stream"
+                );
+            },
+        )
+        .await;
+
+        accept_task.abort();
+        reset_lock_peer_health_for_test(&addr);
+    }
+
     #[test]
     fn lock_mutation_helper_marks_single_and_batch_requests_for_rolling_auth() {
         let mut single = Request::new(GenerallyLockRequest {
@@ -714,6 +1341,7 @@ mod tests {
         let Some((addr, accept_task)) = spawn_hanging_listener().await else {
             return;
         };
+        reset_lock_peer_health_for_test(&addr);
         cache_lazy_channel(&addr).await;
         assert!(runtime_sources::test_node_channel_is_cached(&addr).await);
 
@@ -759,6 +1387,7 @@ mod tests {
         let Some((addr, accept_task)) = spawn_hanging_listener().await else {
             return;
         };
+        reset_lock_peer_health_for_test(&addr);
         cache_lazy_channel(&addr).await;
         assert!(runtime_sources::test_node_channel_is_cached(&addr).await);
 
@@ -805,6 +1434,7 @@ mod tests {
         let Some((addr, accept_task)) = spawn_hanging_listener().await else {
             return;
         };
+        reset_lock_peer_health_for_test(&addr);
         cache_lazy_channel(&addr).await;
         assert!(runtime_sources::test_node_channel_is_cached(&addr).await);
 
@@ -842,6 +1472,7 @@ mod tests {
         let Some((addr, accept_task)) = spawn_hanging_listener().await else {
             return;
         };
+        reset_lock_peer_health_for_test(&addr);
         cache_lazy_channel(&addr).await;
         assert!(runtime_sources::test_node_channel_is_cached(&addr).await);
 
@@ -884,6 +1515,7 @@ mod tests {
         let Some(addr) = closed_listener_addr().await else {
             return;
         };
+        reset_lock_peer_health_for_test(&addr);
         cache_lazy_channel(&addr).await;
         assert!(runtime_sources::test_node_channel_is_cached(&addr).await);
 

crates/ecstore/src/config/com.rs

@@ -449,14 +449,6 @@ where
     Ok(data)
 }
 
-pub(crate) async fn read_config_limited<S>(api: Arc<S>, file: &str, max_bytes: usize) -> Result<Vec<u8>>
-where
-    S: EcstoreObjectIO,
-{
-    let (data, _obj) = read_config_with_metadata_inner(api, file, &ObjectOptions::default(), false, Some(max_bytes)).await?;
-    Ok(data)
-}
-
 pub(crate) async fn read_config_limited_preserve_empty<S>(api: Arc<S>, file: &str, max_bytes: usize) -> Result<Vec<u8>>
 where
     S: EcstoreObjectIO,
@@ -465,6 +457,14 @@ where
     Ok(data)
 }
 
+pub(crate) async fn read_config_limited<S>(api: Arc<S>, file: &str, max_bytes: usize) -> Result<Vec<u8>>
+where
+    S: EcstoreObjectIO,
+{
+    let (data, _obj) = read_config_with_metadata_inner(api, file, &ObjectOptions::default(), false, Some(max_bytes)).await?;
+    Ok(data)
+}
+
 pub(crate) async fn read_config_limited_preserve_empty_with_metadata<S>(
     api: Arc<S>,
     file: &str,
@@ -476,6 +476,18 @@ where
     read_config_with_metadata_inner(api, file, &ObjectOptions::default(), true, Some(max_bytes)).await
 }
 
+pub(crate) async fn read_config_limited_preserve_empty_with_metadata_opts<S>(
+    api: Arc<S>,
+    file: &str,
+    opts: &ObjectOptions,
+    max_bytes: usize,
+) -> Result<(Vec<u8>, ObjectInfo)>
+where
+    S: EcstoreObjectIO,
+{
+    read_config_with_metadata_inner(api, file, opts, true, Some(max_bytes)).await
+}
+
 /// Read an existing config object without treating an empty payload as absent.
 /// Callers that validate their own payload format need to distinguish corruption
 /// from `ConfigNotFound`.

crates/ecstore/src/config/storageclass.rs

@@ -747,18 +747,27 @@ mod tests {
         let mut kvs = KVS::new();
         kvs.insert(CLASS_STANDARD.to_string(), "EC:2".to_string());
 
-        let err = lookup_config_for_pools_with_env(&kvs, &[4, 2], no_env_overrides())
+        for drives in [2, 3] {
-            .expect_err("EC:2 must be rejected by the two-drive pool");
+            let err = lookup_config_for_pools_with_env(&kvs, &[4, drives], no_env_overrides())
-        assert!(
+                .expect_err("EC:2 must be rejected by a pool with fewer than four drives per set");
-            err.to_string().contains("pool 1") && err.to_string().contains("2 drives"),
+            assert!(
-            "error must identify the rejecting pool: {err}"
+                err.to_string().contains("pool 1") && err.to_string().contains(&format!("{drives} drives")),
-        );
+                "error must identify the rejecting pool: {err}"
+            );
+        }
+
+        let cfg =
+            lookup_config_for_pools_with_env(&kvs, &[4, 4], no_env_overrides()).expect("EC:2 is valid for both four-drive pools");
+        assert_eq!(cfg.parities_for_sc(STANDARD), Some(vec![2, 2]));
 
         kvs.insert(CLASS_STANDARD.to_string(), "EC:1".to_string());
-        let cfg = lookup_config_for_pools_with_env(&kvs, &[4, 2], no_env_overrides()).expect("EC:1 is valid for both pools");
+        for drives in [2, 3, 4] {
-        assert_eq!(cfg.parity_for_sc(STANDARD, 4), Some(1));
+            let cfg =
-        assert_eq!(cfg.parity_for_sc(STANDARD, 2), Some(1));
+                lookup_config_for_pools_with_env(&kvs, &[4, drives], no_env_overrides()).expect("EC:1 is valid for both pools");
-        assert_eq!(cfg.get_parity_for_sc(STANDARD), Some(1));
+            assert_eq!(cfg.parity_for_sc(STANDARD, 4), Some(1));
+            assert_eq!(cfg.parity_for_sc(STANDARD, drives), Some(1));
+            assert_eq!(cfg.get_parity_for_sc(STANDARD), Some(1));
+        }
     }
 
     #[test]

crates/ecstore/src/core/pools_test.rs

@@ -39,6 +39,7 @@ mod capacity_dedup_tests {
                 ..Default::default()
             },
             disks: disks.clone(),
+            ..Default::default()
         };
 
         let total = get_total_usable_capacity(&disks, &info);
@@ -73,6 +74,7 @@ mod capacity_dedup_tests {
                 ..Default::default()
             },
             disks: disks.clone(),
+            ..Default::default()
         };
 
         let total = get_total_usable_capacity(&disks, &info);
@@ -150,6 +152,7 @@ mod capacity_dedup_tests {
                 ..Default::default()
             },
             disks: disks.clone(),
+            ..Default::default()
         };
 
         let total = get_total_usable_capacity(&disks, &info);
@@ -567,6 +570,465 @@ mod decommission_lock_order_tests {
             .expect("decommission activation should commit after the probe release");
     }
 
+    #[test]
+    #[serial_test::serial]
+    fn staged_external_put_rechecks_retiring_source_on_another_node() {
+        run_large_stack_current_thread_async_test("staged-retiring-source", || async {
+            let (_temp_dirs, store, other_store) = test_three_pool_stores_with_isolated_node_contexts(None).await;
+            let bucket = test_bucket("staged-source");
+            let object = "selected-before-retirement.bin";
+            let original = b"original source object";
+            store
+                .make_bucket(&bucket, &MakeBucketOptions::default())
+                .await
+                .expect("create staged source bucket");
+            store.pools[0]
+                .put_object(&bucket, object, &mut PutObjReader::from_vec(original.to_vec()), &ObjectOptions::default())
+                .await
+                .expect("seed the source selected before retirement");
+
+            let layout = DecommissionErasureLayout { data: 1, parity: 0 };
+            set_decommission_capacity_info_overrides_for_test(
+                other_store.id,
+                vec![vec![
+                    DecommissionPoolCapacityInfo::for_test(0, layout, 0, 1024, 1024),
+                    DecommissionPoolCapacityInfo::for_test(1, layout, 4096, 4096, 0),
+                    DecommissionPoolCapacityInfo::for_test(2, layout, 0, 4096, 4096),
+                ]],
+            );
+            let barrier = DecommissionCapacityLockOrderBarrier::install(store.id, store.id);
+            barrier.pause_external_object_commit_phase();
+            let put_store = Arc::clone(&store);
+            let put_bucket = bucket.clone();
+            let put = tokio::spawn(async move {
+                put_store
+                    .put_object(
+                        &put_bucket,
+                        object,
+                        &mut PutObjReader::from_vec(b"must not replace a retiring source".to_vec()),
+                        &ObjectOptions::default(),
+                    )
+                    .await
+            });
+            tokio::time::timeout(Duration::from_secs(30), barrier.wait_until_external_object_commit_phase_started())
+                .await
+                .expect("public PUT must stage before its decommission commit probe");
+            assert!(!store.pool_meta.read().await.is_suspended(0));
+            other_store
+                .save_current_pool_meta_for_decommission_start(&[0], Vec::new())
+                .await
+                .expect("the other node should activate retirement before the staged PUT commits");
+            assert!(other_store.pool_meta.read().await.is_suspended(0));
+            assert!(
+                !store.pool_meta.read().await.is_suspended(0),
+                "the writer's local snapshot must remain stale to exercise the durable admission probe"
+            );
+            barrier.release_external_object_commit_phase();
+            let result = tokio::time::timeout(Duration::from_secs(30), put)
+                .await
+                .expect("staged PUT must finish after the commit probe is released")
+                .expect("staged PUT must not panic");
+            assert!(
+                matches!(result, Err(crate::error::Error::SlowDown)),
+                "a staged PUT must retry pool selection instead of committing to a newly retiring source: {result:?}"
+            );
+            let mut reader = store.pools[0]
+                .get_object_reader(&bucket, object, None, HeaderMap::new(), &ObjectOptions::default())
+                .await
+                .expect("the original source must remain readable after admission rejects the replacement");
+            let mut body = Vec::new();
+            reader
+                .stream
+                .read_to_end(&mut body)
+                .await
+                .expect("read the full retained source body");
+            assert_eq!(body, original);
+        });
+    }
+
+    #[test]
+    #[serial_test::serial]
+    fn reserved_target_shares_business_io_and_retains_source_after_capacity_loss() {
+        run_large_stack_current_thread_async_test("shared-decommission-capacity", || async {
+            for lose_capacity in [false, true] {
+                let (_temp_dirs, store, other_store) =
+                    test_three_pool_stores_with_three_disk_sets_with_isolated_node_contexts(None).await;
+                let bucket = test_bucket("shared-capacity");
+                let object = "migrating-source.bin";
+                let business_object = "business-write.bin";
+                let multipart_object = "business-multipart.bin";
+                let source_body = vec![0x35; 256 * 1024];
+                let business_body = vec![0x57; 64 * 1024];
+                store
+                    .make_bucket(&bucket, &MakeBucketOptions::default())
+                    .await
+                    .expect("create shared-capacity bucket");
+                store.pools[0]
+                    .put_object(
+                        &bucket,
+                        object,
+                        &mut PutObjReader::from_vec(source_body.clone()),
+                        &ObjectOptions::default(),
+                    )
+                    .await
+                    .expect("seed the retiring source");
+                store.pools[2]
+                    .put_object(
+                        &bucket,
+                        business_object,
+                        &mut PutObjReader::from_vec(b"previous business value".to_vec()),
+                        &ObjectOptions::default(),
+                    )
+                    .await
+                    .expect("pin the public overwrite to the migration target");
+                let multipart_opts = ObjectOptions {
+                    expected_bucket_incarnation_id: Some(
+                        store
+                            .bucket_incarnation_id(&bucket)
+                            .await
+                            .expect("load the multipart bucket identity"),
+                    ),
+                    ..Default::default()
+                };
+                let routing_upload = new_multipart_upload(&store, 2, &bucket, multipart_object, multipart_opts.clone())
+                    .await
+                    .expect("pin subsequent public multipart creation to the migration target");
+                let layout = DecommissionErasureLayout { data: 1, parity: 0 };
+                let target_total = source_body.len() * 8;
+                let capacities = vec![
+                    DecommissionPoolCapacityInfo::for_test(0, layout, 0, source_body.len() * 2, source_body.len() * 2),
+                    DecommissionPoolCapacityInfo::for_test(1, layout, 0, target_total, target_total),
+                    DecommissionPoolCapacityInfo::for_test(2, layout, target_total, target_total, 0),
+                ];
+                set_decommission_capacity_info_overrides_for_test(store.id, vec![capacities.clone()]);
+                store
+                    .save_current_pool_meta_for_decommission_start(&[0], Vec::new())
+                    .await
+                    .expect("activate source retirement");
+                *other_store.pool_meta.write().await = store.pool_meta.read().await.clone();
+                let before = other_store.pool_meta.read().await.clone();
+                let reservation = before.pools[0]
+                    .decommission
+                    .as_ref()
+                    .expect("active source")
+                    .capacity_reservation
+                    .as_ref()
+                    .expect("durable reservation");
+                assert_eq!(
+                    reservation.model_version, 2,
+                    "exercise migration I/O outside the global metadata write lock"
+                );
+                assert_eq!(reservation.targets[0].pool_index, 2);
+
+                let barrier =
+                    crate::set_disk::rename_fanout_barrier::arm(object, 0, crate::set_disk::rename_fanout_barrier::PHASE_RENAME);
+                let migration_store = Arc::clone(&store);
+                let migration_bucket = bucket.clone();
+                let migration = tokio::spawn(async move {
+                    migration_store
+                        .decommission_entry_for_test_with_bucket_incarnation(
+                            0,
+                            MetaCacheEntry {
+                                name: object.to_string(),
+                                ..Default::default()
+                            },
+                            migration_bucket,
+                            migration_store.pools[0].get_disks_by_key(object),
+                        )
+                        .await
+                });
+                tokio::time::timeout(Duration::from_secs(30), barrier.wait_until_paused())
+                    .await
+                    .expect("migration must reach target publication");
+                assert!(!migration.is_finished());
+                let mut pending = crate::core::pools::PoolMeta::default();
+                pending
+                    .load_no_lock_from_replicas(other_store.pools.clone())
+                    .await
+                    .expect("read migration intent from the other node");
+                let pending_reservation = pending.pools[0]
+                    .decommission
+                    .as_ref()
+                    .expect("active source")
+                    .capacity_reservation
+                    .as_ref()
+                    .expect("pending reservation")
+                    .clone();
+                assert_eq!(pending_reservation.pending_target_physical_bytes, source_body.len());
+                assert_eq!(pending_reservation.consumed_target_physical_bytes, 0);
+
+                tokio::time::timeout(
+                    Duration::from_secs(30),
+                    other_store.put_object(
+                        &bucket,
+                        business_object,
+                        &mut PutObjReader::from_vec(business_body.clone()),
+                        &ObjectOptions::default(),
+                    ),
+                )
+                .await
+                .expect("business PUT must finish without waiting for the migration target gate")
+                .expect("a reserved healthy pool must accept ordinary PUT");
+                tokio::time::timeout(Duration::from_secs(30), async {
+                    let upload = other_store
+                        .new_multipart_upload(&bucket, multipart_object, &ObjectOptions::default())
+                        .await
+                        .expect("the reserved target must accept public multipart creation");
+                    assert_ne!(upload.upload_id, routing_upload.upload_id);
+                    let lifecycle_guard = other_store
+                        .acquire_bucket_lifecycle_read_lock(&bucket)
+                        .await
+                        .expect("fence the exact-pool multipart placement check");
+                    let mut lookup_opts = multipart_opts.clone();
+                    lookup_opts.add_bucket_lifecycle_lock_guard(&lifecycle_guard);
+                    other_store.pools[2]
+                        .get_multipart_info(&bucket, multipart_object, &upload.upload_id, &lookup_opts)
+                        .await
+                        .expect("public multipart creation must actually select the reserved target");
+                    drop(lifecycle_guard);
+                    let mut final_part = None;
+                    for payload in [vec![0x18; business_body.len()], business_body.clone()] {
+                        final_part = Some(
+                            other_store
+                                .put_object_part(
+                                    &bucket,
+                                    multipart_object,
+                                    &upload.upload_id,
+                                    1,
+                                    &mut PutObjReader::from_vec(payload),
+                                    &ObjectOptions::default(),
+                                )
+                                .await
+                                .expect("the reserved target must accept UploadPart and replacement of the same part"),
+                        );
+                    }
+                    let part = final_part.expect("the replacement part must be present");
+                    Arc::clone(&other_store)
+                        .complete_multipart_upload(
+                            &bucket,
+                            multipart_object,
+                            &upload.upload_id,
+                            vec![crate::storage_api_contracts::multipart::CompletePart {
+                                part_num: part.part_num,
+                                etag: part.etag,
+                                ..Default::default()
+                            }],
+                            &ObjectOptions::default(),
+                        )
+                        .await
+                        .expect("the reserved target must accept multipart completion");
+                    other_store
+                        .abort_multipart_upload(&bucket, multipart_object, &routing_upload.upload_id, &ObjectOptions::default())
+                        .await
+                        .expect("ordinary multipart cleanup must not consume the migration's pending intent");
+                })
+                .await
+                .expect("business multipart operations must finish while migration I/O is paused");
+                assert!(!migration.is_finished(), "business publication must overlap paused migration I/O");
+                let mut after_business = crate::core::pools::PoolMeta::default();
+                after_business
+                    .load_no_lock_from_replicas(other_store.pools.clone())
+                    .await
+                    .expect("reload the shared-capacity ledger");
+                assert_eq!(
+                    after_business.pools[0]
+                        .decommission
+                        .as_ref()
+                        .expect("active source")
+                        .capacity_reservation
+                        .as_ref(),
+                    Some(&pending_reservation),
+                    "ordinary PUT and multipart operations must not settle or consume the migration's pending identity"
+                );
+
+                let mut after_capacity = capacities;
+                // Capacity injection is deterministic; the object I/O and durable metadata use real temporary disks.
+                let free = if lose_capacity {
+                    0
+                } else {
+                    target_total - source_body.len() - business_body.len() * 2
+                };
+                after_capacity[2] = DecommissionPoolCapacityInfo::for_test(2, layout, free, target_total, target_total - free);
+                set_decommission_capacity_info_overrides_for_test(store.id, vec![after_capacity]);
+                barrier.release();
+                drop(barrier);
+                let migrated = tokio::time::timeout(Duration::from_secs(30), migration)
+                    .await
+                    .expect("migration must finish after publication resumes")
+                    .expect("migration task must not panic");
+                if lose_capacity {
+                    let err =
+                        migrated.expect_err("capacity loss must prevent source cleanup, even after the target write commits");
+                    assert!(err.to_string().contains("capacity"), "unexpected migration error: {err}");
+                } else {
+                    migrated.expect("shared-capacity migration should finish when space remains sufficient");
+                }
+                let mut persisted = crate::core::pools::PoolMeta::default();
+                persisted
+                    .load_no_lock_from_replicas(other_store.pools.clone())
+                    .await
+                    .expect("reload finalized migration state");
+                let info = persisted.pools[0].decommission.as_ref().expect("source state");
+                let reservation = info.capacity_reservation.as_ref().expect("migration ledger");
+                assert_eq!(reservation.pending_target_physical_bytes, 0);
+                assert_eq!(
+                    reservation.consumed_target_physical_bytes,
+                    source_body.len(),
+                    "foreign writes must not count as committed source bytes"
+                );
+                assert_eq!(reservation.committed_data_bytes, source_body.len());
+                assert_eq!(info.capacity_blocked_reason.is_some(), lose_capacity);
+                for (pool, key, expected) in [
+                    (2, business_object, &business_body),
+                    (2, multipart_object, &business_body),
+                    (2, object, &source_body),
+                ] {
+                    let mut reader = other_store.pools[pool]
+                        .get_object_reader(&bucket, key, None, HeaderMap::new(), &ObjectOptions::default())
+                        .await
+                        .expect("all acknowledged target objects must remain readable");
+                    let mut actual = Vec::new();
+                    reader.read_to_end(&mut actual).await.expect("read the complete target body");
+                    assert_eq!(&actual, expected);
+                }
+                if lose_capacity {
+                    let mut source = other_store.pools[0]
+                        .get_object_reader(&bucket, object, None, HeaderMap::new(), &ObjectOptions::default())
+                        .await
+                        .expect("capacity-blocked migration must retain its source");
+                    let mut actual = Vec::new();
+                    source
+                        .read_to_end(&mut actual)
+                        .await
+                        .expect("read the complete retained source");
+                    assert_eq!(actual, source_body);
+                    other_store
+                        .put_object(
+                            &bucket,
+                            business_object,
+                            &mut PutObjReader::from_vec(business_body.clone()),
+                            &ObjectOptions::default(),
+                        )
+                        .await
+                        .expect("a capacity-blocked migration must not itself make the healthy target read-only");
+                } else {
+                    let err = other_store.pools[0]
+                        .get_object_info(&bucket, object, &ObjectOptions::default())
+                        .await
+                        .expect_err("successful migration must clean the exact source");
+                    assert!(crate::error::is_err_object_not_found(&err));
+                }
+            }
+        });
+    }
+
+    #[test]
+    #[serial_test::serial]
+    fn mixed_batch_delete_admits_only_marker_destinations_during_retirement() {
+        run_large_stack_current_thread_async_test("batch-marker-admission", || async {
+            use crate::storage_api_contracts::object::ObjectToDelete;
+
+            for marker_target in [1, 2] {
+                let (_temp_dirs, store, _other_store) = test_three_pool_stores_with_isolated_node_contexts(None).await;
+                let bucket = test_bucket("batch-marker");
+                store
+                    .make_bucket(
+                        &bucket,
+                        &MakeBucketOptions {
+                            versioning_enabled: true,
+                            ..Default::default()
+                        },
+                    )
+                    .await
+                    .expect("create a versioned batch-delete bucket");
+                let source_version = uuid::Uuid::new_v4();
+                for (pool, object, version) in [
+                    (0, "purge-source", source_version),
+                    (marker_target, "mark-active", uuid::Uuid::new_v4()),
+                ] {
+                    store.pools[pool]
+                        .put_object(
+                            &bucket,
+                            object,
+                            &mut PutObjReader::from_vec(b"version to delete".to_vec()),
+                            &ObjectOptions {
+                                versioned: true,
+                                version_id: Some(version.to_string()),
+                                ..Default::default()
+                            },
+                        )
+                        .await
+                        .expect("seed each exact batch-delete destination");
+                }
+                let layout = DecommissionErasureLayout { data: 1, parity: 0 };
+                set_decommission_capacity_info_overrides_for_test(
+                    store.id,
+                    vec![vec![
+                        DecommissionPoolCapacityInfo::for_test(0, layout, 0, 1024, 1024),
+                        DecommissionPoolCapacityInfo::for_test(1, layout, 4096, 4096, 0),
+                        DecommissionPoolCapacityInfo::for_test(2, layout, 0, 4096, 4096),
+                    ]],
+                );
+                store
+                    .save_current_pool_meta_for_decommission_start(&[0], Vec::new())
+                    .await
+                    .expect("reserve pool 1 while pool 0 retires and pool 2 remains unreserved");
+                let (deleted, errors) = store
+                    .delete_objects(
+                        &bucket,
+                        vec![
+                            ObjectToDelete {
+                                object_name: "mark-active".to_string(),
+                                ..Default::default()
+                            },
+                            ObjectToDelete {
+                                object_name: "purge-source".to_string(),
+                                version_id: Some(source_version),
+                                ..Default::default()
+                            },
+                        ],
+                        ObjectOptions::default(),
+                    )
+                    .await;
+                assert_eq!(errors.len(), 2);
+                assert!(
+                    errors.iter().all(Option::is_none),
+                    "the unrelated retiring/reserved pools must not reject marker admission: {errors:?}"
+                );
+                assert_eq!(deleted.len(), 2);
+                assert_eq!(deleted[0].object_name, "mark-active");
+                assert!(deleted[0].delete_marker);
+                assert!(
+                    deleted[0].version_id.is_none(),
+                    "a latest-version delete does not request an explicit version"
+                );
+                assert!(
+                    deleted[0].delete_marker_version_id.is_some(),
+                    "the newly created marker must have its own version identity"
+                );
+                assert_eq!(deleted[1].object_name, "purge-source");
+                assert!(!deleted[1].delete_marker);
+                assert_eq!(deleted[1].version_id, Some(source_version));
+                assert!(
+                    matches!(
+                        store.pools[0]
+                            .get_object_info(
+                                &bucket,
+                                "purge-source",
+                                &ObjectOptions {
+                                    version_id: Some(source_version.to_string()),
+                                    ..Default::default()
+                                },
+                            )
+                            .await,
+                        Err(crate::error::Error::ObjectNotFound(..) | crate::error::Error::VersionNotFound(..))
+                    ),
+                    "an exact source deletion must retain its capacity-release path"
+                );
+            }
+        });
+    }
+
     #[tokio::test]
     #[serial_test::serial]
     async fn public_upload_part_holds_decommission_capacity_until_rename() {
@@ -4496,6 +4958,470 @@ mod decommission_lock_order_tests {
         }
     }
 
+    #[test]
+    #[serial_test::serial]
+    fn scanner_backlog_cas_keeps_fences_after_waiter_cancellation_until_rename_drains() {
+        run_large_stack_current_thread_async_test("scanner-backlog-canceled-waiter", async || {
+            temp_env::async_with_vars([(crate::set_disk::ENV_RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE, Some("true"))], async {
+                let (_temp_dirs, writer, other) =
+                    test_three_pool_stores_with_three_disk_sets_with_isolated_node_contexts(None).await;
+                let object = "buckets/.scanner-pause-backlog.json";
+                let set_index = 1;
+                let body = vec![0x37; 1024];
+                assert!(
+                    !writer.pools[0].disk_set[0]
+                        .shares_namespace_lock_domain(&writer.pools[0].disk_set[set_index])
+                        .await
+                );
+                let rename_tasks = crate::set_disk::rename_fanout_barrier::observe_tasks(object);
+                let tail =
+                    crate::set_disk::rename_fanout_barrier::arm(object, 0, crate::set_disk::rename_fanout_barrier::PHASE_RENAME);
+                let put_store = Arc::clone(&writer);
+                let put_body = body.clone();
+                let mut put = tokio::spawn(async move {
+                    put_store
+                        .save_scanner_pause_backlog_replica(0, set_index, put_body, Default::default())
+                        .await
+                });
+                tokio::time::timeout(Duration::from_secs(30), tail.wait_until_paused())
+                    .await
+                    .expect("the native write must reach its held rename");
+                tokio::time::timeout(Duration::from_secs(30), async {
+                    while rename_tasks.running() != 1 {
+                        tokio::task::yield_now().await;
+                    }
+                })
+                .await
+                .expect("the other disks must reach quorum before canceling the waiter");
+                assert!(
+                    tokio::time::timeout(Duration::from_millis(100), &mut put).await.is_err(),
+                    "native replica publication must await the entire rename tail"
+                );
+                put.abort();
+                assert!(put.await.expect_err("the scanner waiter must be canceled").is_cancelled());
+
+                let capacity_lock = other
+                    .new_ns_lock(RUSTFS_META_BUCKET, POOL_META_NAME)
+                    .await
+                    .expect("capacity lock probe");
+                let object_lock = other
+                    .new_ns_lock(RUSTFS_META_BUCKET, object)
+                    .await
+                    .expect("fixed object lock probe");
+                let mut capacity_probe = tokio::spawn(async move { capacity_lock.get_write_lock(Duration::from_secs(30)).await });
+                let mut object_probe = tokio::spawn(async move { object_lock.get_write_lock(Duration::from_secs(30)).await });
+                for (label, probe) in [("capacity", &mut capacity_probe), ("fixed object", &mut object_probe)] {
+                    assert!(
+                        tokio::time::timeout(Duration::from_millis(100), probe).await.is_err(),
+                        "canceling the scanner waiter must retain its {label} fence while rename is pending"
+                    );
+                }
+                tail.release();
+                drop(tail);
+                for probe in [capacity_probe, object_probe] {
+                    drop(
+                        tokio::time::timeout(Duration::from_secs(30), probe)
+                            .await
+                            .expect("publication fence must drain after rename")
+                            .expect("lock probe must not panic")
+                            .expect("publication fence must eventually be released"),
+                    );
+                }
+                let mut reader = writer.pools[0].disk_set[set_index]
+                    .get_object_reader(RUSTFS_META_BUCKET, object, None, HeaderMap::new(), &ObjectOptions::default())
+                    .await
+                    .expect("canceled waiter must leave the committed replica readable");
+                let mut actual = Vec::new();
+                reader
+                    .read_to_end(&mut actual)
+                    .await
+                    .expect("read the full native replica after tail drain");
+                assert_eq!(actual, body);
+            })
+            .await;
+        });
+    }
+
+    #[test]
+    #[serial_test::serial]
+    fn scanner_backlog_cas_rejects_lost_capacity_lease_before_publication() {
+        run_large_stack_current_thread_async_test("scanner-backlog-lease-loss", async || {
+            let (_temp_dirs, writer, other) = test_three_pool_stores_with_isolated_node_contexts(None).await;
+            let object = "buckets/.scanner-pause-backlog.json";
+            let body = b"native source before lease loss".to_vec();
+            let original = writer
+                .save_scanner_pause_backlog_replica(2, 1, body.clone(), Default::default())
+                .await
+                .expect("seed the exact native replica set");
+            let (lossy, refresh_calls) = store_with_capacity_lease_loss(&other).await;
+            let barrier = PutObjectCommitBarrier::install(RUSTFS_META_BUCKET, object, PutObjectCommitPause::BeforeQuotaRename);
+            let put = tokio::spawn(async move {
+                lossy
+                    .save_scanner_pause_backlog_replica(
+                        2,
+                        1,
+                        b"must not commit after lease loss".to_vec(),
+                        crate::storage_api_contracts::object::HTTPPreconditions {
+                            if_match: original.etag,
+                            ..Default::default()
+                        },
+                    )
+                    .await
+            });
+            tokio::time::timeout(Duration::from_secs(30), barrier.wait_until_paused())
+                .await
+                .expect("native CAS must reach its commit barrier");
+            tokio::time::pause();
+            tokio::task::yield_now().await;
+            refresh_calls.arm();
+            tokio::time::advance(Duration::from_secs(11)).await;
+            tokio::task::yield_now().await;
+            assert!(
+                refresh_calls.load(Ordering::Acquire) > 0,
+                "the durable metadata lease must lose refresh quorum"
+            );
+            barrier.release();
+            tokio::time::resume();
+            let err = tokio::time::timeout(Duration::from_secs(30), put)
+                .await
+                .expect("native CAS must finish after the barrier release")
+                .expect("native CAS task must not panic")
+                .expect_err("a lost outer capacity lease must reject native publication");
+            assert!(
+                matches!(err, crate::error::Error::NamespaceLockQuorumUnavailable { .. }),
+                "unexpected lease error: {err}"
+            );
+            drop(barrier);
+            let mut reader = writer.pools[2].disk_set[1]
+                .get_object_reader(RUSTFS_META_BUCKET, object, None, HeaderMap::new(), &ObjectOptions::default())
+                .await
+                .expect("the preexisting replica must survive lease loss");
+            let mut actual = Vec::new();
+            reader.read_to_end(&mut actual).await.expect("read the full retained replica");
+            assert_eq!(actual, body);
+        });
+    }
+
+    #[test]
+    #[serial_test::serial]
+    fn scanner_backlog_cas_rejects_a_retiring_source_on_a_stale_node() {
+        run_large_stack_current_thread_async_test("scanner-backlog-source-fence", async || {
+            let (_temp_dirs, store, writer) = test_three_pool_stores_with_three_disk_sets_with_isolated_node_contexts(None).await;
+            let object = "buckets/.scanner-pause-backlog.json";
+            let body = b"frozen native scanner replica".to_vec();
+            let source_set_index = (writer.pools[0].get_disks_by_key(object).set_index + 1) % writer.pools[0].disk_set.len();
+            assert_ne!(
+                source_set_index,
+                writer.pools[0].get_disks_by_key(object).set_index,
+                "exercise a non-routed native set"
+            );
+            let original = writer
+                .save_scanner_pause_backlog_replica(
+                    0,
+                    source_set_index,
+                    body.clone(),
+                    crate::storage_api_contracts::object::HTTPPreconditions {
+                        if_none_match: Some("*".to_string()),
+                        ..Default::default()
+                    },
+                )
+                .await
+                .expect("seed a native scanner replica before retirement");
+            assert!(
+                writer
+                    .scanner_pause_backlog_writable_set_disks()
+                    .await
+                    .iter()
+                    .any(|set| set.pool_index == 0)
+            );
+
+            let layout = DecommissionErasureLayout { data: 1, parity: 0 };
+            let target_total = body.len() * 8;
+            set_decommission_capacity_info_overrides_for_test(
+                store.id,
+                vec![vec![
+                    DecommissionPoolCapacityInfo::for_test(0, layout, 0, body.len() * 2, body.len() * 2),
+                    DecommissionPoolCapacityInfo::for_test(1, layout, 0, target_total, target_total),
+                    DecommissionPoolCapacityInfo::for_test(2, layout, target_total, target_total, 0),
+                ]],
+            );
+            store
+                .save_current_pool_meta_for_decommission_start(&[0], Vec::new())
+                .await
+                .expect("another node durably retires the selected source");
+            assert!(
+                writer.pool_meta.read().await.pools[0].decommission.is_none(),
+                "the writer must retain a stale snapshot"
+            );
+            assert!(
+                writer
+                    .scanner_pause_backlog_writable_set_disks()
+                    .await
+                    .iter()
+                    .any(|set| set.pool_index == 0)
+            );
+
+            let result = writer
+                .save_scanner_pause_backlog_replica(
+                    0,
+                    source_set_index,
+                    b"late native scanner update".to_vec(),
+                    crate::storage_api_contracts::object::HTTPPreconditions {
+                        if_match: original.etag.clone(),
+                        ..Default::default()
+                    },
+                )
+                .await;
+            assert!(
+                matches!(result, Err(crate::error::Error::SlowDown)),
+                "late native source publication must fail: {result:?}"
+            );
+            let mut source = writer.pools[0].disk_set[source_set_index]
+                .get_object_reader(RUSTFS_META_BUCKET, object, None, HeaderMap::new(), &ObjectOptions::default())
+                .await
+                .expect("the original source must remain readable");
+            assert_eq!(source.object_info.etag, original.etag);
+            let mut actual = Vec::new();
+            source
+                .read_to_end(&mut actual)
+                .await
+                .expect("read the entire retained source");
+            assert_eq!(actual, body);
+
+            for set in &writer.pools[2].disk_set {
+                let target_body = format!("surviving native scanner set {}", set.set_index).into_bytes();
+                let committed = writer
+                    .save_scanner_pause_backlog_replica(
+                        2,
+                        set.set_index,
+                        target_body.clone(),
+                        crate::storage_api_contracts::object::HTTPPreconditions {
+                            if_none_match: Some("*".to_string()),
+                            ..Default::default()
+                        },
+                    )
+                    .await
+                    .expect("every reserved healthy target set must still accept scanner replicas");
+                let conflict = writer
+                    .save_scanner_pause_backlog_replica(
+                        2,
+                        set.set_index,
+                        b"must not bypass CAS".to_vec(),
+                        crate::storage_api_contracts::object::HTTPPreconditions {
+                            if_match: Some("stale-native-revision".to_string()),
+                            ..Default::default()
+                        },
+                    )
+                    .await
+                    .expect_err("capacity admission must retain the native writer's CAS");
+                assert!(matches!(conflict, crate::error::Error::PreconditionFailed));
+                let mut target = set
+                    .get_object_reader(RUSTFS_META_BUCKET, object, None, HeaderMap::new(), &ObjectOptions::default())
+                    .await
+                    .expect("read the actual replica set, not the hash-routed set");
+                assert_eq!(target.object_info.etag, committed.etag);
+                let mut actual = Vec::new();
+                target
+                    .read_to_end(&mut actual)
+                    .await
+                    .expect("read the complete native target");
+                assert_eq!(actual, target_body);
+            }
+            for (pool_index, set_index) in [(writer.pools.len(), 0), (0, writer.pools[0].disk_set.len())] {
+                assert!(matches!(
+                    writer
+                        .save_scanner_pause_backlog_replica(pool_index, set_index, Vec::new(), Default::default())
+                        .await,
+                    Err(crate::error::Error::InvalidArgument(_, _, _))
+                ));
+            }
+            store
+                .decommission_cancel(0)
+                .await
+                .expect("cancel retirement before restoring native membership");
+            writer
+                .save_scanner_pause_backlog_replica(
+                    0,
+                    source_set_index,
+                    b"canceled source membership repair".to_vec(),
+                    crate::storage_api_contracts::object::HTTPPreconditions {
+                        if_match: original.etag,
+                        ..Default::default()
+                    },
+                )
+                .await
+                .expect("cancel must retain scanner's existing native membership repair contract");
+        });
+    }
+
+    #[test]
+    #[serial_test::serial]
+    fn scanner_backlog_native_replica_reconciles_capacity_and_cleans_source() {
+        run_large_stack_current_thread_async_test("scanner-backlog-reconcile", async || {
+            let (_temp_dirs, store, other_store) =
+                test_three_pool_stores_with_three_disk_sets_with_isolated_node_contexts(None).await;
+            let object = "buckets/.scanner-pause-backlog.json";
+            let body = br#"{"schemaVersion":1,"generation":2}"#.to_vec();
+            let old_body = br#"{"schemaVersion":1,"generation":1}"#.to_vec();
+            let source_time = time::OffsetDateTime::UNIX_EPOCH + time::Duration::seconds(20);
+            let target_time = time::OffsetDateTime::UNIX_EPOCH + time::Duration::seconds(10);
+            for (pool_index, payload, mod_time) in [(0, body.clone(), source_time), (2, old_body, target_time)] {
+                store.pools[pool_index]
+                    .put_object(
+                        RUSTFS_META_BUCKET,
+                        object,
+                        &mut PutObjReader::from_vec(payload),
+                        &ObjectOptions {
+                            max_parity: true,
+                            mod_time: Some(mod_time),
+                            ..Default::default()
+                        },
+                    )
+                    .await
+                    .expect("seed native scanner replicas with independent write times");
+            }
+            let layout = DecommissionErasureLayout { data: 1, parity: 0 };
+            let target_total = body.len() * 8;
+            let capacities = vec![
+                DecommissionPoolCapacityInfo::for_test(0, layout, 0, body.len() * 2, body.len() * 2),
+                DecommissionPoolCapacityInfo::for_test(1, layout, 0, target_total, target_total),
+                DecommissionPoolCapacityInfo::for_test(2, layout, target_total, target_total, 0),
+            ];
+            set_decommission_capacity_info_overrides_for_test(store.id, vec![capacities.clone()]);
+            store
+                .save_current_pool_meta_for_decommission_start(&[0], Vec::new())
+                .await
+                .expect("activate the source reservation");
+            let owner = decommission_capacity_owner(&*store.pool_meta.read().await);
+            let source_reader = store.pools[0]
+                .get_object_reader(
+                    RUSTFS_META_BUCKET,
+                    object,
+                    None,
+                    HeaderMap::new(),
+                    &ObjectOptions {
+                        no_lock: true,
+                        data_movement: true,
+                        raw_data_movement_read: true,
+                        ..Default::default()
+                    },
+                )
+                .await
+                .expect("read the frozen source replica");
+            let conflict = data_movement::migrate_decommission_object(
+                Arc::clone(&store),
+                0,
+                RUSTFS_META_BUCKET.to_string(),
+                source_reader,
+                None,
+                "scanner_backlog_conflict",
+                Some(owner),
+            )
+            .await
+            .expect_err("a different older native ledger must retain its source and capacity intent");
+            assert!(conflict.to_string().contains("Precondition failed"), "unexpected conflict: {conflict}");
+            let mut persisted = crate::core::pools::PoolMeta::default();
+            persisted
+                .load_no_lock_from_replicas(store.pools.clone())
+                .await
+                .expect("reload the unresolved intent");
+            assert_eq!(
+                persisted.pools[0]
+                    .decommission
+                    .as_ref()
+                    .expect("source state")
+                    .capacity_reservation
+                    .as_ref()
+                    .expect("durable capacity")
+                    .pending_target_physical_bytes,
+                body.len()
+            );
+            let previous = store.pools[2]
+                .get_object_info(RUSTFS_META_BUCKET, object, &ObjectOptions::default())
+                .await
+                .expect("read the native writer's CAS revision");
+            let replacement = store.pools[2]
+                .put_object(
+                    RUSTFS_META_BUCKET,
+                    object,
+                    &mut PutObjReader::from_vec(body.clone()),
+                    &ObjectOptions {
+                        max_parity: true,
+                        mod_time: Some(target_time),
+                        http_preconditions: Some(crate::storage_api_contracts::object::HTTPPreconditions {
+                            if_match: previous.etag,
+                            ..Default::default()
+                        }),
+                        ..Default::default()
+                    },
+                )
+                .await
+                .expect("native scanner CAS converges the payload without a migration marker");
+            assert!(!data_movement::is_owned_data_movement_target(&replacement));
+            *other_store.pool_meta.write().await = persisted;
+            set_decommission_capacity_info_overrides_for_test(other_store.id, vec![capacities]);
+            tokio::time::timeout(
+                Duration::from_secs(30),
+                other_store.decommission_entry_for_test(
+                    0,
+                    MetaCacheEntry {
+                        name: object.to_string(),
+                        ..Default::default()
+                    },
+                    RUSTFS_META_BUCKET.to_string(),
+                    other_store.pools[0].get_disks_by_key(object),
+                ),
+            )
+            .await
+            .expect("replica conflict recovery must be bounded")
+            .expect("identical native replica should finish migration on the reloaded node");
+            let mut reconciled = crate::core::pools::PoolMeta::default();
+            reconciled
+                .load_no_lock_from_replicas(other_store.pools.clone())
+                .await
+                .expect("reload reconciled capacity");
+            let reservation = reconciled.pools[0]
+                .decommission
+                .as_ref()
+                .expect("source state")
+                .capacity_reservation
+                .as_ref()
+                .expect("reconciled capacity");
+            assert_eq!(reservation.pending_target_physical_bytes, 0);
+            assert_eq!(reservation.committed_data_bytes, body.len());
+            assert_eq!(reservation.consumed_target_physical_bytes, body.len());
+            assert!(reservation.targets.iter().all(|target| target.pending_mutation_id.is_none()));
+            assert_eq!(
+                other_store.pool_meta.read().await.pools[0]
+                    .decommission
+                    .as_ref()
+                    .expect("worker progress")
+                    .items_decommission_failed,
+                0
+            );
+            let missing = other_store.pools[0]
+                .get_object_info(RUSTFS_META_BUCKET, object, &ObjectOptions::default())
+                .await
+                .expect_err("the source should be cleaned only after equivalent-target capacity reconciliation");
+            assert!(crate::error::is_err_object_not_found(&missing));
+            let mut target_reader = other_store.pools[2]
+                .get_object_reader(RUSTFS_META_BUCKET, object, None, HeaderMap::new(), &ObjectOptions::default())
+                .await
+                .expect("the surviving replica should remain readable");
+            assert_eq!(
+                target_reader.object_info.mod_time,
+                Some(target_time),
+                "recovery must not overwrite the native target"
+            );
+            let mut actual = Vec::new();
+            target_reader
+                .read_to_end(&mut actual)
+                .await
+                .expect("read surviving ledger bytes");
+            assert_eq!(actual, body);
+        });
+    }
+
     #[test]
     #[serial_test::serial]
     fn data_movement_equivalent_target_reconciles_published_capacity_after_restart() {

crates/ecstore/src/data_movement/mod.rs

@@ -984,6 +984,24 @@ fn is_superseding_unversioned_data_movement_object(source: &ObjectInfo, target:
             .is_some_and(|(source_time, target_time)| target_time > source_time)
 }
 
+fn is_equivalent_scanner_backlog_replica(source: &ObjectInfo, target: &ObjectInfo, compare_part_checksums: bool) -> bool {
+    // Scanner publishes this exact payload to surviving sets with CAS. Each
+    // set assigns its own write time; that timestamp is not a ledger generation.
+    // Accept only an identical, known unversioned identity, never a different
+    // record based on timestamp ordering or a similarly named user object.
+    source.bucket == crate::disk::RUSTFS_META_BUCKET
+        && target.bucket == source.bucket
+        && source.name == "buckets/.scanner-pause-backlog.json"
+        && target.name == source.name
+        && is_unversioned_data_movement_object(source)
+        && is_unversioned_data_movement_object(target)
+        && !source.delete_marker
+        && source.mod_time.is_some()
+        && target.mod_time.is_some()
+        && source.etag.as_ref().is_some_and(|etag| !etag.is_empty())
+        && is_equivalent_data_movement_object_identity(source, target, false, compare_part_checksums)
+}
+
 fn is_data_movement_upload_takeover_target(source: &ObjectInfo, target: &ObjectInfo, compare_part_checksums: bool) -> bool {
     let identity = data_movement_upload_identity(source);
     source.mod_time.is_some()
@@ -1453,7 +1471,9 @@ fn resolve_data_movement_overwrite_resume_result_for(
         return Ok(true);
     }
 
-    Ok(matches!(err, Error::PreconditionFailed) && is_superseding_unversioned_data_movement_object(source, &target))
+    Ok(matches!(err, Error::PreconditionFailed)
+        && (is_equivalent_scanner_backlog_replica(source, &target, compare_part_checksums)
+            || is_superseding_unversioned_data_movement_object(source, &target)))
 }
 
 #[derive(Clone, Copy)]
@@ -2673,7 +2693,7 @@ mod tests {
             ]),
             ..Default::default()
         };
-        assert!(!object_info.is_multipart());
+        assert!(object_info.is_multipart());
         assert!(should_use_multipart_data_movement(&object_info, false));
 
         let single_nonstandard_part = ObjectInfo {
@@ -3050,7 +3070,7 @@ mod tests {
             ..Default::default()
         };
 
-        assert!(!object_info.is_multipart());
+        assert!(object_info.is_multipart());
         assert!(object_info.parts.iter().any(|part| part.checksums.is_some()));
         let opts = data_movement_put_object_opts(&object_info, 0);
         assert!(!rustfs_utils::http::contains_key_str(&opts.user_defined, SUFFIX_PART_CHECKSUMS));
@@ -3288,6 +3308,132 @@ mod tests {
         assert!(overwrite_resume_for_target(&source, source.clone()));
     }
 
+    fn scanner_backlog_replica_pair() -> (ObjectInfo, ObjectInfo) {
+        let source = ObjectInfo {
+            bucket: crate::disk::RUSTFS_META_BUCKET.to_string(),
+            name: "buckets/.scanner-pause-backlog.json".to_string(),
+            version_id: None,
+            mod_time: Some(OffsetDateTime::UNIX_EPOCH + time::Duration::SECOND),
+            ..overwrite_equivalence_source()
+        };
+        let target = ObjectInfo {
+            mod_time: Some(OffsetDateTime::UNIX_EPOCH),
+            ..source.clone()
+        };
+        (source, target)
+    }
+
+    fn scanner_backlog_precondition_resumes(source: &ObjectInfo, target: ObjectInfo) -> bool {
+        resolve_data_movement_overwrite_resume_result_for(&Error::PreconditionFailed, Ok(Some(target)), source, 0, 1, true)
+            .expect("scanner replica conflict should be adjudicated")
+    }
+
+    #[test]
+    fn test_scanner_backlog_resume_accepts_identical_native_replica_with_older_write_time() {
+        let (source, target) = scanner_backlog_replica_pair();
+        assert!(!is_owned_data_movement_target(&target), "native scanner writes are not migration copies");
+        assert!(!is_equivalent_data_movement_object(&source, &target));
+        assert!(
+            scanner_backlog_precondition_resumes(&source, target),
+            "identical ledger payloads have replica-local write times, not distinct committed generations"
+        );
+    }
+
+    #[test]
+    fn test_scanner_backlog_resume_rejects_changed_payload_or_metadata() {
+        let (source, target) = scanner_backlog_replica_pair();
+        let mut different_etag = target.clone();
+        different_etag.etag = Some("different-ledger-generation".to_string());
+        let mut different_size = target.clone();
+        different_size.size += 1;
+        let mut different_checksum = target.clone();
+        different_checksum.checksum = Some(Bytes::from_static(b"different-checksum"));
+        let mut different_metadata = target.clone();
+        Arc::make_mut(&mut different_metadata.user_defined).insert("x-amz-meta-key".to_string(), "different".to_string());
+        let mut different_tags = target.clone();
+        different_tags.user_tags = Arc::new("tag=changed".to_string());
+        let mut different_parts = target.clone();
+        Arc::make_mut(&mut different_parts.parts)[0].etag = "different-part".to_string();
+        let mut different_tier = target;
+        different_tier.transitioned_object.tier = "different-tier".to_string();
+        for (label, different) in [
+            ("etag", different_etag),
+            ("size", different_size),
+            ("checksum", different_checksum),
+            ("metadata", different_metadata),
+            ("tags", different_tags),
+            ("parts", different_parts),
+            ("tier", different_tier),
+        ] {
+            assert!(
+                !scanner_backlog_precondition_resumes(&source, different),
+                "replica-local timestamps do not authorize a changed {label}"
+            );
+        }
+    }
+
+    #[test]
+    fn test_scanner_backlog_resume_rejects_other_namespaces_and_incomplete_identity() {
+        let (source, target) = scanner_backlog_replica_pair();
+        for (bucket, name) in [
+            ("user-bucket", "buckets/.scanner-pause-backlog.json"),
+            (crate::disk::RUSTFS_META_BUCKET, "buckets/.scanner-pause-backlog.json.bkp"),
+            (crate::disk::RUSTFS_META_BUCKET, "buckets/.usage-cache.bin"),
+        ] {
+            let mut source = source.clone();
+            let mut target = target.clone();
+            for replica in [&mut source, &mut target] {
+                replica.bucket = bucket.to_string();
+                replica.name = name.to_string();
+            }
+            assert!(!scanner_backlog_precondition_resumes(&source, target), "out-of-scope key {bucket}/{name}");
+        }
+        for missing in ["etag", "empty-etag", "source-time", "target-time", "version", "delete-marker"] {
+            let mut source = source.clone();
+            let mut target = target.clone();
+            match missing {
+                "etag" => {
+                    source.etag = None;
+                    target.etag = None;
+                }
+                "empty-etag" => {
+                    source.etag = Some(String::new());
+                    target.etag = Some(String::new());
+                }
+                "source-time" => source.mod_time = None,
+                "target-time" => target.mod_time = None,
+                "version" => {
+                    source.version_id = Some(Uuid::from_u128(1));
+                    target.version_id = source.version_id;
+                }
+                "delete-marker" => {
+                    source.delete_marker = true;
+                    target.delete_marker = true;
+                }
+                _ => unreachable!("all identity variants are enumerated above"),
+            }
+            assert!(!scanner_backlog_precondition_resumes(&source, target), "unsupported identity: {missing}");
+        }
+    }
+
+    #[test]
+    fn test_scanner_backlog_resume_requires_a_cross_pool_precondition_conflict() {
+        let (source, target) = scanner_backlog_replica_pair();
+        for (err, target_pool) in [
+            (Error::PreconditionFailed, 0),
+            (Error::SlowDown, 1),
+            (
+                Error::InvalidUploadID(source.bucket.clone(), source.name.clone(), "upload".to_string()),
+                1,
+            ),
+        ] {
+            assert!(
+                !resolve_data_movement_overwrite_resume_result_for(&err, Ok(Some(target.clone())), &source, 0, target_pool, true)
+                    .expect("non-resumable conflict should return false")
+            );
+        }
+    }
+
     #[test]
     fn test_data_movement_overwrite_resume_accepts_part_mod_time_drift() {
         let source = overwrite_equivalence_source();

crates/ecstore/src/disk/disk_store.rs

@@ -195,6 +195,13 @@ fn resolve_drive_timeout_profile_from_env() -> DriveTimeoutProfile {
     DriveTimeoutProfile::parse(rustfs_config::DEFAULT_DRIVE_TIMEOUT_PROFILE).unwrap_or(DriveTimeoutProfile::Default)
 }
 
+#[cfg(test)]
+tokio::task_local! {
+    /// Artificial `disk_info` latency for tests that pin how the admin storage
+    /// walk composes per-drive probe time.
+    pub(crate) static DISK_INFO_PROBE_DELAY_FOR_TEST: Duration;
+}
+
 fn get_drive_timeout_profile() -> DriveTimeoutProfile {
     #[cfg(test)]
     {
@@ -324,6 +331,46 @@ impl DiskStoreRenameDataExt for LocalDiskWrapper {
 }
 
 impl LocalDiskWrapper {
+    pub(in crate::disk) async fn delete_version_with_namespace_owner(
+        &self,
+        volume: &str,
+        path: &str,
+        fi: FileInfo,
+        force_del_marker: bool,
+        opts: DeleteOptions,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Result<()> {
+        self.track_disk_health_mutation(
+            "delete_version",
+            DiskMetricMutation::Delete,
+            || async {
+                Box::pin(
+                    self.disk
+                        .delete_version_with_namespace_owner(volume, path, fi, force_del_marker, opts, namespace_owner),
+                )
+                .await
+            },
+            get_max_timeout_duration(),
+        )
+        .await
+    }
+
+    pub(in crate::disk) async fn delete_with_namespace_owner(
+        &self,
+        volume: &str,
+        path: &str,
+        opts: DeleteOptions,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Result<()> {
+        self.track_disk_health_mutation(
+            "delete",
+            DiskMetricMutation::Delete,
+            || async { Box::pin(self.disk.delete_with_namespace_owner(volume, path, opts, namespace_owner)).await },
+            get_max_timeout_duration(),
+        )
+        .await
+    }
+
     pub(in crate::disk) async fn undo_write_with_namespace_owner(
         &self,
         volume: &str,
@@ -1996,6 +2043,10 @@ impl DiskAPI for LocalDiskWrapper {
             .track_disk_health_with_op_and_timeout_action(
                 "disk_info",
                 || async {
+                    #[cfg(test)]
+                    if let Ok(delay) = DISK_INFO_PROBE_DELAY_FOR_TEST.try_with(|delay| *delay) {
+                        tokio::time::sleep(delay).await;
+                    }
                     let result = self.disk.disk_info(opts).await?;
 
                     if let Some(current_disk_id) = *self.disk_id.read().await

crates/ecstore/src/disk/error.rs

@@ -425,6 +425,7 @@ impl From<rustfs_filemeta::Error> for DiskError {
             rustfs_filemeta::Error::FileVersionNotFound => DiskError::FileVersionNotFound,
             rustfs_filemeta::Error::FileCorrupt => DiskError::FileCorrupt,
             rustfs_filemeta::Error::MethodNotAllowed => DiskError::MethodNotAllowed,
+            rustfs_filemeta::Error::MaxVersionsExceeded => DiskError::MaxVersionsExceeded,
             e => DiskError::other(e),
         }
     }

crates/ecstore/src/disk/mod.rs

@@ -732,6 +732,46 @@ impl Disk {
         }
     }
 
+    pub(crate) async fn delete_version_with_namespace_owner(
+        &self,
+        volume: &str,
+        path: &str,
+        fi: FileInfo,
+        force_del_marker: bool,
+        opts: DeleteOptions,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Result<()> {
+        match self {
+            Self::Local(disk) => {
+                disk.delete_version_with_namespace_owner(volume, path, fi, force_del_marker, opts, namespace_owner)
+                    .await
+            }
+            Self::Remote(disk) => {
+                let result = disk.delete_version(volume, path, fi, force_del_marker, opts).await;
+                // This is sender lifetime only, not proof of a remote physical drain.
+                drop(namespace_owner);
+                result
+            }
+        }
+    }
+
+    pub(crate) async fn delete_with_namespace_owner(
+        &self,
+        volume: &str,
+        path: &str,
+        opts: DeleteOptions,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Result<()> {
+        match self {
+            Self::Local(disk) => disk.delete_with_namespace_owner(volume, path, opts, namespace_owner).await,
+            Self::Remote(disk) => {
+                let result = disk.delete(volume, path, opts).await;
+                drop(namespace_owner);
+                result
+            }
+        }
+    }
+
     /// Keep local undo publication owned independently of the wrapper deadline.
     /// Remote undo retains its existing RPC contract; this is not a remote drain proof.
     pub(crate) async fn undo_write_with_namespace_owner(

crates/ecstore/src/disk/os.rs

@@ -92,6 +92,9 @@ pub(crate) mod fsync_dir_recorder {
     static LIMITED: Mutex<Vec<PathBuf>> = Mutex::new(Vec::new());
     static GROUPED: Mutex<Vec<(PathBuf, usize)>> = Mutex::new(Vec::new());
     #[cfg(unix)]
+    static FAILURES: std::sync::LazyLock<Mutex<HashMap<PathBuf, io::ErrorKind>>> =
+        std::sync::LazyLock::new(|| Mutex::new(HashMap::new()));
+    #[cfg(unix)]
     static BEFORE_LIMITED: std::sync::LazyLock<Mutex<HashMap<PathBuf, Hook>>> =
         std::sync::LazyLock::new(|| Mutex::new(HashMap::new()));
     static BEFORE_GROUP_BATCH: std::sync::LazyLock<Mutex<HashMap<PathBuf, Hook>>> =
@@ -151,6 +154,19 @@ pub(crate) mod fsync_dir_recorder {
         contains_path(&RECORDED.lock().expect("fsync dir recorder poisoned"), dir)
     }
 
+    #[cfg(unix)]
+    pub(crate) fn set_failure(dir: &Path, kind: io::ErrorKind) {
+        FAILURES
+            .lock()
+            .expect("fsync dir failure hook poisoned")
+            .insert(dir.to_path_buf(), kind);
+    }
+
+    #[cfg(unix)]
+    pub(crate) fn take_failure(dir: &Path) -> Option<io::ErrorKind> {
+        remove_path_keyed(&FAILURES, dir, "fsync dir failure hook poisoned")
+    }
+
     #[cfg(unix)]
     pub(crate) fn record_limited(dir: &Path) {
         record_path(&LIMITED, dir, "limited fsync dir recorder");
@@ -291,6 +307,46 @@ pub(crate) mod prepared_publication_test_hooks {
             hook();
         }
     }
+
+    #[cfg(test)]
+    type RenameDestinationHook = Box<dyn FnOnce(&Path) + Send>;
+    #[cfg(test)]
+    static RENAME_DESTINATIONS: LazyLock<Mutex<HashMap<PathBuf, RenameDestinationHook>>> =
+        LazyLock::new(|| Mutex::new(HashMap::new()));
+
+    #[cfg(test)]
+    pub(crate) struct RenameDestinationGuard(PathBuf);
+
+    #[cfg(test)]
+    impl Drop for RenameDestinationGuard {
+        fn drop(&mut self) {
+            RENAME_DESTINATIONS.lock().remove(&self.0);
+        }
+    }
+
+    #[cfg(test)]
+    pub(crate) fn observe_rename_destination(source: &Path, hook: impl FnOnce(&Path) + Send + 'static) -> RenameDestinationGuard {
+        assert!(
+            RENAME_DESTINATIONS
+                .lock()
+                .insert(source.to_path_buf(), Box::new(hook))
+                .is_none()
+        );
+        RenameDestinationGuard(source.to_path_buf())
+    }
+
+    #[cfg(test)]
+    pub(crate) async fn drain_namespace_key(path: &Path) {
+        drop(super::acquire_namespace_mutation_lease(path).await);
+    }
+
+    #[cfg(test)]
+    pub(super) fn run_rename_destination(source: &Path, destination: &Path) {
+        let hook = RENAME_DESTINATIONS.lock().remove(source);
+        if let Some(hook) = hook {
+            hook(destination);
+        }
+    }
 }
 
 /// Controlled application-test pause at an existing physical executor boundary.
@@ -434,6 +490,10 @@ pub fn fsync_dir_std(dir: impl AsRef<Path>) -> io::Result<()> {
     fsync_dir_recorder::record(dir.as_ref());
     #[cfg(unix)]
     {
+        #[cfg(test)]
+        if let Some(kind) = fsync_dir_recorder::take_failure(dir.as_ref()) {
+            return Err(io::Error::from(kind));
+        }
         std::fs::File::open(dir.as_ref())?.sync_all()?;
     }
     #[cfg(not(unix))]
@@ -1409,7 +1469,7 @@ async fn acquire_namespace_mutation_lease(path: &Path) -> Arc<NamespaceMutationL
     acquire_namespace_mutation_lease_with_owner(path, None).await
 }
 
-async fn acquire_namespace_mutation_lease_with_owner(
+pub(in crate::disk) async fn acquire_namespace_mutation_lease_with_owner(
     path: &Path,
     namespace_owner: Option<Arc<dyn Send + Sync>>,
 ) -> Arc<NamespaceMutationLease> {
@@ -1984,6 +2044,42 @@ pub(crate) async fn remove_dir_with_owner(
     run_blocking_namespace_operation(lease, move || std::fs::remove_dir(path)).await
 }
 
+/// Preserve raw rename semantics while retaining a counted owner in the syscall.
+/// Unlike reliable rename, this never creates parents or retries a missing source.
+pub(in crate::disk) async fn rename_with_namespace_owner(
+    src: &Path,
+    dst: &Path,
+    namespace_owner: Option<Arc<dyn Send + Sync>>,
+) -> io::Result<()> {
+    if namespace_owner.is_none() {
+        return tokio::fs::rename(src, dst).await;
+    }
+    let src = src.to_path_buf();
+    let dst = dst.to_path_buf();
+    let lease = acquire_namespace_mutation_lease_with_owner(&dst, namespace_owner).await;
+    run_blocking_namespace_operation(lease, move || {
+        #[cfg(all(test, not(windows)))]
+        {
+            prepared_publication_test_hooks::run(prepared_publication_test_hooks::Stage::Rename, &src);
+            prepared_publication_test_hooks::run(prepared_publication_test_hooks::Stage::Rename, &dst);
+        }
+        std::fs::rename(src, dst)
+    })
+    .await
+}
+
+pub(in crate::disk) async fn create_dir_all_with_namespace_owner(
+    path: &Path,
+    namespace_owner: Option<Arc<dyn Send + Sync>>,
+) -> io::Result<()> {
+    if namespace_owner.is_none() {
+        return tokio::fs::create_dir_all(path).await;
+    }
+    let path = path.to_path_buf();
+    let lease = acquire_namespace_mutation_lease_with_owner(&path, namespace_owner).await;
+    run_blocking_namespace_operation(lease, move || std::fs::create_dir_all(path)).await
+}
+
 #[tracing::instrument(name = "rename_all", level = "debug", skip_all)]
 pub(crate) async fn rename_all_with_owner(
     src_file_path: impl AsRef<Path>,
@@ -2319,6 +2415,8 @@ async fn reliable_rename_inner_with_lease(
         let base_dir = base_dir.clone();
         move || {
             let preparation = prepare_rename_with_retry(&src_file_path, &dst_file_path, &base_dir, &publication_root)?;
+            #[cfg(all(test, not(windows)))]
+            prepared_publication_test_hooks::run_rename_destination(&src_file_path, &dst_file_path);
             #[cfg(all(any(test, feature = "test-util"), not(windows)))]
             {
                 prepared_publication_test_hooks::run(prepared_publication_test_hooks::Stage::Rename, &src_file_path);

crates/ecstore/src/erasure/coding/encode.rs

@@ -431,7 +431,7 @@ impl<'a> MultiWriter<'a> {
             errs = ?self.errs,
             "Erasure encode write quorum unavailable: {summary_text}"
         );
-        Err(std::io::Error::other(format!("Failed to write data: {summary_text}")))
+        Err(write_err.into())
     }
 
     async fn shutdown_writer(writer_opt: &mut Option<BitrotWriterWrapper>, err: &mut Option<Error>) {
@@ -503,7 +503,7 @@ impl<'a> MultiWriter<'a> {
             errs = ?self.errs,
             "Erasure encode shutdown quorum unavailable: {summary_text}"
         );
-        Err(std::io::Error::other(format!("Failed to shutdown writers: {summary_text}")))
+        Err(write_err.into())
     }
 }
 
@@ -1002,6 +1002,7 @@ impl Erasure {
 mod tests {
     use super::*;
     use crate::erasure::coding::{BitrotWriterWrapper, CustomWriter};
+    use crate::error::StorageError;
     use rustfs_rio::HardLimitReader;
     use rustfs_utils::HashAlgorithm;
     use std::future::Future;
@@ -1451,7 +1452,14 @@ mod tests {
             Ok(_) => panic!("writer quorum failure should fail the encode pipeline"),
             Err(err) => err,
         };
-        assert!(err.to_string().contains("Failed to write data"));
+        let err = StorageError::from(err);
+        assert!(matches!(
+            &err,
+            StorageError::Io(source)
+                if source.kind() == std::io::ErrorKind::Other
+                    && source.to_string() == "injected write failure after producer blocks"
+        ));
+        assert!(!err.is_quorum_error());
         tokio::time::timeout(Duration::from_secs(1), reader_dropped)
             .await
             .expect("writer failure should abort the blocked producer")
@@ -1644,7 +1652,7 @@ mod tests {
 
     #[tokio::test]
     async fn multi_writer_short_write_fails_before_shutdown() {
-        let mut writers = vec![Some(bitrot_writer(ShortWriteWriter, 16))];
+        let mut writers = vec![Some(bitrot_writer(ShortWriteWriter, 32))];
         let err = {
             let mut writer = MultiWriter::new(&mut writers, 1);
             writer
@@ -1653,63 +1661,93 @@ mod tests {
                 .expect_err("short writes must fail the shard writer")
         };
 
-        assert!(err.to_string().contains("Failed to write data"));
+        let err = StorageError::from(err);
+        assert!(matches!(&err, StorageError::Io(source) if source.kind() == std::io::ErrorKind::WriteZero));
+        assert!(!err.is_quorum_error());
         assert!(writers[0].is_none(), "short-write shard must be removed before commit");
     }
 
     #[tokio::test]
     async fn multi_writer_reports_fallback_summary_when_only_offline_writers_remain() {
         let mut writers = vec![None, None];
-        let err = {
+        let (err, summary) = {
             let mut writer = MultiWriter::new(&mut writers, 1);
-            writer
+            let err = writer
                 .write(vec![Bytes::from_static(b"offline-a"), Bytes::from_static(b"offline-b")])
                 .await
-                .expect_err("offline writers cannot satisfy write quorum")
+                .expect_err("offline writers cannot satisfy write quorum");
+            let summary = build_write_quorum_failure_summary(&writer.errs, OBJECT_OP_IGNORED_ERRS, writer.write_quorum);
+            (err, format_write_quorum_failure(&summary))
         };
 
-        let err = err.to_string();
+        assert_eq!(
-        assert!(err.contains("Failed to write data"));
+            err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
-        assert!(err.contains("offline-disks=2/2"));
+            Some(&Error::ErasureWriteQuorum),
-        assert!(err.contains("required=1"));
+        );
+        let err = StorageError::from(err);
+        assert_eq!(err, StorageError::ErasureWriteQuorum);
+        assert!(err.is_quorum_error());
+        assert!(summary.contains("offline-disks=2/2"));
+        assert!(summary.contains("required=1"));
 
-        let shutdown_err = {
+        let (shutdown_err, summary) = {
             let mut writer = MultiWriter::new(&mut writers, 1);
-            writer
+            let err = writer
                 .shutdown()
                 .await
-                .expect_err("offline writers cannot satisfy shutdown quorum")
+                .expect_err("offline writers cannot satisfy shutdown quorum");
+            let summary = build_write_quorum_failure_summary(&writer.errs, OBJECT_OP_IGNORED_ERRS, writer.write_quorum);
+            (err, format_write_quorum_failure(&summary))
         };
 
-        let shutdown_err = shutdown_err.to_string();
+        assert_eq!(
-        assert!(shutdown_err.contains("Failed to shutdown writers"));
+            shutdown_err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
-        assert!(shutdown_err.contains("offline-disks=2/2"));
+            Some(&Error::ErasureWriteQuorum),
-        assert!(shutdown_err.contains("required=1"));
+        );
+        let shutdown_err = StorageError::from(shutdown_err);
+        assert_eq!(shutdown_err, StorageError::ErasureWriteQuorum);
+        assert!(shutdown_err.is_quorum_error());
+        assert!(summary.contains("offline-disks=2/2"));
+        assert!(summary.contains("required=1"));
     }
 
     #[tokio::test]
     async fn multi_writer_reports_quorum_failure_when_quorum_exceeds_writer_count() {
         let committed = Arc::new(Mutex::new(Vec::new()));
-        let mut writers = vec![Some(bitrot_writer(DeferredCommitWriter::new(committed), 16))];
+        let mut writers = vec![Some(bitrot_writer(DeferredCommitWriter::new(committed), 32))];
         let mut writer = MultiWriter::new(&mut writers, 2);
 
         let err = writer
             .write(vec![Bytes::from_static(b"quorum impossible")])
             .await
             .expect_err("write quorum above writer count must fail");
-        let err = err.to_string();
+        assert_eq!(
-        assert!(err.contains("Failed to write data"));
+            err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
-        assert!(err.contains("required=2"));
+            Some(&Error::ErasureWriteQuorum),
-        assert!(err.contains("erasure write quorum"));
+        );
+        let err = StorageError::from(err);
+        assert_eq!(err, StorageError::ErasureWriteQuorum);
+        assert!(err.is_quorum_error());
+        let summary = build_write_quorum_failure_summary(&writer.errs, OBJECT_OP_IGNORED_ERRS, writer.write_quorum);
+        let summary = format_write_quorum_failure(&summary);
+        assert!(summary.contains("required=2"));
+        assert!(summary.contains("erasure write quorum"));
 
         let shutdown_err = writer
             .shutdown()
             .await
             .expect_err("shutdown quorum above writer count must fail");
-        let shutdown_err = shutdown_err.to_string();
+        assert_eq!(
-        assert!(shutdown_err.contains("Failed to shutdown writers"));
+            shutdown_err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
-        assert!(shutdown_err.contains("required=2"));
+            Some(&Error::ErasureWriteQuorum),
-        assert!(shutdown_err.contains("erasure write quorum"));
+        );
+        let shutdown_err = StorageError::from(shutdown_err);
+        assert_eq!(shutdown_err, StorageError::ErasureWriteQuorum);
+        assert!(shutdown_err.is_quorum_error());
+        let summary = build_write_quorum_failure_summary(&writer.errs, OBJECT_OP_IGNORED_ERRS, writer.write_quorum);
+        let summary = format_write_quorum_failure(&summary);
+        assert!(summary.contains("required=2"));
+        assert!(summary.contains("erasure write quorum"));
     }
 
     // The production wiring (`MultiWriter::new`) must arm a real deadline by
@@ -1794,7 +1832,13 @@ mod tests {
             .write(four_shards())
             .await
             .expect_err("two stalled writers must fail the write quorum instead of hanging");
-        assert!(err.to_string().contains("Failed to write data"));
+        assert_eq!(
+            err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
+            Some(&Error::ErasureWriteQuorum),
+        );
+        let err = StorageError::from(err);
+        assert_eq!(err, StorageError::ErasureWriteQuorum);
+        assert!(err.is_quorum_error());
     }
 
     // A small object whose bytes were fully buffered leaves `write` succeeding
@@ -1839,7 +1883,13 @@ mod tests {
             .shutdown()
             .await
             .expect_err("two shutdown stalls must fail the shutdown quorum instead of hanging");
-        assert!(err.to_string().contains("Failed to shutdown writers"));
+        assert_eq!(
+            err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
+            Some(&Error::ErasureWriteQuorum),
+        );
+        let err = StorageError::from(err);
+        assert_eq!(err, StorageError::ErasureWriteQuorum);
+        assert!(err.is_quorum_error());
     }
 
     // A slow-but-honest writer that keeps completing shards (delay < stall
@@ -2121,7 +2171,13 @@ mod tests {
             .await
             .expect_err("streaming encode must fail when write quorum is unavailable");
 
-        assert!(err.to_string().contains("Failed to write data"));
+        assert_eq!(
+            err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
+            Some(&Error::ErasureWriteQuorum),
+        );
+        let err = StorageError::from(err);
+        assert_eq!(err, StorageError::ErasureWriteQuorum);
+        assert!(err.is_quorum_error());
     }
 
     #[tokio::test]
@@ -2145,7 +2201,13 @@ mod tests {
             .await
             .expect_err("write quorum failure must fail the inline encode");
 
-        assert!(err.to_string().contains("Failed to write data"));
+        assert_eq!(
+            err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
+            Some(&Error::ErasureWriteQuorum),
+        );
+        let err = StorageError::from(err);
+        assert_eq!(err, StorageError::ErasureWriteQuorum);
+        assert!(err.is_quorum_error());
         assert!(
             committed.lock().expect("committed buffer should be lockable").is_empty(),
             "successful writer must not be committed when write quorum fails before shutdown"
@@ -2173,7 +2235,13 @@ mod tests {
             .await
             .expect_err("shutdown quorum failure must fail the inline encode");
 
-        assert!(err.to_string().contains("Failed to shutdown writers"));
+        let err = StorageError::from(err);
+        assert!(matches!(
+            &err,
+            StorageError::Io(source)
+                if source.kind() == std::io::ErrorKind::Other && source.to_string() == "injected shutdown failure"
+        ));
+        assert!(!err.is_quorum_error());
         assert!(
             !committed.lock().expect("committed buffer should be lockable").is_empty(),
             "the successful writer should have committed before shutdown quorum failure was reported"
@@ -2395,7 +2463,13 @@ mod tests {
             .await
             .expect_err("batched encode must fail when write quorum is unavailable");
 
-        assert!(err.to_string().contains("Failed to write data"));
+        assert_eq!(
+            err.get_ref().and_then(|source| source.downcast_ref::<Error>()),
+            Some(&Error::ErasureWriteQuorum),
+        );
+        let err = StorageError::from(err);
+        assert_eq!(err, StorageError::ErasureWriteQuorum);
+        assert!(err.is_quorum_error());
     }
 
     #[tokio::test]

crates/ecstore/src/error/mod.rs

@@ -23,17 +23,59 @@ use s3s::S3ErrorCode;
 pub type Error = StorageError;
 pub type Result<T> = core::result::Result<T, Error>;
 
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum PoolMetadataFailure {
+    ReadUnavailable,
+    RecoveryRequired,
+    TransactionUnknown,
+    FenceLost,
+}
+
+impl PoolMetadataFailure {
+    fn recovery_hint(self) -> &'static str {
+        match self {
+            Self::ReadUnavailable => "read unavailable; retry after the replicas are readable",
+            Self::TransactionUnknown => "writes remain blocked pending fenced transaction recovery",
+            Self::RecoveryRequired | Self::FenceLost => {
+                "writes remain blocked after a recovery-required replica state; restart after all replicas are readable and consistent, with compatible formats"
+            }
+        }
+    }
+
+    pub fn as_str(self) -> &'static str {
+        match self {
+            Self::ReadUnavailable => "read_unavailable",
+            Self::RecoveryRequired => "recovery_required",
+            Self::TransactionUnknown => "transaction_unknown",
+            Self::FenceLost => "fence_lost",
+        }
+    }
+}
+
+/// Local control-plane context. Keep the existing storage error wire codes;
+/// the HTTP boundary recognizes this typed source, not an error-message prefix.
+#[derive(Debug, Clone, thiserror::Error)]
+#[error("{operation}: pool metadata {hint} ({reason}, {phase}): {detail}", hint = kind.recovery_hint(), reason = kind.as_str(), detail = source.as_ref().map(ToString::to_string).unwrap_or_default())]
+pub struct PoolMetadataError {
+    pub kind: PoolMetadataFailure,
+    pub operation: String,
+    pub phase: &'static str,
+    pub since: time::OffsetDateTime,
+    #[source]
+    pub source: Option<std::sync::Arc<StorageError>>,
+}
+
 /// Keeps high-cardinality diagnostic detail in the error source while making
 /// the rendered `io::Error` stable for quorum aggregation.
 #[derive(Debug)]
 struct StableIoContextError {
-    message: &'static str,
+    message: std::borrow::Cow<'static, str>,
     source: Box<dyn std::error::Error + Send + Sync>,
 }
 
 impl std::fmt::Display for StableIoContextError {
     fn fmt(&self, formatter: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        formatter.write_str(self.message)
+        formatter.write_str(&self.message)
     }
 }
 
@@ -48,7 +90,7 @@ where
     E: Into<Box<dyn std::error::Error + Send + Sync>>,
 {
     std::io::Error::other(StableIoContextError {
-        message,
+        message: message.into(),
         source: source.into(),
     })
 }
@@ -203,6 +245,19 @@ pub enum StorageError {
     NotFirstDisk,
     #[error("first disk wait")]
     FirstDiskWait,
+    #[error(
+        "unsupported pool expansion: an existing single-node single-drive (SNSD) deployment cannot be expanded in place (configured {configured_drives} drive endpoints); restart with the original single local path, or create a new multi-drive deployment and migrate data through S3"
+    )]
+    UnsupportedSnsdExpansion { configured_drives: usize },
+    #[error(
+        "pool topology mismatch: stored {stored_drives} drives with {stored_set_drive_count} drives per erasure set, configured {configured_drives} drives with {configured_set_drive_count} drives per erasure set; an existing pool's drive count and erasure set width cannot be changed in place; restore its original endpoints and RUSTFS_ERASURE_SET_DRIVE_COUNT setting; to expand a multi-drive deployment, append a new pool with at least 2 drive endpoints"
+    )]
+    PoolTopologyMismatch {
+        stored_drives: usize,
+        stored_set_drive_count: usize,
+        configured_drives: usize,
+        configured_set_drive_count: usize,
+    },
 
     // ── Operational ──────────────────────────────────────────────────
     #[error("Storage reached its minimum free drive threshold.")]
@@ -287,6 +342,22 @@ impl From<crate::erasure::coding::ErasureConstructionError> for StorageError {
 }
 
 impl StorageError {
+    pub fn pool_metadata_failure(&self) -> Option<&PoolMetadataError> {
+        let mut current: Option<&(dyn std::error::Error + 'static)> = Some(self);
+        while let Some(error) = current {
+            if let Some(context) = error.downcast_ref::<PoolMetadataError>() {
+                return Some(context);
+            }
+            // io::Error::source skips its boxed context itself.
+            current = if let Some(io) = error.downcast_ref::<std::io::Error>() {
+                io.get_ref().map(|inner| inner as &(dyn std::error::Error + 'static))
+            } else {
+                error.source()
+            };
+        }
+        None
+    }
+
     pub fn other<E>(error: E) -> Self
     where
         E: Into<Box<dyn std::error::Error + Send + Sync>>,
@@ -517,6 +588,7 @@ impl From<rustfs_filemeta::Error> for StorageError {
             rustfs_filemeta::Error::FileVersionNotFound => StorageError::FileVersionNotFound,
             rustfs_filemeta::Error::FileCorrupt => StorageError::FileCorrupt,
             rustfs_filemeta::Error::Unexpected => StorageError::Unexpected,
+            rustfs_filemeta::Error::MaxVersionsExceeded => StorageError::MaxVersionsExceeded,
             rustfs_filemeta::Error::Io(io_error) => io_error.into(),
             _ => StorageError::Io(std::io::Error::other(e)),
         }
@@ -535,7 +607,19 @@ impl PartialEq for StorageError {
 impl Clone for StorageError {
     fn clone(&self) -> Self {
         match self {
-            StorageError::Io(e) => StorageError::Io(std::io::Error::new(e.kind(), e.to_string())),
+            StorageError::Io(e) => {
+                if let Some(context) = self.pool_metadata_failure() {
+                    Self::Io(std::io::Error::new(
+                        e.kind(),
+                        StableIoContextError {
+                            message: e.to_string().into(),
+                            source: Box::new(context.clone()),
+                        },
+                    ))
+                } else {
+                    StorageError::Io(std::io::Error::new(e.kind(), e.to_string()))
+                }
+            }
             StorageError::FaultyDisk => StorageError::FaultyDisk,
             StorageError::DiskFull => StorageError::DiskFull,
             StorageError::VolumeNotFound => StorageError::VolumeNotFound,
@@ -629,6 +713,20 @@ impl Clone for StorageError {
             StorageError::ErasureWriteQuorum => StorageError::ErasureWriteQuorum,
             StorageError::NotFirstDisk => StorageError::NotFirstDisk,
             StorageError::FirstDiskWait => StorageError::FirstDiskWait,
+            StorageError::UnsupportedSnsdExpansion { configured_drives } => StorageError::UnsupportedSnsdExpansion {
+                configured_drives: *configured_drives,
+            },
+            StorageError::PoolTopologyMismatch {
+                stored_drives,
+                stored_set_drive_count,
+                configured_drives,
+                configured_set_drive_count,
+            } => StorageError::PoolTopologyMismatch {
+                stored_drives: *stored_drives,
+                stored_set_drive_count: *stored_set_drive_count,
+                configured_drives: *configured_drives,
+                configured_set_drive_count: *configured_set_drive_count,
+            },
             StorageError::TooManyOpenFiles => StorageError::TooManyOpenFiles,
             StorageError::NoHealRequired => StorageError::NoHealRequired,
             StorageError::Lock(e) => StorageError::Lock(e.clone()),
@@ -662,7 +760,8 @@ impl Clone for StorageError {
 }
 
 impl StorageError {
-    fn code(&self) -> StorageErrorCode {
+    /// Stable classification without error payloads or storage paths.
+    pub fn code(&self) -> StorageErrorCode {
         match self {
             StorageError::Io(_) => StorageErrorCode::Io,
             StorageError::FaultyDisk => StorageErrorCode::FaultyDisk,
@@ -735,6 +834,11 @@ impl StorageError {
             StorageError::ErasureWriteQuorum => StorageErrorCode::ErasureWriteQuorum,
             StorageError::NotFirstDisk => StorageErrorCode::NotFirstDisk,
             StorageError::FirstDiskWait => StorageErrorCode::FirstDiskWait,
+            // Topology diagnostics reuse the existing wire code; they are
+            // not disk errors and must retain their local identity for retry classification.
+            StorageError::UnsupportedSnsdExpansion { .. } | StorageError::PoolTopologyMismatch { .. } => {
+                StorageErrorCode::InvalidArgument
+            }
             StorageError::ConfigNotFound => StorageErrorCode::ConfigNotFound,
             StorageError::TooManyOpenFiles => StorageErrorCode::TooManyOpenFiles,
             StorageError::NoHealRequired => StorageErrorCode::NoHealRequired,
@@ -1215,6 +1319,29 @@ mod tests {
     use super::*;
     use std::io::{Error as IoError, ErrorKind};
 
+    #[test]
+    fn startup_topology_errors_preserve_identity_and_guidance() {
+        for error in [
+            StorageError::UnsupportedSnsdExpansion { configured_drives: 4 },
+            StorageError::PoolTopologyMismatch {
+                stored_drives: 4,
+                stored_set_drive_count: 4,
+                configured_drives: 8,
+                configured_set_drive_count: 8,
+            },
+        ] {
+            let io_error: IoError = error.clone().into();
+            let restored = StorageError::from(io_error);
+            assert_eq!(std::mem::discriminant(&restored), std::mem::discriminant(&error));
+            assert_eq!(restored.to_string(), error.to_string());
+            assert_eq!(restored.code(), StorageErrorCode::InvalidArgument);
+            assert!(
+                restored.narrow_to_disk().is_err(),
+                "startup diagnostics must not become disk/quorum errors"
+            );
+        }
+    }
+
     #[test]
     fn other_preserves_erasure_construction_source_chain() {
         use crate::erasure::coding::ErasureConstructionError;

crates/ecstore/src/layout/disks_layout.rs

@@ -25,6 +25,20 @@ pub(crate) const MAX_ERASURE_SET_DRIVE_COUNT: usize = 16;
 const SET_SIZES: [usize; 15] = [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, MAX_ERASURE_SET_DRIVE_COUNT];
 const ENV_RUSTFS_ERASURE_SET_DRIVE_COUNT: &str = "RUSTFS_ERASURE_SET_DRIVE_COUNT";
 
+#[derive(Debug, thiserror::Error)]
+enum PoolDriveCountError {
+    #[error(
+        "Incorrect number of endpoints provided, size {size}; an erasure pool requires at least {} drive endpoints on one or more nodes; for a standalone single-drive deployment, use a single local path without ellipses",
+        SET_SIZES[0]
+    )]
+    BelowMinimum { size: usize },
+    #[error(
+        "Incorrect number of endpoints provided, size {size}; {}={set_drive_count} requires at least {set_drive_count} drive endpoints per pool",
+        ENV_RUSTFS_ERASURE_SET_DRIVE_COUNT
+    )]
+    BelowSetWidth { size: usize, set_drive_count: usize },
+}
+
 #[derive(Deserialize, Debug, Default)]
 pub struct PoolDisksLayout {
     cmd_line: String,
@@ -132,7 +146,7 @@ impl DisksLayout {
         for arg in args.iter() {
             if !has_ellipses(&[arg]) && args.len() > 1 {
                 return Err(Error::other(
-                    "all args must have ellipses for pool expansion (Invalid arguments specified)",
+                    "all args must have ellipses for pool expansion (Invalid arguments specified); each pool must expand to at least 2 drive endpoints on one or more nodes; a single-drive pool cannot be added to a multi-pool deployment",
                 ));
             }
 
@@ -396,9 +410,11 @@ fn get_set_indexes<T: AsRef<str>>(
     }
 
     for &size in total_sizes {
-        // Check if total_sizes has minimum range upto set_size
+        if size < SET_SIZES[0] {
-        if size < SET_SIZES[0] || size < set_drive_count {
+            return Err(Error::other(PoolDriveCountError::BelowMinimum { size }));
-            return Err(Error::other(format!("Incorrect number of endpoints provided, size {size}")));
+        }
+        if size < set_drive_count {
+            return Err(Error::other(PoolDriveCountError::BelowSetWidth { size, set_drive_count }));
         }
     }
 
@@ -707,7 +723,7 @@ mod test {
                 arg: "http://rustfs{2...3}/export/set{1...0}",
                 ..Default::default()
             },
-            // Range cannot be smaller than 4 minimum.
+            // Ranges must use three dots.
             TestCase {
                 num: 4,
                 arg: "/export{1..2}",
@@ -926,11 +942,146 @@ mod test {
         }
     }
 
+    #[test]
+    fn pool_expansion_accepts_single_node_multi_drive_pools() {
+        temp_env::with_var(ENV_RUSTFS_ERASURE_SET_DRIVE_COUNT, Some("0"), || {
+            for (volumes, drives) in [
+                (["http://node1:9000/data{1...2}", "http://node2:9000/data{1...2}"], 2),
+                (["http://node1:9000/data{1...4}", "http://node2:9000/data{1...4}"], 4),
+                (["http://node{1...4}:9000/data", "http://node5:9000/data{1...4}"], 4),
+                (["http://node5:9000/data{1...4}", "http://node{1...4}:9000/data"], 4),
+            ] {
+                let layout = DisksLayout::from_volumes(&volumes).expect("single-node multi-drive pools are valid");
+
+                assert!(!layout.legacy);
+                assert_eq!(layout.pools.len(), 2);
+                for (index, volume) in volumes.iter().enumerate() {
+                    assert_eq!(layout.get_set_count(index), 1);
+                    assert_eq!(layout.get_drives_per_set(index), drives);
+                    assert_eq!(layout.get_cmd_line(index), *volume);
+                }
+            }
+        });
+    }
+
+    #[test]
+    fn pool_expansion_accepts_multi_node_single_drive_pools() {
+        temp_env::with_var(ENV_RUSTFS_ERASURE_SET_DRIVE_COUNT, Some("0"), || {
+            for nodes in [2, 3, 4] {
+                let volumes = [
+                    format!("http://pool1-node{{1...{nodes}}}:9000/data"),
+                    format!("http://pool2-node{{1...{nodes}}}:9000/data"),
+                ];
+                let layout = DisksLayout::from_volumes(&volumes).expect("each node may contribute one drive to a pool");
+
+                assert_eq!(layout.pools.len(), 2);
+                for pool in 0..2 {
+                    assert_eq!(layout.get_set_count(pool), 1);
+                    assert_eq!(layout.get_drives_per_set(pool), nodes);
+                    let expected = (1..=nodes)
+                        .map(|node| format!("http://pool{}-node{node}:9000/data", pool + 1))
+                        .collect::<Vec<_>>();
+                    assert_eq!(layout.pools[pool].layout, vec![expected]);
+                }
+            }
+        });
+    }
+
+    #[test]
+    fn explicit_endpoints_without_ellipses_form_one_pool() {
+        temp_env::with_var(ENV_RUSTFS_ERASURE_SET_DRIVE_COUNT, Some("0"), || {
+            let volumes = ["http://node1:9000/data", "http://node2:9000/data"];
+            let layout = DisksLayout::from_volumes(&volumes).expect("explicit endpoints form one legacy pool");
+
+            assert!(layout.legacy);
+            assert_eq!(layout.pools.len(), 1);
+            assert_eq!(layout.pools[0].layout, vec![volumes.to_vec()]);
+        });
+    }
+
+    #[test]
+    fn standalone_single_drive_path_remains_supported() {
+        temp_env::with_var(ENV_RUSTFS_ERASURE_SET_DRIVE_COUNT, Some("0"), || {
+            let layout = DisksLayout::from_volumes(&["/data"]).expect("standalone single-drive deployment is valid");
+
+            assert!(layout.is_single_drive_layout());
+            assert_eq!(layout.get_single_drive_layout(), "/data");
+        });
+    }
+
+    #[test]
+    fn pool_expansion_rejects_plain_single_drive_pool_with_notice() {
+        temp_env::with_var(ENV_RUSTFS_ERASURE_SET_DRIVE_COUNT, Some("0"), || {
+            for volumes in [
+                ["http://node{1...2}:9000/data", "http://node3:9000/data"],
+                ["http://node3:9000/data", "http://node{1...2}:9000/data"],
+            ] {
+                let err = DisksLayout::from_volumes(&volumes).expect_err("a plain endpoint cannot be an expansion pool");
+                let message = err.to_string();
+
+                assert!(message.contains("all args must have ellipses for pool expansion"), "{message}");
+                assert!(message.contains("at least 2 drive endpoints"), "{message}");
+            }
+        });
+    }
+
+    #[test]
+    fn pool_expansion_rejects_singleton_ellipsis_pool_with_notice() {
+        temp_env::with_var(ENV_RUSTFS_ERASURE_SET_DRIVE_COUNT, Some("0"), || {
+            for singleton in ["http://node{3...3}:9000/data", "http://node3:9000/data{1...1}"] {
+                for volumes in [
+                    vec!["http://node{1...2}:9000/data", singleton],
+                    vec![singleton, "http://node{1...2}:9000/data"],
+                    vec![singleton],
+                ] {
+                    let err = DisksLayout::from_volumes(&volumes).expect_err("a singleton range still contains one drive");
+                    let message = err.to_string();
+
+                    assert_eq!(err.kind(), std::io::ErrorKind::Other);
+                    assert!(matches!(
+                        err.get_ref().and_then(|source| source.downcast_ref::<PoolDriveCountError>()),
+                        Some(PoolDriveCountError::BelowMinimum { size: 1 })
+                    ));
+                    assert!(message.contains("at least 2 drive endpoints"), "{message}");
+                    assert!(message.contains("single local path without ellipses"), "{message}");
+                }
+            }
+        });
+    }
+
+    #[test]
+    fn explicit_set_size_counts_drives_not_nodes() {
+        for volume in ["http://node1:9000/data{1...4}", "http://node{1...4}:9000/data"] {
+            let sets = get_all_sets(2, true, &[volume]).expect("four endpoints can form two two-drive sets");
+            assert_eq!(sets.iter().map(Vec::len).collect::<Vec<_>>(), vec![2, 2]);
+        }
+    }
+
+    #[test]
+    fn undersized_pool_error_identifies_requested_set_size() {
+        let err =
+            get_all_sets(4, true, &["http://node{1...2}:9000/data"]).expect_err("two endpoints cannot fill a four-drive set");
+        let message = err.to_string();
+
+        assert_eq!(err.kind(), std::io::ErrorKind::Other);
+        assert!(matches!(
+            err.get_ref().and_then(|source| source.downcast_ref::<PoolDriveCountError>()),
+            Some(PoolDriveCountError::BelowSetWidth {
+                size: 2,
+                set_drive_count: 4
+            })
+        ));
+        assert!(message.contains("size 2"), "{message}");
+        assert!(message.contains("RUSTFS_ERASURE_SET_DRIVE_COUNT=4"), "{message}");
+    }
+
     #[test]
     fn layout_errors_do_not_echo_url_credentials() {
         for volumes in [
             vec!["http://:duplicate-secret@server/path", "http://:duplicate-secret@server/path"],
             vec!["http://:ellipsis...secret@server/path"],
+            vec!["http://server{1...2}/data", "http://:plain-secret@server3/data"],
+            vec!["http://server{1...2}/data", "http://:singleton-secret@server{3...3}/data"],
         ] {
             let err = DisksLayout::from_volumes(&volumes).unwrap_err();
             assert!(!err.to_string().contains("secret"), "layout error leaked endpoint credentials: {err}");

crates/ecstore/src/layout/endpoints.rs

@@ -2432,6 +2432,41 @@ mod test {
         assert_eq!(local_endpoints[0].pool_idx, 1);
     }
 
+    #[tokio::test]
+    async fn pool_expansion_resolves_single_node_multi_drive_and_multi_node_single_drive_pools() {
+        for (additional_pool, expected_nodes) in [
+            ("http://rustfs-5.example.invalid:9000/data{1...4}", 5),
+            ("http://rustfs-{5...8}.example.invalid:9000/data", 8),
+        ] {
+            let layout = temp_env::with_var("RUSTFS_ERASURE_SET_DRIVE_COUNT", Some("0"), || {
+                DisksLayout::from_volumes(&["http://rustfs-{1...4}.example.invalid:9000/data", additional_pool])
+            })
+            .expect("both single-node multi-drive and multi-node single-drive pools should parse");
+
+            let (pools, setup_type) = EndpointServerPools::create_server_endpoints_with(
+                "0.0.0.0:9000",
+                &layout,
+                Some(orchestrated_test_policy()),
+                Some("rustfs-1.example.invalid"),
+            )
+            .await
+            .expect("pool admission must not impose a minimum node count or drives per node");
+
+            assert_eq!(setup_type, SetupType::DistErasure);
+            assert_eq!(pools.0.len(), 2);
+            assert_eq!(pools.get_nodes().len(), expected_nodes);
+            for (pool_index, pool) in (0_i32..).zip(&pools.0) {
+                assert_eq!((pool.set_count, pool.drives_per_set), (1, 4));
+                assert_eq!(pool.endpoints.as_ref().len(), 4);
+                for (disk_index, endpoint) in (0_i32..).zip(pool.endpoints.as_ref()) {
+                    assert_eq!(endpoint.pool_idx, pool_index);
+                    assert_eq!(endpoint.set_idx, 0);
+                    assert_eq!(endpoint.disk_idx, disk_index);
+                }
+            }
+        }
+    }
+
     #[tokio::test]
     async fn explicit_local_endpoint_host_fails_closed_for_invalid_context_or_zero_match() {
         let args = vec![