test(heal): avoid needless collect in legacy root recovery test (1.0.2 backport) (#8551)

test(heal): avoid needless collect in legacy root recovery test

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

@@ -45,11 +45,13 @@ Update this file only when an advisory adds or changes a reusable lesson, affect
 - `GHSA-6r96-hmgc-726c`: request headers collided with lowercase server-derived condition keys such as `userid`, `groups`, `versionid`, and JWT/LDAP claims. Lesson: never let caller-controlled headers append to or replace server-derived policy context; reserve trusted condition keys and keep intentional request-header keys separate.
 - `GHSA-v9cp-qfw9-9pfp`: quantified negated string conditions applied negation after aggregation, transposing `ForAllValues` and `ForAnyValue` semantics. Lesson: push negation into the per-value predicate for quantified operators and test partially overlapping multi-value sets.
 - `GHSA-5w8r-p896-6vq2`: OPA policy mode skipped `ExistingObjectTag/*` loading, so tagged objects looked untagged to external policies. Lesson: external authorization plugins need the same object-tag and request context as built-in policy evaluation before they decide.
+- `GHSA-9wgh-5r68-49p7`: substituted policy-variable values could recreate `${...}` and loop inside a single resolution pass, never reaching the outer iteration limit. Lesson: bound substitutions within each pass and test cycles formed across otherwise literal resolver values; confirm whether attacker-controlled claims can reach those values before assigning remote impact.
 
 ### S3 object actions, copy, multipart, and upload policy validation
 
 - `GHSA-g8w9-qw9q-fghr` and `GHSA-xm99-m3gq-83g8`: a valid presigned `PutObject` accepted extra `x-amz-*` headers omitted from `SignedHeaders`; `x-amz-copy-source` could turn that upload capability into a cross-bucket read performed as the signer. Lesson: a presigned URL is a bounded capability; inspect all received security-sensitive headers and reject unsigned ones before selecting the S3 operation or reaching storage.
 - `GHSA-3ppv-fx5m-m749`: explicit `versionId` reads and copy sources authorized `s3:GetObject` instead of `s3:GetObjectVersion`. Lesson: version-specific object access must select version-specific actions for direct reads, `CopyObject`, and `UploadPartCopy`, with tests proving the backend is not reached on denial.
+- `GHSA-cg33-pxfp-qcj3` and `GHSA-65wq-gjw8-xr9w`: `HeadObject` and object-tagging paths authorized current-object actions while operating on an explicit historical `versionId`. Lesson: select the corresponding version action for metadata and tag reads and writes too; audit every operation that accepts a version ID.
 - `GHSA-gjph-xcgw-4p7w`: `PutObject` and browser POST accepted a caller-supplied `versionId` and replaced an unlocked historical version with only `s3:PutObject`. Lesson: only replication-authorized writes may choose an existing version ID; ordinary uploads must not rewrite version history.
 - `GHSA-x298-9x87-fvjq`: anonymous `ListObjectVersions` fell back to `ListBucket` and returned before public-access-block gates. Lesson: compatibility fallbacks must converge on the same post-authorization checks as direct grants, especially `RestrictPublicBuckets` and anonymous data-plane denies.
 - `GHSA-mx42-j6wv-px98`: `UploadPartCopy` missed source authorization and allowed cross-bucket object exfiltration. Lesson: multipart copy must enforce the same source and destination contract as `CopyObject`.
@@ -110,6 +112,7 @@ Update this file only when an advisory adds or changes a reusable lesson, affect
 
 - `GHSA-j548-9grx-fh4f`: Object Lock enforcement treated unreadable, fabricated, or unparsable bucket metadata as absent configuration and allowed retained objects to be deleted or expired. Lesson: retention must fail closed unless Object Lock absence is authoritative, and every delete, lifecycle, scanner, force-delete, and default-retention path needs the same state distinction.
 - `GHSA-28w4-rp2p-636w` and `GHSA-c7hx-vr97-45rg`: an authorized `DeleteBucket` with `x-rustfs-force-delete: true` removed active COMPLIANCE-retained versions even though ordinary version deletion was denied. Lesson: bucket force-delete must enforce retention across every version, including against root/owner callers.
+- `GHSA-q23w-fp7p-472v`: `PutObjectRetention` policy conditions read `x-amz-object-lock-*` headers while the handler applied mode and date from the XML body. Lesson: authorize the parsed retention that will be stored, or reject mismatched header/body values; compare retention dates as instants so offsets cannot change the applied bound.
 
 ### Serde deserialization and input validation
 
@@ -143,13 +146,13 @@ rg -n "deny_unknown_fields|serde.default|as u32|as usize|as i32" rustfs crates
 - IAM fixes: include import/update/list service-account cases with attacker-controlled parent, claims, access key, secret key, and policy.
 - Copy/upload fixes: include cross-bucket, cross-user, source-denied, destination-denied, copy-source-condition, and multipart completion cases.
 - Presigned upload fixes: include a valid presign with extra unsigned tagging, redirect, storage-class, and cross-bucket copy-source headers; require rejection before source or destination storage access, and verify explicitly signed equivalents still work.
-- Version-action fixes: include historical UUID, explicit current version, `null`, range, partNumber, presigned, STS/session, service-account, anonymous bucket-policy, copy source, and multipart-copy source cases.
+- Version-action fixes: include historical UUID, explicit current version, `null`, range, partNumber, presigned, STS/session, service-account, anonymous bucket-policy, copy source, multipart-copy source, `HeadObject`, and tag read/write cases.
 - Version-write fixes: a `PutObject`-only principal must not replace an existing version through `PutObject` or browser POST `?versionId=`; verify replication-authorized writes and versioned, suspended, and unversioned buckets separately.
-- Policy-condition fixes: include reserved-key header collisions, missing keys, partially overlapping multi-value sets, plugin mode, and built-in policy mode.
+- Policy-condition fixes: include reserved-key header collisions, missing keys, partially overlapping multi-value sets, cyclic variable substitutions, plugin mode, and built-in policy mode.
 - Path fixes: include encoded traversal, absolute path, nested traversal, archive entries with `..`, valid object keys that resemble traversal text but should be rejected, and canonical bucket/prefix boundary checks; test per-key `DeleteObjects` errors and prefix-scoped denials on versioned buckets.
 - Logging fixes: assert redacted output for structs and response bodies that may contain credentials.
 - IAM export fixes: assert exported archives omit plaintext user and service-account secrets unless the format deliberately encrypts or seals them.
 - RPC auth fixes: include captured metadata replay across two concrete methods, stale timestamps, wrong path, wrong method surrogate, wrong secret, and valid same-method calls.
 - Browser/CORS fixes: assert no credentials on reflected/default origins, correct behavior for explicit allowlists, and no same-origin script execution for previewed object content.
 - SSE fixes: inspect stored bytes and verify API metadata, read-back behavior, and on-disk ciphertext together; for ETags, upload equal plaintext under different keys and assert list-only callers cannot derive or correlate the plaintext digest, including multipart objects.
-- Object Lock fixes: include unreadable metadata, fabricated metadata defaults, unparsable config, confirmed absent config, COMPLIANCE/GOVERNANCE retention, lifecycle expiry, scanner sweeps, and force-delete paths; an authorized force-delete must preserve active COMPLIANCE versions.
+- Object Lock fixes: include unreadable metadata, fabricated metadata defaults, unparsable config, confirmed absent config, COMPLIANCE/GOVERNANCE retention, lifecycle expiry, scanner sweeps, and force-delete paths; an authorized force-delete must preserve active COMPLIANCE versions. For `PutObjectRetention`, test conflicting header/body values and offset dates against policy bounds and read-back retention.

.config/e2e-full-selection.txt

@@ -1,2 +1,2 @@
-sha256-darwin=98dff0961c6eb21d6128bfd2d4a6100a624ddf1fb7db05dba360a5d3d0a310c2
-sha256-linux=890f5925b87e1bf3710c356253f3b2276d901e156df9ccc04ec9315374de6d22
+sha256-darwin=1520cd3027c3a6924afc6a204eddd5475bc30f0109d4c2847456a66177263244
+sha256-linux=26155b7f1c5bf6bf596ff24a4136e6767ea4ee6461c32db7fbe1400791da2c28

.config/e2e-nightly-selection.txt

@@ -1,2 +1,2 @@
-sha256-darwin=efe68675f4676a0d2901242c5815b1f7af5cdec8231f7cccd408f6dacf5c75db
-sha256-linux=003ed6480935a0a95680cc061be143722fbd37bd477b8ff98607b1a064f3aef1
+sha256-darwin=347a0d4e5e547d5388703062ce66a66167a845b6e57ec591ccb4c732a567c68d
+sha256-linux=4f9efa47194279185bb68249e82c788fa147abe944b6c764050e3d47f31e7926

.config/make/tests.mak

@@ -52,6 +52,9 @@ script-tests: ## Run shell script tests
 	./scripts/check_embedded_secrets.sh --self-test
 	$(RUSTFS_PYTHON_BIN) ./scripts/check_test_wiring.py --self-test
 	$(RUSTFS_PYTHON_BIN) ./scripts/test_e2e_binary.py
+	$(RUSTFS_PYTHON_BIN) ./scripts/test_release_build_workflow.py
+	$(RUSTFS_PYTHON_BIN) ./scripts/test_release_glibc.py
+	$(RUSTFS_PYTHON_BIN) ./scripts/test_upgrade_workflow.py
 	$(RUSTFS_PYTHON_BIN) ./scripts/test_migration_gate_evidence.py
 	$(RUSTFS_PYTHON_BIN) ./scripts/ci_gate.py --self-test
 	$(RUSTFS_PYTHON_BIN) ./scripts/check_security_coverage.py --self-test

.config/nextest.toml

@@ -88,6 +88,11 @@ threads-required = "num-test-threads"
 filter = 'package(rustfs-ecstore) & test(/^(bucket::lifecycle::bucket_lifecycle_ops::tests::manual_transition_worker_result_recovery_marks_unknown_for_corrupt_marker|services::rebalance::entry::tests::real_rebalance_run_fence_loss_blocks_multipart_publication|store::init::tests::(batch_transitioned_delete_uses_free_version_per_item|decommission_entry_(allows_free_version_consumed_before_source_lock|rejects_subquorum_free_version_conflict_and_retains_source|skips_cleanup_only_marker_when_free_version_is_present)|dispatched_tier_delete_recovery_(checks_later_pool_then_commits_after_source_removal|finds_directory_source_on_encoded_set|retains_journal_on_source_metadata_error)|force_tier_remove_blocks_on_physical_free_version_hidden_by_other_pool|legacy_unknown_transition_delete_falls_back_for_single_batch_and_blocks_prefix|multi_pool_(recursive_prefix_rejects_legacy_or_hidden_merge_loser_before_delete|same_remote_tuple_(batch|single)_delete_waits_for_all_sources|same_tuple_recursive_prefix_uses_one_journal_owner|transitioned_delete_persists_one_free_version_per_remote_tuple)|recursive_prefix_partial_(pool|set)_failure_keeps_prepared_cleanup_owners|restored_transitioned_delete_uses_free_version_as_cleanup_owner|stable_transitioned_recursive_prefix_delete_uses_journal_owners|suspended_null_transition_delete_uses_free_version_as_sole_owner|tier_mutation_peer_handler_applies_prepare_commit_and_abort_idempotently|transition_response_loss_persists_unknown_outcome_for_provider_recovery|transition_transaction_recovery_(drops_record_after_confirmed_local_commit|keeps_cleanup_pending_local_commit)|transitioned_delete_(free_version_replays_after_store_restart|local_quorum_failure_rolls_back_without_cleanup_owner|uses_free_version_as_cleanup_owner)|versioned_delete_marker_keeps_transitioned_source_and_remote_object|versioned_explicit_transition_delete_preserves_other_version_then_allows_bucket_delete))$/)'
 setup = 'ecstore-large-stack'
 
+# These cross-crate real-storage fixtures need the same bounded stack as ECStore tests.
+[[profile.default.scripts]]
+filter = '(package(rustfs) & test(/^(admin::handlers::site_replication::tests::durable_iam_state_apply_iam_item_rejects_a_stale_recreate_after_a_replicated_delete|admin::service::federated_identity::tests::(binding_keeps_explicit_iam_mapping_strict|markerless_legacy_oidc_session_crosses_auth_and_iam_deletion_boundaries)|storage::scanner_dirty_journal::tests::(clear_callback_keeps_unknown_coverage_until_all_pending_work_is_confirmed|owned_journals_keep_peer_records_after_interleaved_writes_and_clear))$/)) | (package(rustfs-heal) & ((binary(heal_b5_versioned_regression_test) & test(/^serial_tests::(test_heal_unversioned_bucket_e2e|test_historical_null_recursive_heal_restores_physical_versions|test_recursive_deep_heal_reports_stale_delete_marker_repair)$/)) | (binary(usage_observation_read_repair_test) & test(=usage_observation_read_repair_survives_snapshot_cleanup)))) | (package(rustfs-test-utils) & test(=data_usage_snapshot_tests::data_usage_snapshot_recovery_wires_primary_backup_and_negative_cache))'
+setup = 'ecstore-base-stack'
+
 [[profile.default.scripts]]
 filter = 'package(rustfs-ecstore) | package(rustfs-s3select-api) | package(rustfs-scanner) | (package(rustfs) & test(/^(app::multipart_usecase::tests::concurrent_completions_share_durable_bucket_quota_reservations|app::object::delete::tests::compressed_delete_requests_update_observed_usage_without_releasing_quota_floor|app::object::internal_put::tests::internal_multipart_roundtrip_completes_and_abort_leaves_nothing|app::object::restore::tests::execute_restore_object_maps_failures_to_typed_s3_errors|storage::access::tests::(delete_object_access_captures_authorized_bucket_incarnation|copy_operations_reject_recreated_source_bucket_after_authorization|request_slot_keeps_bucket_policy_bound_to_its_store))$/))'
 setup = 'ecstore-base-stack'
@@ -339,6 +344,11 @@ threads-required = "num-test-threads"
 filter = 'package(rustfs-ecstore) & test(/^(bucket::lifecycle::bucket_lifecycle_ops::tests::manual_transition_worker_result_recovery_marks_unknown_for_corrupt_marker|services::rebalance::entry::tests::real_rebalance_run_fence_loss_blocks_multipart_publication|store::init::tests::(batch_transitioned_delete_uses_free_version_per_item|decommission_entry_(allows_free_version_consumed_before_source_lock|rejects_subquorum_free_version_conflict_and_retains_source|skips_cleanup_only_marker_when_free_version_is_present)|dispatched_tier_delete_recovery_(checks_later_pool_then_commits_after_source_removal|finds_directory_source_on_encoded_set|retains_journal_on_source_metadata_error)|force_tier_remove_blocks_on_physical_free_version_hidden_by_other_pool|legacy_unknown_transition_delete_falls_back_for_single_batch_and_blocks_prefix|multi_pool_(recursive_prefix_rejects_legacy_or_hidden_merge_loser_before_delete|same_remote_tuple_(batch|single)_delete_waits_for_all_sources|same_tuple_recursive_prefix_uses_one_journal_owner|transitioned_delete_persists_one_free_version_per_remote_tuple)|recursive_prefix_partial_(pool|set)_failure_keeps_prepared_cleanup_owners|restored_transitioned_delete_uses_free_version_as_cleanup_owner|stable_transitioned_recursive_prefix_delete_uses_journal_owners|suspended_null_transition_delete_uses_free_version_as_sole_owner|tier_mutation_peer_handler_applies_prepare_commit_and_abort_idempotently|transition_response_loss_persists_unknown_outcome_for_provider_recovery|transition_transaction_recovery_(drops_record_after_confirmed_local_commit|keeps_cleanup_pending_local_commit)|transitioned_delete_(free_version_replays_after_store_restart|local_quorum_failure_rolls_back_without_cleanup_owner|uses_free_version_as_cleanup_owner)|versioned_delete_marker_keeps_transitioned_source_and_remote_object|versioned_explicit_transition_delete_preserves_other_version_then_allows_bucket_delete))$/)'
 setup = 'ecstore-large-stack'
 
+# These cross-crate real-storage fixtures need the same bounded stack as ECStore tests.
+[[profile.ci.scripts]]
+filter = '(package(rustfs) & test(/^(admin::handlers::site_replication::tests::durable_iam_state_apply_iam_item_rejects_a_stale_recreate_after_a_replicated_delete|admin::service::federated_identity::tests::(binding_keeps_explicit_iam_mapping_strict|markerless_legacy_oidc_session_crosses_auth_and_iam_deletion_boundaries)|storage::scanner_dirty_journal::tests::(clear_callback_keeps_unknown_coverage_until_all_pending_work_is_confirmed|owned_journals_keep_peer_records_after_interleaved_writes_and_clear))$/)) | (package(rustfs-heal) & ((binary(heal_b5_versioned_regression_test) & test(/^serial_tests::(test_heal_unversioned_bucket_e2e|test_historical_null_recursive_heal_restores_physical_versions|test_recursive_deep_heal_reports_stale_delete_marker_repair)$/)) | (binary(usage_observation_read_repair_test) & test(=usage_observation_read_repair_survives_snapshot_cleanup)))) | (package(rustfs-test-utils) & test(=data_usage_snapshot_tests::data_usage_snapshot_recovery_wires_primary_backup_and_negative_cache))'
+setup = 'ecstore-base-stack'
+
 [[profile.ci.scripts]]
 filter = 'package(rustfs-ecstore) | package(rustfs-s3select-api) | package(rustfs-scanner) | (package(rustfs) & test(/^(app::multipart_usecase::tests::concurrent_completions_share_durable_bucket_quota_reservations|app::object::delete::tests::compressed_delete_requests_update_observed_usage_without_releasing_quota_floor|app::object::internal_put::tests::internal_multipart_roundtrip_completes_and_abort_leaves_nothing|app::object::restore::tests::execute_restore_object_maps_failures_to_typed_s3_errors|storage::access::tests::(delete_object_access_captures_authorized_bucket_incarnation|copy_operations_reject_recreated_source_bucket_after_authorization|request_slot_keeps_bucket_policy_bound_to_its_store))$/))'
 setup = 'ecstore-base-stack'

.docker/observability/grafana/dashboards/rustfs.json

@@ -1748,7 +1748,7 @@
             "uid": "${datasource}"
           },
           "editorMode": "code",
-          "expr": "max by (job, bucket) (rustfs_cluster_usage_buckets_objects_count{job=~\"$job\", bucket=~\"$bucket\"})",
+          "expr": "max by (rustfs_cluster_id, job, bucket) (rustfs_cluster_usage_buckets_objects_count{rustfs_cluster_id=\"$storage_cluster\", job=~\"$job\", bucket=~\"$bucket\"})",
           "legendFormat": "{{bucket}}",
           "range": true,
           "refId": "A"
@@ -1848,7 +1848,7 @@
             "uid": "${datasource}"
           },
           "editorMode": "code",
-          "expr": "max by (job, bucket) (rustfs_cluster_usage_buckets_total_bytes{job=~\"$job\", bucket=~\"$bucket\"})",
+          "expr": "max by (rustfs_cluster_id, job, bucket) (rustfs_cluster_usage_buckets_total_bytes{rustfs_cluster_id=\"$storage_cluster\", job=~\"$job\", bucket=~\"$bucket\"})",
           "legendFormat": "{{bucket}}",
           "range": true,
           "refId": "A"

.docker/observability/prometheus.yml

@@ -61,6 +61,7 @@ scrape_configs:
 otlp:
   promote_resource_attributes:
     - service.instance.id
+    - process.pid
     - service.name
     - service.namespace
     - cloud.availability_zone

.docker/observability/tests/bucket-usage.test.yml

@@ -0,0 +1,48 @@
+evaluation_interval: 1m
+tests:
+- name: bucket usage keeps clusters separate and deduplicates reporting nodes
+  interval: 1m
+  input_series:
+  - series: rustfs_cluster_usage_buckets_total_bytes{rustfs_cluster_id="a",job="rustfs",bucket="shared",service_instance_id="a1"}
+    values: '4096 1024'
+  - series: rustfs_cluster_usage_buckets_total_bytes{rustfs_cluster_id="a",job="rustfs",bucket="shared",service_instance_id="a2"}
+    values: '4096 1024'
+  - series: rustfs_cluster_usage_buckets_total_bytes{rustfs_cluster_id="b",job="rustfs",bucket="shared",service_instance_id="b1"}
+    values: '65536 65536'
+  - series: rustfs_cluster_usage_buckets_total_bytes{rustfs_cluster_id="a",job="rustfs",bucket="empty",service_instance_id="a1"}
+    values: '0 0'
+  - series: rustfs_cluster_usage_buckets_objects_count{rustfs_cluster_id="a",job="rustfs",bucket="shared",service_instance_id="a1"}
+    values: '3 1'
+  - series: rustfs_cluster_usage_buckets_objects_count{rustfs_cluster_id="a",job="rustfs",bucket="shared",service_instance_id="a2"}
+    values: '3 1'
+  - series: rustfs_cluster_usage_buckets_objects_count{rustfs_cluster_id="b",job="rustfs",bucket="shared",service_instance_id="b1"}
+    values: '70 70'
+  - series: rustfs_cluster_usage_buckets_objects_count{rustfs_cluster_id="a",job="rustfs",bucket="empty",service_instance_id="a1"}
+    values: '0 0'
+  promql_expr_test:
+  - expr: max by (rustfs_cluster_id, job, bucket) (rustfs_cluster_usage_buckets_total_bytes{rustfs_cluster_id="a",job=~"rustfs",bucket=~".*"})
+    eval_time: 1m
+    exp_samples:
+    - labels: '{rustfs_cluster_id="a",job="rustfs",bucket="shared"}'
+      value: 1024
+    - labels: '{rustfs_cluster_id="a",job="rustfs",bucket="empty"}'
+      value: 0
+  - expr: max by (rustfs_cluster_id, job, bucket) (rustfs_cluster_usage_buckets_objects_count{rustfs_cluster_id="a",job=~"rustfs",bucket=~".*"})
+    eval_time: 1m
+    exp_samples:
+    - labels: '{rustfs_cluster_id="a",job="rustfs",bucket="shared"}'
+      value: 1
+    - labels: '{rustfs_cluster_id="a",job="rustfs",bucket="empty"}'
+      value: 0
+  - expr: max by (rustfs_cluster_id, job, bucket) (rustfs_cluster_usage_buckets_total_bytes)
+    eval_time: 1m
+    exp_samples:
+    - labels: '{rustfs_cluster_id="a",job="rustfs",bucket="shared"}'
+      value: 1024
+    - labels: '{rustfs_cluster_id="a",job="rustfs",bucket="empty"}'
+      value: 0
+    - labels: '{rustfs_cluster_id="b",job="rustfs",bucket="shared"}'
+      value: 65536
+  - expr: rustfs_cluster_usage_buckets_total_bytes{rustfs_cluster_id="a",bucket="unknown"}
+    eval_time: 1m
+    exp_samples: []

.github/actions/setup-docker-compose/action.yml

@@ -0,0 +1,39 @@
+# 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.
+
+name: Set up Docker Compose
+description: Install the verified Compose CLI used by native Linux x86_64 Connect acceptance
+
+runs:
+  using: composite
+  steps:
+    - name: Install verified Docker Compose CLI plugin
+      shell: bash
+      run: |
+        set -euo pipefail
+        [[ $(uname -s) == Linux && $(uname -m) == x86_64 ]]
+        plugin_dir="${DOCKER_CONFIG:-$HOME/.docker}/cli-plugins"
+        mkdir -p "$plugin_dir"
+        compose_download=$(mktemp "${RUNNER_TEMP:?}/connect-compose.XXXXXX")
+        compose_staged=$(mktemp "$plugin_dir/.docker-compose.XXXXXX")
+        trap 'rm -f "$compose_download" "$compose_staged"' EXIT
+        curl --fail --location --retry 3 \
+          https://github.com/docker/compose/releases/download/v2.40.3/docker-compose-linux-x86_64 \
+          --output "$compose_download"
+        printf '%s  %s\n' \
+          dba9d98e1ba5bfe11d88c99b9bd32fc4a0624a30fafe68eea34d61a3e42fd372 \
+          "$compose_download" | sha256sum --check --strict
+        install -m 0755 "$compose_download" "$compose_staged"
+        mv -f "$compose_staged" "$plugin_dir/docker-compose"
+        [[ $(docker compose version --short) == 2.40.3 ]]

.github/actions/setup/action.yml

@@ -36,6 +36,16 @@ inputs:
     description: "Install cross-compilation tools"
     required: false
     default: "false"
+  zig-version:
+    description: >-
+      Zig version for cargo-zigbuild builds. Pinned so the `<target>.<glibc>`
+      baseline semantics cannot drift with an untested zig release.
+    required: false
+    default: "0.17.0"
+  cargo-zigbuild-version:
+    description: "cargo-zigbuild version installed for cross builds."
+    required: false
+    default: "0.23.4"
   target:
     description: "Target architecture to add"
     required: false
@@ -99,10 +109,14 @@ runs:
     - name: Install Zig
       if: inputs.install-cross-tools == 'true'
       uses: mlugg/setup-zig@d1434d08867e3ee9daa34448df10607b98908d29 # v2
+      with:
+        version: ${{ inputs.zig-version }}
 
     - name: Install cargo-zigbuild
       if: inputs.install-cross-tools == 'true'
       uses: taiki-e/install-action@a21ae4029b089b9ddc45704028756f51ab8abe48 # cargo-zigbuild
+      with:
+        tool: cargo-zigbuild@${{ inputs.cargo-zigbuild-version }}
 
     - name: Install cargo-nextest
       if: inputs.install-test-tools == 'true'
@@ -126,5 +140,5 @@ runs:
         # normalises, so cargo fingerprints stay valid.
         cache-all-crates: false
         cache-on-failure: true
-        shared-key: ${{ inputs.cache-shared-key }}
+        shared-key: ${{ inputs.cache-shared-key }}${{ inputs.install-cross-tools == 'true' && format('-zig-{0}-zigbuild-{1}', inputs.zig-version, inputs.cargo-zigbuild-version) || '' }}
         save-if: ${{ inputs.cache-save-if }}

.github/workflows/build.yml

@@ -213,12 +213,12 @@ jobs:
           fi
 
           all='{"include":[
-            {"target_id":"linux-x86_64-musl","os":"sm-standard-4","target":"x86_64-unknown-linux-musl","cross":false,"platform":"linux","rustflags":"","features":""},
-            {"target_id":"linux-aarch64-musl","os":"sm-standard-4-arm","target":"aarch64-unknown-linux-musl","cross":false,"platform":"linux","rustflags":"","features":""},
-            {"target_id":"linux-x86_64-gnu","os":"sm-standard-4","target":"x86_64-unknown-linux-gnu","cross":false,"platform":"linux","rustflags":"","features":"pyroscope"},
-            {"target_id":"linux-aarch64-gnu","os":"sm-standard-4-arm","target":"aarch64-unknown-linux-gnu","cross":false,"platform":"linux","rustflags":"","features":"pyroscope"},
-            {"target_id":"macos-aarch64","os":"macos-latest","target":"aarch64-apple-darwin","cross":false,"platform":"macos","rustflags":"","features":"pyroscope"},
-            {"target_id":"windows-x86_64","os":"windows-latest","target":"x86_64-pc-windows-msvc","cross":false,"platform":"windows","rustflags":"","features":""}
+            {"target_id":"linux-x86_64-musl","os":"sm-standard-4","target":"x86_64-unknown-linux-musl","cross":false,"platform":"linux","rustflags":"","features":"","glibc":""},
+            {"target_id":"linux-aarch64-musl","os":"sm-standard-4-arm","target":"aarch64-unknown-linux-musl","cross":false,"platform":"linux","rustflags":"","features":"","glibc":""},
+            {"target_id":"linux-x86_64-gnu","os":"sm-standard-4","target":"x86_64-unknown-linux-gnu","cross":true,"platform":"linux","rustflags":"","features":"pyroscope","glibc":"2.31"},
+            {"target_id":"linux-aarch64-gnu","os":"sm-standard-4-arm","target":"aarch64-unknown-linux-gnu","cross":true,"platform":"linux","rustflags":"","features":"pyroscope","glibc":"2.31"},
+            {"target_id":"macos-aarch64","os":"macos-latest","target":"aarch64-apple-darwin","cross":false,"platform":"macos","rustflags":"","features":"pyroscope","glibc":""},
+            {"target_id":"windows-x86_64","os":"windows-latest","target":"x86_64-pc-windows-msvc","cross":false,"platform":"windows","rustflags":"","features":"","glibc":""}
           ]}'
 
           if [[ "${selected}" == "all" ]]; then
@@ -246,7 +246,8 @@ jobs:
           echo "Selected platforms: ${selected}"
 
   build-rustfs:
-    name: Build RustFS
+    # Acceptance workflows select artifacts by this stable platform name.
+    name: Build RustFS (${{ matrix.target_id }})
     needs: [ build-check, prepare-platform-matrix ]
     if: needs.build-check.outputs.should_build == 'true' && needs.prepare-platform-matrix.result == 'success'
     runs-on: ${{ matrix.os }}
@@ -273,7 +274,7 @@ jobs:
         with:
           rust-version: stable
           target: ${{ matrix.target }}
-          cache-shared-key: build-${{ matrix.target }}
+          cache-shared-key: build-${{ matrix.target }}${{ matrix.glibc && format('-glibc-{0}', matrix.glibc) || '' }}
           # main only. A cache saved on refs/tags/X is scoped to that tag: no
           # other tag, no main run and no PR can restore it, so every release
           # cycle wrote up to 12 entries of 1-2GB (preview tag plus final tag,
@@ -410,35 +411,53 @@ jobs:
         shell: bash
         env:
           BUILD_TARGET: ${{ matrix.target }}
+          BUILD_OS: ${{ matrix.platform }}-${{ startsWith(matrix.target, 'aarch64-') && 'aarch64' || 'x86_64' }}
         run: |
           # Force rebuild by touching build.rs
           touch rustfs/build.rs
 
-          BINARY_NAMES=(rustfs)
-          if [[ "${{ matrix.platform }}" == "linux" ]]; then
-            BINARY_NAMES+=(rustfs-cli)
-          fi
           FEATURE_ARGS=()
           if [[ -n "${{ matrix.features }}" ]]; then
             FEATURE_ARGS=(--features "${{ matrix.features }}")
           fi
 
-          # Build one binary at a time so release LTO links cannot exhaust a
-          # self-hosted runner by running concurrently.
-          for binary in "${BINARY_NAMES[@]}"; do
-            PROFILE_ARGS=()
-            if [[ "$binary" == "rustfs" && "$BUILD_TARGET" == *-linux-gnu && "${{ matrix.features }}" == "pyroscope" ]]; then
-              # The in-process profiler resolves names from the final executable.
-              # Retain function symbols without enabling DWARF source information.
-              PROFILE_ARGS=(--config 'profile.release.strip="none"')
-            fi
-            if [[ "${{ matrix.cross }}" == "true" ]]; then
-              # All cross targets in the matrix are Linux; zigbuild handles them.
-              cargo zigbuild --release --target ${{ matrix.target }} "${FEATURE_ARGS[@]}" "${PROFILE_ARGS[@]}" -p rustfs --bin "$binary"
-            else
-              cargo build --release --target ${{ matrix.target }} "${FEATURE_ARGS[@]}" "${PROFILE_ARGS[@]}" -p rustfs --bin "$binary"
-            fi
-          done
+          # gnu targets pin an explicit glibc baseline (`<target>.<glibc>` for
+          # cargo-zigbuild) so release binaries link no symbols newer than it
+          # and run on older distros such as RHEL/Rocky 9 (glibc 2.34); #8062.
+          TARGET_ARG="$BUILD_TARGET"
+          if [[ -n "${{ matrix.glibc }}" ]]; then
+            TARGET_ARG="${TARGET_ARG}.${{ matrix.glibc }}"
+          fi
+
+          PROFILE_ARGS=()
+          if [[ "$BUILD_TARGET" == *-linux-gnu && "${{ matrix.features }}" == "pyroscope" ]]; then
+            # The in-process profiler resolves names from the final executable.
+            # Retain function symbols without enabling DWARF source information.
+            PROFILE_ARGS=(--config 'profile.release.strip="none"')
+          fi
+          if [[ "${{ matrix.cross }}" == "true" ]]; then
+            cargo zigbuild --release --locked --target "$TARGET_ARG" "${FEATURE_ARGS[@]}" "${PROFILE_ARGS[@]}" -p rustfs --bin rustfs
+          else
+            cargo build --release --locked --target "$TARGET_ARG" "${FEATURE_ARGS[@]}" "${PROFILE_ARGS[@]}" -p rustfs --bin rustfs
+          fi
+
+          # All release targets use same-architecture runners, including GNU
+          # builds that use Zig only to select the glibc ABI floor.
+          bin="target/${{ matrix.target }}/release/rustfs"
+          if [[ -f "${bin}.exe" ]]; then
+            bin="${bin}.exe"
+          fi
+          version_output=$("$bin" --version)
+          if ! grep -Fxq "build target : ${BUILD_TARGET}" <<< "$version_output" || \
+             ! grep -Fxq "build os     : ${BUILD_OS}" <<< "$version_output" || \
+             ! grep -Fxq "rust channel : stable-${BUILD_TARGET}" <<< "$version_output"; then
+            echo "release binary reports an unexpected compile target" >&2
+            exit 1
+          fi
+          if [[ -n "${{ matrix.glibc }}" ]]; then
+            python3 scripts/release/check_glibc.py "${{ matrix.glibc }}" \
+              "target/${{ matrix.target }}/release/rustfs"
+          fi
 
       - name: Verify macOS binary has no dynamic liblzma dependency
         if: matrix.platform == 'macos'
@@ -554,9 +573,6 @@ jobs:
           fi
 
           BINARY_FILES=("$BINARY_NAME")
-          if [[ "${{ matrix.platform }}" == "linux" ]]; then
-            BINARY_FILES+=("rustfs-cli")
-          fi
 
           if [[ "$TARGET" == *-linux-gnu && "${{ matrix.features }}" == "pyroscope" ]]; then
             BINARY_FILES+=(rustfs.cpu-symbol-catalog.json rustfs.cpu-symbol-catalog.sha256)
@@ -671,7 +687,7 @@ jobs:
           echo "🔧 Build type: ${BUILD_TYPE}"
           echo "📊 Version: ${VERSION}"
 
-      - name: Verify packaged Linux binaries
+      - name: Verify packaged Linux binary
         if: matrix.platform == 'linux'
         shell: bash
         run: |
@@ -679,7 +695,7 @@ jobs:
 
           package_file="${{ steps.package.outputs.package_file }}"
           archive_members="$(unzip -Z1 "$package_file" | sort)"
-          expected_members=$'rustfs\nrustfs-cli'
+          expected_members='rustfs'
           target="${{ matrix.target }}"
           features="${{ matrix.features }}"
           if [[ "$target" == *-linux-gnu && "$features" == "pyroscope" ]]; then
@@ -691,13 +707,14 @@ jobs:
             exit 1
           fi
 
-          # Every Linux target builds natively, so executing the binaries here
+          # musl targets build natively; the gnu legs are same-arch
+          # cargo-zigbuild builds pinned to the release glibc baseline. Either
+          # way the binaries run on their runner, so executing them here
           # proves each artifact matches its runner architecture.
-          cli_dir="$(mktemp -d)"
-          trap 'rm -rf "$cli_dir"' EXIT
-          unzip -q "$package_file" rustfs rustfs-cli -d "$cli_dir"
-          "$cli_dir/rustfs" --version
-          "$cli_dir/rustfs-cli" --help >/dev/null
+          binary_dir="$(mktemp -d)"
+          trap 'rm -rf "$binary_dir"' EXIT
+          unzip -q "$package_file" rustfs -d "$binary_dir"
+          "$binary_dir/rustfs" --version
 
       - name: Verify packaged console
         if: matrix.target == 'x86_64-unknown-linux-gnu' || matrix.target == 'aarch64-unknown-linux-gnu'

.github/workflows/cache-warm.yml

@@ -99,7 +99,7 @@ env:
   RUST_BACKTRACE: 1
 
 jobs:
-  # Readers: test-and-lint, test-ilm-integration-serial, build-rustfs-debug-binary,
+  # Readers: test-and-lint (workspace and ILM), build-rustfs-debug-binary,
   # e2e-tests, e2e-full.
   warm-ci-dev:
     name: Warm ci-dev

.github/workflows/ci.yml

@@ -168,13 +168,10 @@ jobs:
             echo "started_at=$(date --utc --iso-8601=seconds)"
           } > artifacts/test-and-lint/run-metadata.txt
 
-      # Clippy runs before the test pass: lint failures are the most common
-      # CI-only breakage and should surface in minutes, not after 20+ minutes
-      # of tests.
-      # Sampled too: clippy is the natural control arm for any CARGO_BUILD_JOBS
-      # experiment, since --all-targets is check-only for workspace members and
-      # never links the ~100 test binaries the limit exists to throttle.
+      # Keep full Clippy validation on main, merge groups, and scheduled/manual
+      # runs. PRs retain the complete default test suite and compile-free checks.
       - name: Run clippy lints
+        if: github.event_name != 'pull_request'
         run: |
           ./scripts/ci/resource_sampler.sh start clippy
           trap './scripts/ci/resource_sampler.sh stop' EXIT
@@ -283,6 +280,49 @@ jobs:
       - name: Verify rebalance/decommission migration proofs
         run: ./scripts/check_migration_gate_count.sh evidence artifacts/test-and-lint/core-test-listing.json target/nextest/ci/junit.xml
 
+      # The workspace already compiled these ignored tests. Keep the same Cargo
+      # package selection to reuse their binaries, then run them serially in
+      # separate processes because they share fixed ports and global state.
+      # test_transition_and_restore_flows was re-enabled by rustfs/backlog#1303:
+      # its "missing xl.meta on disk2" was a test-util bug (open_disk hardcoded
+      # disk_index 0), not an EC metadata-distribution issue.
+      # restore_object_usecase_reports_ongoing_conflict_and_completion was
+      # re-enabled by backlog#1304 (restore accepts serialize on a short CAS
+      # guard; the copy-back no longer holds the #4877 whole-copy-back lock,
+      # so the mid-restore ongoing read and fast 409 rejection it asserts are
+      # the implemented contract).
+      - name: Run ignored ILM integration tests serially
+        if: needs.classify-changes.outputs.mode == 'full'
+        env:
+          # Match the preceding workspace build and reuse its feature graph.
+          CARGO_BUILD_JOBS: ${{ (github.event_name == 'push' || github.event_name == 'workflow_dispatch') && '3' || '2' }}
+        run: |
+          mkdir -p artifacts/ilm-integration
+          set +e
+          NEXTEST_HIDE_PROGRESS_BAR=1 timeout --verbose --signal=TERM --kill-after=30s 80m \
+            cargo nextest run --profile default -j1 --run-ignored ignored-only \
+            --all --exclude e2e_test \
+            -E 'binary(lifecycle_integration_test) or (package(rustfs) and test(lifecycle_transition_api_test))' \
+            --status-level all --final-status-level all \
+            2>&1 | tee artifacts/ilm-integration/nextest.log
+          status=${PIPESTATUS[0]}
+          {
+            echo "exit_status=${status}"
+            echo "finished_at=$(date --utc --iso-8601=seconds)"
+            echo
+            echo "Remaining test-related processes:"
+            pgrep -af 'cargo|nextest|target/.*/deps/' || true
+          } > artifacts/ilm-integration/diagnostics.txt
+          exit "${status}"
+
+      - name: Upload ILM test diagnostics
+        if: always() && needs.classify-changes.outputs.mode == 'full'
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        with:
+          name: ilm-integration-${{ github.run_number }}-${{ github.run_attempt }}
+          path: artifacts/ilm-integration
+          retention-days: 3
+
   # This gate builds into a fresh target directory to isolate the E2E root.
   # Give its cold build a separate budget from workspace tests and migration proofs.
   offline-enrollment-e2e:
@@ -314,83 +354,11 @@ jobs:
       - name: Check offline enrollment E2E root boundary
         run: ./scripts/check_offline_enrollment_e2e.sh
 
-  # Dedicated serial lane for the ILM / lifecycle integration tests. These tests
-  # drive the object layer through process-global singletons (the GLOBAL_ENV
-  # ECStore, the global tier-config manager, background-expiry workers) and bind
-  # fixed ports (9002 for the scanner suite, 9003 for the app suite), so they are
-  # marked #[ignore] to keep them out of the default parallel `cargo nextest run
-  # --all` pass. Run them here explicitly with `--run-ignored ignored-only` and
-  # `-j1` so no two of them share global state or race for a port at once.
-  # serial_test's #[serial] does NOT serialize across nextest's process-per-test
-  # boundary (see .config/nextest.toml); `-j1` is what actually serializes them.
-  # See rustfs/backlog#1148 (ilm-1) and #1155.
-  test-ilm-integration-serial:
-    name: ILM Integration (serial)
-    if: needs.classify-changes.outputs.mode == 'full' && (github.event_name != 'pull_request' || github.event.action != 'closed')
-    needs: [ quick-checks, classify-changes ]
-    runs-on: sm-standard-4
-    timeout-minutes: 90
-    env:
-      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@f548e57e544e1ff5a4c46bf1e1b8685f8e4a348a # v7
-        with:
-          persist-credentials: false
-
-      - name: Setup Rust environment
-        uses: ./.github/actions/setup
-        with:
-          rust-version: stable
-          cache-shared-key: ci-dev
-          cache-save-if: 'false'
-          install-build-packaging-tools: 'false'
-
-      # test_transition_and_restore_flows was re-enabled by rustfs/backlog#1303:
-      # its "missing xl.meta on disk2" was a test-util bug (open_disk hardcoded
-      # disk_index 0), not an EC metadata-distribution issue.
-      # restore_object_usecase_reports_ongoing_conflict_and_completion was
-      # re-enabled by backlog#1304 (restore accepts serialize on a short CAS
-      # guard; the copy-back no longer holds the #4877 whole-copy-back lock,
-      # so the mid-restore ongoing read and fast 409 rejection it asserts are
-      # the implemented contract).
-      - name: Run ignored ILM integration tests serially
-        env:
-          # Match the measured Test and Lint link budget. The default exposed
-          # all 14 pod CPUs and a cold cache spent the full 80m compiling
-          # without starting one ILM test (main run 32982910990).
-          CARGO_BUILD_JOBS: ${{ (github.event_name == 'push' || github.event_name == 'workflow_dispatch') && '3' || '2' }}
-        run: |
-          mkdir -p artifacts/ilm-integration
-          set +e
-          NEXTEST_HIDE_PROGRESS_BAR=1 timeout --verbose --signal=TERM --kill-after=30s 80m \
-            cargo nextest run -j1 --run-ignored ignored-only \
-            -p rustfs-scanner -p rustfs \
-            -E 'binary(lifecycle_integration_test) or (package(rustfs) and test(lifecycle_transition_api_test))' \
-            --status-level all --final-status-level all \
-            2>&1 | tee artifacts/ilm-integration/nextest.log
-          status=${PIPESTATUS[0]}
-          {
-            echo "exit_status=${status}"
-            echo "finished_at=$(date --utc --iso-8601=seconds)"
-            echo
-            echo "Remaining test-related processes:"
-            pgrep -af 'cargo|nextest|target/.*/deps/' || true
-          } > artifacts/ilm-integration/diagnostics.txt
-          exit "${status}"
-
-      - name: Upload ILM test diagnostics
-        if: always()
-        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
-        with:
-          name: ilm-integration-${{ github.run_number }}-${{ github.run_attempt }}
-          path: |
-            artifacts/ilm-integration
-            target/nextest/ci/junit.xml
-
   test-and-lint-rio-v2:
     name: Test and Lint (rio-v2)
-    if: needs.classify-changes.outputs.mode == 'full' && (github.event_name != 'pull_request' || github.event.action != 'closed')
+    # Alternate feature configurations run on main, merge groups, and scheduled
+    # or manual validation; the default workspace remains the PR gate.
+    if: needs.classify-changes.outputs.mode == 'full' && github.event_name != 'pull_request'
     needs: [ quick-checks, classify-changes ]
     runs-on: sm-standard-4
     timeout-minutes: 90
@@ -506,7 +474,7 @@ jobs:
 
   test-and-lint-protocols:
     name: "Test and Lint (${{ matrix.features.name }})"
-    if: needs.classify-changes.outputs.mode == 'full' && (github.event_name != 'pull_request' || github.event.action != 'closed')
+    if: needs.classify-changes.outputs.mode == 'full' && github.event_name != 'pull_request'
     needs: [ quick-checks, classify-changes ]
     runs-on: sm-standard-4
     timeout-minutes: 90
@@ -1264,7 +1232,6 @@ jobs:
       - quick-checks
       - test-and-lint
       - offline-enrollment-e2e
-      - test-ilm-integration-serial
       - test-and-lint-rio-v2
       - connect-short-credential-boundary
       - test-and-lint-protocols
@@ -1298,7 +1265,6 @@ jobs:
       - quick-checks
       - test-and-lint
       - offline-enrollment-e2e
-      - test-ilm-integration-serial
       - test-and-lint-rio-v2
       - test-and-lint-protocols
       - build-rustfs-debug-binary

.github/workflows/connect-diagnostic-schedule-acceptance.yml

@@ -57,6 +57,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -94,7 +102,7 @@ jobs:
           ' <<<"$run" >/dev/null
           jobs=$(gh api --paginate --slurp "repos/${GITHUB_REPOSITORY}/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e '
-            [.[].jobs[] | select(.name | test("^Build RustFS \\(linux-x86_64-gnu, [a-z0-9-]+, x86_64-unknown-linux-gnu, false, linux, pyroscope\\)$"))] as $matches
+            [.[].jobs[] | select(.name | test("^Build RustFS \\(linux-x86_64-gnu\\)$"))] as $matches
             | ($matches | length) == 1 and $matches[0].conclusion == "success"
           ' <<<"$jobs" >/dev/null
           artifact=$(gh api "repos/${GITHUB_REPOSITORY}/actions/artifacts/${ARTIFACT_ID}")
@@ -137,7 +145,7 @@ jobs:
               assert outer.testzip() is None
               with zipfile.ZipFile(io.BytesIO(outer.read(package))) as inner:
                   entries = inner.infolist()
-                  assert {item.filename for item in entries} == {'rustfs', 'rustfs-cli'} and len(entries) == 2
+                  assert {item.filename for item in entries} == {'rustfs', 'rustfs.cpu-symbol-catalog.json', 'rustfs.cpu-symbol-catalog.sha256'} and len(entries) == 3
                   assert all(regular(item) and 0 < item.file_size <= 1073741824 for item in entries)
                   assert inner.testzip() is None
                   pathlib.Path('binary').mkdir()
@@ -162,7 +170,7 @@ jobs:
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
           (cd connect-harness && ./tests/connectivity/diagnostic-schedule-evidence.test.sh)
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_DIAGNOSTIC_SCHEDULE_EVIDENCE="$GITHUB_WORKSPACE/diagnostic-schedule-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=diagnostic-schedule

.github/workflows/connect-health-service-acceptance.yml

@@ -57,6 +57,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -127,7 +135,7 @@ jobs:
           ' <<<"$run" >/dev/null
           jobs=$(gh api --paginate --slurp "repos/rustfs/rustfs/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e '
-            [.[].jobs[] | select(.name | test("^Build RustFS \\(linux-x86_64-gnu, [a-z0-9-]+, x86_64-unknown-linux-gnu, false, linux, pyroscope\\)$"))] as $matches
+            [.[].jobs[] | select(.name | test("^Build RustFS \\(linux-x86_64-gnu\\)$"))] as $matches
             | ($matches | length) == 1 and $matches[0].conclusion == "success"
           ' <<<"$jobs" >/dev/null
           artifact=$(gh api "repos/rustfs/rustfs/actions/artifacts/${ARTIFACT_ID}")
@@ -172,7 +180,7 @@ jobs:
               assert outer.testzip() is None
               with zipfile.ZipFile(io.BytesIO(outer.read(package))) as inner:
                   entries = inner.infolist()
-                  assert {entry.filename for entry in entries} == {'rustfs', 'rustfs-cli', 'rustfs.cpu-symbol-catalog.json', 'rustfs.cpu-symbol-catalog.sha256'} and len(entries) == 4
+                  assert {entry.filename for entry in entries} == {'rustfs', 'rustfs.cpu-symbol-catalog.json', 'rustfs.cpu-symbol-catalog.sha256'} and len(entries) == 3
                   assert all(regular(entry) and 0 < entry.file_size <= 1073741824 for entry in entries)
                   assert inner.testzip() is None
                   pathlib.Path('binary').mkdir()
@@ -199,7 +207,7 @@ jobs:
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
           (cd connect-harness && ./tests/connectivity/health-service-evidence.test.sh)
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_HEALTH_SERVICE_EVIDENCE="$GITHUB_WORKSPACE/health-service-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=health-service

.github/workflows/connect-performance-drive-acceptance.yml

@@ -73,6 +73,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -107,7 +115,7 @@ jobs:
           [[ $(jq -r '.head_repository.full_name' <<<"$run") == "$GITHUB_REPOSITORY" ]]
           [[ $(jq -r '.name' <<<"$run") == "Build and Release" ]]
 
-          expected_job='Build RustFS (linux-x86_64-gnu, sm-standard-4, x86_64-unknown-linux-gnu, false, linux, pyroscope)'
+          expected_job='Build RustFS (linux-x86_64-gnu)'
           jobs=$(gh api --paginate --slurp "repos/${GITHUB_REPOSITORY}/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e --arg name "$expected_job" '
             [.[].jobs[] | select(.name == $name)] as $matches
@@ -154,7 +162,7 @@ jobs:
           npm --prefix connect-harness/web ci
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_PERFORMANCE_DRIVE_EVIDENCE="$GITHUB_WORKSPACE/performance-drive-service-job-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=performance-drive

.github/workflows/connect-performance-network-acceptance.yml

@@ -73,6 +73,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -107,7 +115,7 @@ jobs:
           [[ $(jq -r '.head_repository.full_name' <<<"$run") == "$GITHUB_REPOSITORY" ]]
           [[ $(jq -r '.name' <<<"$run") == "Build and Release" ]]
 
-          expected_job='Build RustFS (linux-x86_64-gnu, sm-standard-4, x86_64-unknown-linux-gnu, false, linux, pyroscope)'
+          expected_job='Build RustFS (linux-x86_64-gnu)'
           jobs=$(gh api --paginate --slurp "repos/${GITHUB_REPOSITORY}/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e --arg name "$expected_job" '
             [.[].jobs[] | select(.name == $name)] as $matches
@@ -154,7 +162,7 @@ jobs:
           npm --prefix connect-harness/web ci
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_PERFORMANCE_NETWORK_EVIDENCE="$GITHUB_WORKSPACE/performance-network-service-job-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=performance-network

.github/workflows/connect-profile-cpu-acceptance.yml

@@ -77,6 +77,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -118,7 +126,7 @@ jobs:
           [[ $(jq -r '.head_repository.full_name' <<<"$run") == "$GITHUB_REPOSITORY" ]]
           [[ $(jq -r '.name' <<<"$run") == "Build and Release" ]]
 
-          expected_job='Build RustFS (linux-x86_64-gnu, sm-standard-4, x86_64-unknown-linux-gnu, false, linux, pyroscope)'
+          expected_job='Build RustFS (linux-x86_64-gnu)'
           jobs=$(github_api "repos/${GITHUB_REPOSITORY}/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e --arg name "$expected_job" '
             [.jobs[] | select(.name == $name)] as $matches
@@ -173,7 +181,7 @@ jobs:
           npm --prefix connect-harness/web ci
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_CPU_CATALOG="$GITHUB_WORKSPACE/binary/rustfs.cpu-symbol-catalog.json" \
             CONNECT_E2E_CPU_CATALOG_SHA256="$CATALOG_SHA256" \

.github/workflows/connect-profile-memory-acceptance.yml

@@ -57,6 +57,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -127,7 +135,7 @@ jobs:
           ' <<<"$run" >/dev/null
           jobs=$(gh api --paginate --slurp "repos/rustfs/rustfs/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e '
-            [.[].jobs[] | select(.name | test("^Build RustFS \\(linux-x86_64-gnu, [a-z0-9-]+, x86_64-unknown-linux-gnu, false, linux, pyroscope\\)$"))] as $matches
+            [.[].jobs[] | select(.name | test("^Build RustFS \\(linux-x86_64-gnu\\)$"))] as $matches
             | ($matches | length) == 1 and $matches[0].conclusion == "success"
           ' <<<"$jobs" >/dev/null
           artifact=$(gh api "repos/rustfs/rustfs/actions/artifacts/${ARTIFACT_ID}")
@@ -172,7 +180,7 @@ jobs:
               assert outer.testzip() is None
               with zipfile.ZipFile(io.BytesIO(outer.read(package))) as inner:
                   entries = inner.infolist()
-                  expected = {'rustfs', 'rustfs-cli', 'rustfs.cpu-symbol-catalog.json', 'rustfs.cpu-symbol-catalog.sha256'}
+                  expected = {'rustfs', 'rustfs.cpu-symbol-catalog.json', 'rustfs.cpu-symbol-catalog.sha256'}
                   assert {entry.filename for entry in entries} == expected and len(entries) == len(expected)
                   assert all(regular(entry) and 0 < entry.file_size <= 1073741824 for entry in entries)
                   assert inner.testzip() is None
@@ -198,7 +206,7 @@ jobs:
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
           (cd connect-harness && ./tests/connectivity/profile-memory-evidence.test.sh)
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_PROFILE_MEMORY_EVIDENCE="$GITHUB_WORKSPACE/profile-memory-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=profile-memory

.github/workflows/connect-profile-threads-acceptance.yml

@@ -81,6 +81,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -148,7 +156,7 @@ jobs:
           [[ $(jq -r '.head_repository.full_name' <<<"$run") == "$GITHUB_REPOSITORY" ]]
           [[ $(jq -r '.name' <<<"$run") == "Build and Release" ]]
 
-          expected_job='Build RustFS (linux-x86_64-gnu, sm-standard-4, x86_64-unknown-linux-gnu, false, linux, pyroscope)'
+          expected_job='Build RustFS (linux-x86_64-gnu)'
           jobs=$(gh api --paginate --slurp "repos/${GITHUB_REPOSITORY}/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e --arg name "$expected_job" '
             [.[].jobs[] | select(.name == $name)] as $matches
@@ -201,7 +209,7 @@ jobs:
           npm --prefix connect-harness/web ci
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_PROFILE_THREADS_EVIDENCE="$GITHUB_WORKSPACE/profile-threads-service-job-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=profile-threads

.github/workflows/connect-top-api-acceptance.yml

@@ -73,6 +73,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -107,7 +115,7 @@ jobs:
           [[ $(jq -r '.head_repository.full_name' <<<"$run") == "$GITHUB_REPOSITORY" ]]
           [[ $(jq -r '.name' <<<"$run") == "Build and Release" ]]
 
-          expected_job='Build RustFS (linux-x86_64-gnu, sm-standard-4, x86_64-unknown-linux-gnu, false, linux, pyroscope)'
+          expected_job='Build RustFS (linux-x86_64-gnu)'
           jobs=$(gh api --paginate --slurp "repos/${GITHUB_REPOSITORY}/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e --arg name "$expected_job" '
             [.[].jobs[] | select(.name == $name)] as $matches
@@ -154,7 +162,7 @@ jobs:
           npm --prefix connect-harness/web ci
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_TOP_API_EVIDENCE="$GITHUB_WORKSPACE/top-api-service-job-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=top-api

.github/workflows/connect-top-disk-acceptance.yml

@@ -68,6 +68,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -138,7 +146,7 @@ jobs:
           ' <<<"$run" >/dev/null
           jobs=$(gh api --paginate --slurp "repos/rustfs/rustfs/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e '
-            [.[].jobs[] | select(.name | test("^Build RustFS \\(linux-x86_64-gnu, [a-z0-9-]+, x86_64-unknown-linux-gnu, false, linux, pyroscope\\)$"))] as $matches
+            [.[].jobs[] | select(.name | test("^Build RustFS \\(linux-x86_64-gnu\\)$"))] as $matches
             | ($matches | length) == 1 and $matches[0].conclusion == "success"
           ' <<<"$jobs" >/dev/null
           artifact=$(gh api "repos/rustfs/rustfs/actions/artifacts/${ARTIFACT_ID}")
@@ -183,7 +191,7 @@ jobs:
               assert outer.testzip() is None
               with zipfile.ZipFile(io.BytesIO(outer.read(package))) as inner:
                   entries = inner.infolist()
-                  expected = {'rustfs', 'rustfs-cli', 'rustfs.cpu-symbol-catalog.json', 'rustfs.cpu-symbol-catalog.sha256'}
+                  expected = {'rustfs', 'rustfs.cpu-symbol-catalog.json', 'rustfs.cpu-symbol-catalog.sha256'}
                   assert {entry.filename for entry in entries} == expected and len(entries) == len(expected)
                   assert all(regular(entry) and 0 < entry.file_size <= 1073741824 for entry in entries)
                   assert inner.testzip() is None
@@ -213,7 +221,7 @@ jobs:
           npm --prefix connect-harness/web ci
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_TOP_DISK_EVIDENCE="$GITHUB_WORKSPACE/top-disk-service-job-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=top-disk

.github/workflows/connect-top-locks-acceptance.yml

@@ -73,6 +73,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -107,7 +115,7 @@ jobs:
           [[ $(jq -r '.head_repository.full_name' <<<"$run") == "$GITHUB_REPOSITORY" ]]
           [[ $(jq -r '.name' <<<"$run") == "Build and Release" ]]
 
-          expected_job='Build RustFS (linux-x86_64-gnu, sm-standard-4, x86_64-unknown-linux-gnu, false, linux, pyroscope)'
+          expected_job='Build RustFS (linux-x86_64-gnu)'
           jobs=$(gh api --paginate --slurp "repos/${GITHUB_REPOSITORY}/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e --arg name "$expected_job" '
             [.[].jobs[] | select(.name == $name)] as $matches
@@ -154,7 +162,7 @@ jobs:
           npm --prefix connect-harness/web ci
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_TOP_LOCKS_EVIDENCE="$GITHUB_WORKSPACE/top-locks-service-job-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=top-locks

.github/workflows/connect-top-rpc-acceptance.yml

@@ -73,6 +73,14 @@ jobs:
           ref: ${{ inputs.connect_sha }}
           token: ${{ secrets.PF_TESTING_GH_TOKEN }}
 
+      - name: Set up Docker Compose
+        uses: ./.github/actions/setup-docker-compose
+
+      - name: Set up Python
+        uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
+        with:
+          python-version: "3.14"
+
       - name: Set up Node.js
         uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
         with:
@@ -107,7 +115,7 @@ jobs:
           [[ $(jq -r '.head_repository.full_name' <<<"$run") == "$GITHUB_REPOSITORY" ]]
           [[ $(jq -r '.name' <<<"$run") == "Build and Release" ]]
 
-          expected_job='Build RustFS (linux-x86_64-gnu, sm-standard-4, x86_64-unknown-linux-gnu, false, linux, pyroscope)'
+          expected_job='Build RustFS (linux-x86_64-gnu)'
           jobs=$(gh api --paginate --slurp "repos/${GITHUB_REPOSITORY}/actions/runs/${BUILD_RUN_ID}/jobs?per_page=100")
           jq -e --arg name "$expected_job" '
             [.[].jobs[] | select(.name == $name)] as $matches
@@ -154,7 +162,7 @@ jobs:
           npm --prefix connect-harness/web ci
           connect-harness/web/node_modules/.bin/playwright install --with-deps chromium
           make -C connect-harness e2e-connected-dispatch-check
-          RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
+          TMPDIR="$GITHUB_WORKSPACE" RUSTFS_BINARY="$GITHUB_WORKSPACE/binary/rustfs" \
             RUSTFS_WORKTREE="$GITHUB_WORKSPACE/rustfs-source" \
             CONNECT_E2E_TOP_RPC_EVIDENCE="$GITHUB_WORKSPACE/top-rpc-service-job-evidence.json" \
             make -C connect-harness e2e-connected E2E_SCENARIO=top-rpc

.github/workflows/e2e-upgrade.yml

@@ -96,38 +96,8 @@ jobs:
           overwrite: true
 
   upgrade:
-    name: ${{ matrix.name }}
+    name: Upgrade compatibility
     needs: build
-    strategy:
-      fail-fast: false
-      matrix:
-        include:
-          # The two `_from_rc2_` tests keep their names: they assert
-          # release-independent object contracts and pass unchanged against the
-          # newer pinned source, so renaming them would only churn history and
-          # the CI required-check names. UPGRADE_SOURCE_VERSION above is the
-          # single source of truth for which release they actually run against.
-          - name: Direct upgrade from the previous release
-            test: direct_upgrade_from_rc2_preserves_object_contracts
-            artifact: direct-upgrade
-          - name: Mixed-version rolling upgrade from the previous release
-            test: rolling_upgrade_from_rc2_preserves_mixed_version_contracts
-            artifact: mixed-version-upgrade
-          - name: Bucket configuration survives the upgrade
-            test: direct_upgrade_from_previous_release_preserves_bucket_configuration
-            artifact: bucket-config-upgrade
-          - name: Rollback reads current bucket metadata
-            test: rollback_to_previous_release_reads_current_bucket_metadata
-            artifact: bucket-config-rollback
-          - name: ODM configuration recovery after rc.5 rollback
-            test: rc5_rollback_requires_restoring_odm_configuration
-            artifact: odm-config-rollback
-          - name: Multipart layouts survive the rc.5 upgrade
-            test: direct_upgrade_from_rc5_preserves_multipart_layouts
-            artifact: multipart-layout-upgrade
-          - name: rc.5 multipart replication baseline
-            test: rc5_baseline_replicates_multipart_layouts
-            artifact: multipart-layout-baseline
     runs-on: sm-standard-2
     timeout-minutes: 60
     env:
@@ -142,8 +112,7 @@ jobs:
         uses: ./.github/actions/setup
         with:
           cache-shared-key: e2e-upgrade-tests
-          # One main-branch consumer warms the shared test-harness dependencies.
-          cache-save-if: ${{ github.ref == 'refs/heads/main' && matrix.artifact == 'direct-upgrade' }}
+          cache-save-if: ${{ github.ref == 'refs/heads/main' }}
           install-build-packaging-tools: "false"
           install-test-tools: "false"
 
@@ -170,22 +139,49 @@ jobs:
           echo "RUSTFS_UPGRADE_SOURCE_BINARY=$SOURCE_DIR/rustfs" >> "$GITHUB_ENV"
           echo "RUSTFS_E2E_LOG_DIR=$RUNNER_TEMP/rustfs-upgrade-logs" >> "$GITHUB_ENV"
 
-      - name: Run upgrade compatibility test
+      # Reuse one compiled harness, while keeping each case in its own process.
+      # Continue after a failure to retain the former matrix's complete verdict.
+      - name: Run upgrade compatibility tests
+        shell: bash
         env:
-          RUSTFS_SCANNER_HEAL_G09_EVIDENCE_DIR: ${{ runner.temp }}/rustfs-upgrade-g09-evidence/${{ matrix.artifact }}
           NO_PROXY: 127.0.0.1,localhost,::1
         run: |
           export no_proxy="$NO_PROXY"
-          python3 scripts/e2e_binary.py run -- cargo test --locked -p e2e_test \
-            "upgrade_compatibility_test::${{ matrix.test }}" \
-            -- --ignored --exact --nocapture
+          cases=(
+            direct-upgrade:direct_upgrade_from_rc2_preserves_object_contracts
+            mixed-version-upgrade:rolling_upgrade_from_rc2_preserves_mixed_version_contracts
+            bucket-config-upgrade:direct_upgrade_from_previous_release_preserves_bucket_configuration
+            bucket-config-rollback:rollback_to_previous_release_reads_current_bucket_metadata
+            odm-config-rollback:rc5_rollback_requires_restoring_odm_configuration
+            multipart-layout-upgrade:direct_upgrade_from_rc5_preserves_multipart_layouts
+            multipart-layout-baseline:rc5_baseline_replicates_multipart_layouts
+          )
+          status=0
+          for entry in "${cases[@]}"; do
+            artifact="${entry%%:*}"
+            test="${entry#*:}"
+            export RUSTFS_SCANNER_HEAL_G09_EVIDENCE_DIR="$RUNNER_TEMP/rustfs-upgrade-g09-evidence/$artifact"
+            export RUSTFS_E2E_LOG_DIR="$RUNNER_TEMP/rustfs-upgrade-logs/$artifact"
+            mkdir -p "$RUSTFS_E2E_LOG_DIR"
+            echo "::group::$test"
+            if python3 scripts/e2e_binary.py run -- cargo test --locked -p e2e_test \
+              "upgrade_compatibility_test::$test" -- --ignored --exact --nocapture \
+              2>&1 | tee "$RUSTFS_E2E_LOG_DIR/test.log"; then
+              echo "PASS: $test"
+            else
+              echo "::error::Upgrade compatibility failed: $test"
+              status=1
+            fi
+            echo "::endgroup::"
+          done
+          exit "$status"
 
       - name: Upload scanner/heal G09 evidence
         if: always()
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
         with:
-          name: ${{ matrix.artifact }}-g09-evidence-${{ github.run_number }}
-          path: ${{ runner.temp }}/rustfs-upgrade-g09-evidence/${{ matrix.artifact }}
+          name: upgrade-g09-evidence-${{ github.run_number }}
+          path: ${{ runner.temp }}/rustfs-upgrade-g09-evidence
           if-no-files-found: ignore
           retention-days: 14
 
@@ -193,7 +189,7 @@ jobs:
         if: always()
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
         with:
-          name: ${{ matrix.artifact }}-server-logs-${{ github.run_number }}
+          name: upgrade-server-logs-${{ github.run_number }}
           path: ${{ runner.temp }}/rustfs-upgrade-logs
           if-no-files-found: warn
           retention-days: 14

.github/workflows/nightly-gnu.yml

@@ -80,13 +80,24 @@ jobs:
       - name: Setup Rust environment
         uses: ./.github/actions/setup
         with:
-          cache-shared-key: build-x86_64-unknown-linux-gnu
+          cache-shared-key: build-x86_64-unknown-linux-gnu-glibc-2.31
           cache-save-if: 'false'
+          install-cross-tools: 'true'
+          target: x86_64-unknown-linux-gnu
           install-build-packaging-tools: 'false'
           install-test-tools: 'false'
 
       - name: Build RustFS
-        run: cargo build --release --locked --target x86_64-unknown-linux-gnu -p rustfs --bins
+        # Pin the glibc baseline (`<target>.<glibc>` for cargo-zigbuild) so
+        # nightlies run on older distros such as RHEL/Rocky 9 (glibc 2.34); #8062.
+        run: cargo zigbuild --release --locked --target x86_64-unknown-linux-gnu.2.31 -p rustfs --bins
+
+      - name: Verify glibc symbol baseline
+        run: |
+          python3 scripts/release/check_glibc.py 2.31 \
+            target/x86_64-unknown-linux-gnu/release/rustfs \
+            target/x86_64-unknown-linux-gnu/release/rustfs-cli
+          target/x86_64-unknown-linux-gnu/release/rustfs --version | grep -Fx 'build target : x86_64-unknown-linux-gnu'
 
       - name: Build DEB package
         id: deb

.github/workflows/oidc-keycloak.yml

@@ -20,8 +20,9 @@ on:
       - ".github/workflows/oidc-keycloak.yml"
       - "crates/config/src/constants/oidc.rs"
       - "crates/iam/src/federation/**"
-      - "crates/iam/src/oidc.rs"
+      - "crates/iam/src/oidc/**"
       - "rustfs/src/admin/handlers/oidc.rs"
+      - "rustfs/src/admin/service/oidc_config.rs"
       - "rustfs/src/admin/handlers/sts.rs"
       - "scripts/test/oidc_keycloak_live.sh"
       - "scripts/test/fixtures/keycloak-rustfs-ci-realm.json"
@@ -31,8 +32,9 @@ on:
       - ".github/workflows/oidc-keycloak.yml"
       - "crates/config/src/constants/oidc.rs"
       - "crates/iam/src/federation/**"
-      - "crates/iam/src/oidc.rs"
+      - "crates/iam/src/oidc/**"
       - "rustfs/src/admin/handlers/oidc.rs"
+      - "rustfs/src/admin/service/oidc_config.rs"
       - "rustfs/src/admin/handlers/sts.rs"
       - "scripts/test/oidc_keycloak_live.sh"
       - "scripts/test/fixtures/keycloak-rustfs-ci-realm.json"

.gitignore

@@ -90,3 +90,4 @@ worktrees/*
 # insta scratch files; the accepted .snap files ARE the assertions and are committed
 *.snap.new
 *.pending-snap
+.mimosa

.vscode/launch.json

@@ -7,7 +7,9 @@
             "name": "Debug(only) executable 'rustfs'",
             "env": {
               "RUST_LOG": "rustfs=info,ecstore=info,s3s=info,iam=info",
-              "RUSTFS_SKIP_BACKGROUND_TASK": "on"
+              "RUSTFS_SKIP_BACKGROUND_TASK": "on",
+              // Local test volumes share one physical disk.
+              "RUSTFS_UNSAFE_BYPASS_DISK_CHECK": "true"
               //"RUSTFS_OBS_LOG_DIRECTORY": "./deploy/logs",
               // "RUSTFS_POLICY_PLUGIN_URL":"http://localhost:8181/v1/data/rustfs/authz/allow",
               // "RUSTFS_POLICY_PLUGIN_AUTH_TOKEN":"your-opa-token"
@@ -44,6 +46,7 @@
             "env": {
                 "RUST_LOG": "rustfs=debug,ecstore=info,s3s=debug,iam=debug",
                 "RUSTFS_SKIP_BACKGROUND_TASK": "on",
+                "RUSTFS_UNSAFE_BYPASS_DISK_CHECK": "true",
                 //"RUSTFS_OBS_LOG_DIRECTORY": "./deploy/logs",
                 // "RUSTFS_POLICY_PLUGIN_URL":"http://localhost:8181/v1/data/rustfs/authz/allow",
                 // "RUSTFS_POLICY_PLUGIN_AUTH_TOKEN":"your-opa-token" 
@@ -117,7 +120,8 @@
             "env": {
                 "RUSTFS_ACCESS_KEY": "rustfsadmin",
                 "RUSTFS_SECRET_KEY": "rustfsadmin",
-                //"RUSTFS_VOLUMES": "./target/volume/test{1...4}",
+                "RUSTFS_VOLUMES": "./target/volume/test{1...4}",
+                "RUSTFS_UNSAFE_BYPASS_DISK_CHECK": "true",
                 "RUSTFS_ADDRESS": ":9000",
                 "RUSTFS_CONSOLE_ENABLE": "true",
                 // "RUSTFS_OBS_TRACE_ENDPOINT": "http://127.0.0.1:4318/v1/traces", // jeager otlp http endpoint
@@ -135,41 +139,30 @@
         {
             "type": "lldb",
             "request": "launch",
-            "name": "Debug test_lifecycle_transition_basic",
+            "name": "Debug test_transition_and_restore_flows",
             "cargo": {
                 "args": [
                     "test",
+                    "--no-run",
                     "-p",
                     "rustfs-scanner",
                     "--test",
                     "lifecycle_integration_test",
-                    "serial_tests::test_lifecycle_transition_basic",
                     "-j",
                     "1"
-                ]
+                ],
+                "filter": {
+                    "name": "lifecycle_integration_test",
+                    "kind": "test"
+                }
             },
-            "args": [],
-            "cwd": "${workspaceFolder}"
-        },
-        {
-            "name": "Debug executable target/debug/test",
-            "type":  "lldb",
-            "request": "launch",
-            "program": "${workspaceFolder}/target/debug/deps/lifecycle_integration_test-5915cbfcab491b3b",
             "args": [
-              "--skip",
-              "test_lifecycle_expiry_basic",
-              "--skip",
-              "test_lifecycle_expiry_deletemarker",
-              //"--skip",
-              //"test_lifecycle_transition_basic",
-            ],
-            "cwd": "${workspaceFolder}",
-            //"stopAtEntry": false,
-            //"preLaunchTask": "cargo build",
-            "sourceLanguages": [
-                "rust"
+                "serial_tests::test_transition_and_restore_flows",
+                "--exact",
+                "--ignored",
+                "--test-threads=1"
             ],
+            "cwd": "${workspaceFolder}"
         },
         {
             "name": "Debug executable target/debug/rustfs with sse kms",
@@ -285,7 +278,7 @@
                 "RUSTFS_COMPRESSION_ENABLED": "true",
                 "RUSTFS_KMS_ENABLE": "true",
                 "RUSTFS_KMS_BACKEND": "local",
-                "RUSTFS_KMS_KEY_DIR": "./target/rio-v2/kms-key-dir",
+                "RUSTFS_KMS_KEY_DIR": "${workspaceFolder}/target/rio-v2/kms-key-dir",
                 "RUSTFS_KMS_LOCAL_MASTER_KEY": "my-secret-key", // Some Password
                 "RUSTFS_KMS_DEFAULT_KEY_ID": "rustfs-master-key",
             },
@@ -307,6 +300,7 @@
                 "RUSTFS_ACCESS_KEY": "rustfsadmin",
                 "RUSTFS_SECRET_KEY": "rustfsadmin",
                 "RUSTFS_VOLUMES": "./target/e2e-test/test{1...4}",
+                "RUSTFS_UNSAFE_BYPASS_DISK_CHECK": "true",
                 "RUSTFS_REGION": "us-east-1",
                 "RUSTFS_ADDRESS": ":9000",
                 "RUSTFS_CONSOLE_ENABLE": "true",
@@ -315,7 +309,7 @@
 
                 "RUSTFS_KMS_ENABLE": "true",
                 "RUSTFS_KMS_BACKEND": "local",
-                "RUSTFS_KMS_KEY_DIR": "./target/e2e-key-dir",
+                "RUSTFS_KMS_KEY_DIR": "${workspaceFolder}/target/e2e-key-dir",
                 "RUSTFS_KMS_LOCAL_MASTER_KEY": "my-secret-key", // Some Password
                 "RUSTFS_KMS_DEFAULT_KEY_ID": "rustfs-master-key",
             },

Cargo.toml

@@ -18,7 +18,6 @@ members = [
     "crates/audit", # Audit target management system with multi-target fan-out
     "crates/checksums", # client checksums
     "crates/common", # Shared utilities and data structures
-    "crates/capture-authority", # Durable capture authority storage
     "crates/data-usage", # Shared data usage models and algorithms
     "crates/config", # Configuration management
     "crates/connect-symbol-catalog", # Release-only CPU profile symbol catalog generator
@@ -75,7 +74,7 @@ edition = "2024"
 license = "Apache-2.0"
 repository = "https://github.com/rustfs/rustfs"
 rust-version = "1.99.0"
-version = "1.0.1"
+version = "1.0.2"
 homepage = "https://rustfs.com"
 description = "RustFS is a high-performance distributed object storage software built using Rust, one of the most popular languages worldwide. "
 keywords = ["RustFS", "Minio", "object-storage", "filesystem", "s3"]
@@ -92,56 +91,56 @@ redundant_clone = "warn"
 
 [workspace.dependencies]
 # RustFS Internal Crates
-rustfs = { path = "./rustfs", version = "1.0.1" }
-rustfs-heal = { path = "crates/heal", version = "1.0.1" }
-rustfs-heal-contracts = { path = "crates/heal-contracts", version = "1.0.1" }
-rustfs-scanner-metrics = { path = "crates/scanner-metrics", version = "1.0.1" }
-rustfs-audit = { path = "crates/audit", version = "1.0.1" }
-rustfs-checksums = { path = "crates/checksums", version = "1.0.1" }
-rustfs-common = { path = "crates/common", version = "1.0.1" }
-rustfs-data-usage = { path = "crates/data-usage", version = "1.0.1" }
-rustfs-config = { path = "./crates/config", version = "1.0.1" }
-rustfs-concurrency = { path = "./crates/concurrency", version = "1.0.1" }
-rustfs-credentials = { path = "crates/credentials", version = "1.0.1" }
-rustfs-crypto = { path = "crates/crypto", version = "1.0.1" }
-rustfs-ecstore = { path = "crates/ecstore", version = "1.0.1" }
-rustfs-filemeta = { path = "crates/filemeta", version = "1.0.1" }
-rustfs-iam = { path = "crates/iam", version = "1.0.1" }
-rustfs-keystone = { path = "crates/keystone", version = "1.0.1" }
-rustfs-license = { path = "crates/license", version = "1.0.1" }
-rustfs-lifecycle = { path = "crates/lifecycle", version = "1.0.1" }
-rustfs-kms = { path = "crates/kms", version = "1.0.1" }
-rustfs-lock = { path = "crates/lock", version = "1.0.1" }
-rustfs-madmin = { path = "crates/madmin", version = "1.0.1" }
-rustfs-notify = { path = "crates/notify", version = "1.0.1" }
-rustfs-io-metrics = { path = "crates/io-metrics", version = "1.0.1" }
-rustfs-io-core = { path = "crates/io-core", version = "1.0.1" }
-rustfs-object-capacity = { path = "crates/object-capacity", version = "1.0.1" }
-rustfs-object-data-cache = { path = "crates/object-data-cache", version = "1.0.1", default-features = false }
-rustfs-log-analyzer = { path = "crates/log-analyzer", version = "1.0.1" }
-rustfs-obs = { path = "crates/obs", version = "1.0.1" }
-rustfs-policy = { path = "crates/policy", version = "1.0.1" }
-rustfs-protos = { path = "crates/protos", version = "1.0.1" }
-rustfs-protocols = { path = "crates/protocols", version = "1.0.1" }
-rustfs-replication = { path = "crates/replication", version = "1.0.1" }
-rustfs-rio = { path = "crates/rio", version = "1.0.1" }
-rustfs-rio-v2 = { path = "crates/rio-v2", version = "1.0.1" }
-rustfs-s3-client = { path = "crates/s3-client", version = "1.0.1" }
-rustfs-s3-types = { path = "crates/s3-types", version = "1.0.1" }
-rustfs-s3-ops = { path = "crates/s3-ops", version = "1.0.1" }
-rustfs-s3select-api = { path = "crates/s3select-api", version = "1.0.1" }
-rustfs-s3select-query = { path = "crates/s3select-query", version = "1.0.1" }
-rustfs-scanner = { path = "crates/scanner", version = "1.0.1" }
-rustfs-security-governance = { path = "crates/security-governance", version = "1.0.1" }
-rustfs-extension-schema = { path = "crates/extension-schema", version = "1.0.1" }
-rustfs-signer = { path = "crates/signer", version = "1.0.1" }
-rustfs-storage-api = { path = "crates/storage-api", version = "1.0.1" }
-rustfs-trusted-proxies = { path = "crates/trusted-proxies", version = "1.0.1" }
-rustfs-targets = { path = "crates/targets", version = "1.0.1" }
-rustfs-test-utils = { path = "crates/test-utils", version = "1.0.1" }
-rustfs-tls-runtime = { path = "crates/tls-runtime", version = "1.0.1" }
-rustfs-utils = { path = "crates/utils", version = "1.0.1" }
-rustfs-zip = { path = "./crates/zip", version = "1.0.1" }
+rustfs = { path = "./rustfs", version = "1.0.2" }
+rustfs-heal = { path = "crates/heal", version = "1.0.2" }
+rustfs-heal-contracts = { path = "crates/heal-contracts", version = "1.0.2" }
+rustfs-scanner-metrics = { path = "crates/scanner-metrics", version = "1.0.2" }
+rustfs-audit = { path = "crates/audit", version = "1.0.2" }
+rustfs-checksums = { path = "crates/checksums", version = "1.0.2" }
+rustfs-common = { path = "crates/common", version = "1.0.2" }
+rustfs-data-usage = { path = "crates/data-usage", version = "1.0.2" }
+rustfs-config = { path = "./crates/config", version = "1.0.2" }
+rustfs-concurrency = { path = "./crates/concurrency", version = "1.0.2" }
+rustfs-credentials = { path = "crates/credentials", version = "1.0.2" }
+rustfs-crypto = { path = "crates/crypto", version = "1.0.2" }
+rustfs-ecstore = { path = "crates/ecstore", version = "1.0.2" }
+rustfs-filemeta = { path = "crates/filemeta", version = "1.0.2" }
+rustfs-iam = { path = "crates/iam", version = "1.0.2" }
+rustfs-keystone = { path = "crates/keystone", version = "1.0.2" }
+rustfs-license = { path = "crates/license", version = "1.0.2" }
+rustfs-lifecycle = { path = "crates/lifecycle", version = "1.0.2" }
+rustfs-kms = { path = "crates/kms", version = "1.0.2" }
+rustfs-lock = { path = "crates/lock", version = "1.0.2" }
+rustfs-madmin = { path = "crates/madmin", version = "1.0.2" }
+rustfs-notify = { path = "crates/notify", version = "1.0.2" }
+rustfs-io-metrics = { path = "crates/io-metrics", version = "1.0.2" }
+rustfs-io-core = { path = "crates/io-core", version = "1.0.2" }
+rustfs-object-capacity = { path = "crates/object-capacity", version = "1.0.2" }
+rustfs-object-data-cache = { path = "crates/object-data-cache", version = "1.0.2", default-features = false }
+rustfs-log-analyzer = { path = "crates/log-analyzer", version = "1.0.2" }
+rustfs-obs = { path = "crates/obs", version = "1.0.2" }
+rustfs-policy = { path = "crates/policy", version = "1.0.2" }
+rustfs-protos = { path = "crates/protos", version = "1.0.2" }
+rustfs-protocols = { path = "crates/protocols", version = "1.0.2" }
+rustfs-replication = { path = "crates/replication", version = "1.0.2" }
+rustfs-rio = { path = "crates/rio", version = "1.0.2" }
+rustfs-rio-v2 = { path = "crates/rio-v2", version = "1.0.2" }
+rustfs-s3-client = { path = "crates/s3-client", version = "1.0.2" }
+rustfs-s3-types = { path = "crates/s3-types", version = "1.0.2" }
+rustfs-s3-ops = { path = "crates/s3-ops", version = "1.0.2" }
+rustfs-s3select-api = { path = "crates/s3select-api", version = "1.0.2" }
+rustfs-s3select-query = { path = "crates/s3select-query", version = "1.0.2" }
+rustfs-scanner = { path = "crates/scanner", version = "1.0.2" }
+rustfs-security-governance = { path = "crates/security-governance", version = "1.0.2" }
+rustfs-extension-schema = { path = "crates/extension-schema", version = "1.0.2" }
+rustfs-signer = { path = "crates/signer", version = "1.0.2" }
+rustfs-storage-api = { path = "crates/storage-api", version = "1.0.2" }
+rustfs-trusted-proxies = { path = "crates/trusted-proxies", version = "1.0.2" }
+rustfs-targets = { path = "crates/targets", version = "1.0.2" }
+rustfs-test-utils = { path = "crates/test-utils", version = "1.0.2" }
+rustfs-tls-runtime = { path = "crates/tls-runtime", version = "1.0.2" }
+rustfs-utils = { path = "crates/utils", version = "1.0.2" }
+rustfs-zip = { path = "./crates/zip", version = "1.0.2" }
 
 # Async Runtime and Networking
 async-channel = "2.5.0"
@@ -157,8 +156,9 @@ futures-lite = "2.6.1"
 futures-util = "0.3.34"
 pollster = "1.0.1"
 pulsar = { default-features = false, version = "6.9.0" }
-lapin = { default-features = false, version = "4.12.0" }
+lapin = { default-features = false, version = "4.12.2" }
 hyper = { version = "1.12.0" }
+h2 = { version = "0.4.20" }
 hyper-rustls = { default-features = false, version = "0.27.10" }
 hyper-util = { version = "0.1.21" }
 http = "1.5.0"
@@ -169,11 +169,11 @@ minlz = "1.2.3"
 reqwest = "0.13.5"
 rustfs-kafka-async = { version = "1.4.0" }
 socket2 = { version = "0.6.5" }
-tokio = { version = "1.53.2" }
+tokio = { version = "1.53.3" }
 tokio-rustls = { default-features = false, version = "0.26.6" }
 tokio-stream = { version = "0.1.19" }
 tokio-test = "0.4.6"
-tokio-util = { version = "0.7.19" }
+tokio-util = { version = "0.7.20" }
 tonic = { version = "0.14.6" }
 tonic-prost = { version = "0.14.6" }
 tonic-prost-build = { version = "0.14.6" }
@@ -181,7 +181,7 @@ tower = { version = "0.5.3" }
 tower-http = { version = "0.7.1" }
 h3 = "0.0.8"
 h3-quinn = "0.0.10"
-quinn = { version = "0.11.11", default-features = false, features = [
+quinn = { version = "0.11.12", default-features = false, features = [
       "futures-io",
       "runtime-tokio",
       "rustls-aws-lc-rs",
@@ -201,7 +201,7 @@ 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_json = { version = "1.0.154" }
 serde_urlencoded = "0.7.1"
 
 # Cryptography and Security
@@ -225,7 +225,7 @@ rustls-native-certs = "0.8"
 rustls-pki-types = "1.15.1"
 x509-parser = "0.18.1"
 sha1 = "0.11.0"
-sha2 = "0.11.0"
+sha2 = "0.11.1"
 subtle = "2.6"
 zeroize = { version = "1.9.1" }
 proptest = "1"
@@ -269,7 +269,7 @@ crossbeam-queue = "0.3.14"
 crossbeam-channel = "0.5.17"
 crossbeam-deque = "0.8.8"
 crossbeam-utils = "0.8.23"
-datafusion = { default-features = false, version = "55.1.0" }
+datafusion = { default-features = false, version = "55.2.0" }
 derive_builder = "0.20.2"
 ed25519-dalek = "3.0.0"
 enumset = "1.1.14"
@@ -315,21 +315,19 @@ ratelimit = "2.0.0"
 rayon = "1.12.0"
 rustfs-erasure-codec = { version = "9.0.0" }
 reed-solomon-simd = "3.1.0"
-openraft = { version = "=0.9.25", default-features = false, features = ["serde"] }
-redb = "=4.3.0"
 regex = { version = "1.13.1" }
 rumqttc = { package = "rumqttc-next", version = "0.34.0" }
 redis = { version = "1.7.1" }
 rustify = { version = "0.7", default-features = false }
 rustix = { version = "1.1.5" }
-rust-embed = { version = "8.12.0" }
+rust-embed = { version = "8.13.0" }
 rustc-hash = { version = "2.1.3" }
 s3s = { git = "https://github.com/s3s-project/s3s", rev = "5761ddfe4c509fcac87e3b99fd5bd9e176b442c0", features = ["minio"] }
 s3s-sigv4 = { git = "https://github.com/s3s-project/s3s", rev = "5761ddfe4c509fcac87e3b99fd5bd9e176b442c0" }
 serial_test = "4.0.1"
 shadow-rs = { default-features = false, version = "2.0.0" }
 siphasher = "1.0.4"
-smallvec = { version = "1.16.2" }
+smallvec = { version = "1.16.3" }
 compact_str = "0.10.0"
 snap = "1.1.2"
 starshard = { version = "2.4.0" }
@@ -343,12 +341,12 @@ tracing = { version = "0.1.44" }
 tracing-appender = "0.2.5"
 tracing-core = "0.1.36"
 tracing-error = "0.2.1"
-tracing-opentelemetry = { version = "0.34" }
+tracing-opentelemetry = { version = "0.34.0" }
 tracing-subscriber = { version = "0.3.23" }
 transform-stream = "0.3.1"
 url = "2.5.8"
 urlencoding = "2.1.3"
-uuid = { version = "1.27.0" }
+uuid = { version = "1.28.0" }
 vaultrs = { version = "0.8.0" }
 tar = "0.4.46"
 walkdir = "2.5.0"
@@ -356,26 +354,26 @@ winapi-util = "0.1.11"
 windows = { version = "0.62.2" }
 windows-sys = "0.61.2"
 xxhash-rust = { version = "0.8.19" }
-zip = "8.6.0"
-zstd = "0.14.0"
+zip = "9.0.3"
+zstd = "0.14.2"
 
 # Observability and Metrics
 metrics = "0.24.6"
 metrics-util = "0.20"
-dial9 = { version = "0.5.3", default-features = false }
-opentelemetry = { version = "0.33.0" }
-opentelemetry-appender-tracing = { version = "0.33.0" }
-opentelemetry-otlp = { version = "0.33.0" }
-opentelemetry-proto = { version = "0.33.0", default-features = false, features = ["metrics", "gen-tonic-messages"] }
-opentelemetry_sdk = { version = "0.33.0" }
-opentelemetry-semantic-conventions = { version = "0.33.0" }
-opentelemetry-stdout = { version = "0.33.0" }
-pyroscope = { version = "2.1.1" }
+dial9 = { version = "0.5.4", default-features = false }
+opentelemetry = { version = "0.33.1" }
+opentelemetry-appender-tracing = { version = "0.33.1" }
+opentelemetry-otlp = { version = "0.33.1" }
+opentelemetry-proto = { version = "0.33.1", default-features = false, features = ["metrics", "gen-tonic-messages"] }
+opentelemetry_sdk = { version = "0.33.1" }
+opentelemetry-semantic-conventions = { version = "0.33.1" }
+opentelemetry-stdout = { version = "0.33.1" }
+pyroscope = { version = "2.1.2" }
 
 # FTP and SFTP
 libunftp = { version = "0.23.1" }
 unftp-core = "0.1.0"
-suppaftp = { version = "12.1.1" }
+suppaftp = { version = "12.2.0" }
 rcgen = { version = "0.14.10", default-features = false, features = ["aws_lc_rs", "crypto", "pem"] }
 russh = { version = "0.64.1" }
 russh-sftp = "3.0.1"
@@ -386,7 +384,7 @@ dav-server = "0.11.0"
 # Performance Analysis and Memory Profiling
 rustfs-mimalloc = { version = "0.5.6" }
 # Preserve Unicode focus filters until rustfs/backlog#2302 is resolved.
-hotpath = { version = "0.28.5", default-features = false }
+hotpath = { version = "0.28.6", default-features = false }
 # Snapshot testing for output format regression detection
 insta = { version = "1.49" }
 

_typos.toml

@@ -51,6 +51,8 @@ unparseable = "unparseable"
 # Disaster-recovery objectives: recovery time and recovery point.
 RTO = "RTO"
 rto = "rto"
+# rustix OFlags variant (POSIX O_WRONLY equivalent).
+WRONLY = "WRONLY"
 
 [files]
 extend-exclude = []

crates/capture-authority/Cargo.toml

@@ -1,23 +0,0 @@
-[package]
-name = "rustfs-capture-authority"
-version.workspace = true
-edition.workspace = true
-license.workspace = true
-rust-version.workspace = true
-repository.workspace = true
-description = "Durable local storage for capture authority"
-
-[dependencies]
-openraft = { workspace = true }
-redb = { workspace = true }
-serde = { workspace = true, features = ["derive"] }
-thiserror = { workspace = true }
-tokio = { workspace = true }
-url = { workspace = true }
-
-[dev-dependencies]
-serde_json = { workspace = true }
-tempfile = { workspace = true }
-
-[lints]
-workspace = true

crates/capture-authority/src/application.rs

@@ -1,862 +0,0 @@
-fn read_native_range_in<R: std::ops::RangeBounds<u64>>(
-    tx: &redb::ReadTransaction,
-    range: R,
-) -> Result<Vec<crate::CaptureRaftEntry>, StoreError> {
-    use std::ops::Bound::{Excluded, Included, Unbounded};
-    let logs = tx.open_table(LOGS)?;
-    let start = match range.start_bound() {
-        Included(value) => *value,
-        Excluded(value) => match value.checked_add(1) {
-            Some(value) => value,
-            None => return Ok(Vec::new()),
-        },
-        Unbounded => 0,
-    };
-    let end = match range.end_bound() {
-        Included(value) => *value,
-        Excluded(value) => match value.checked_sub(1) {
-            Some(value) => value,
-            None => return Ok(Vec::new()),
-        },
-        Unbounded => u64::MAX,
-    };
-    if start > end {
-        return Ok(Vec::new());
-    }
-    let mut entries = Vec::new();
-    for row in logs.range(start..=end)? {
-        let (key, value) = row?;
-        let entry = decode_log(value.value())?;
-        if entry.id.index != key.value() {
-            return Err(StoreError::CorruptRecord);
-        }
-        entries.push(decode_native_entry(entry)?);
-    }
-    Ok(entries)
-}
-
-use crate::{
-    LOGS, LogId, META, Store, StoreError, committed::validate_committed, created::*, decode_log, encode_id, encode_log,
-    next_index, number,
-};
-use redb::{ReadTransaction, ReadableDatabase, ReadableTable, ReadableTableMetadata, TableDefinition};
-use std::collections::BTreeMap;
-
-pub(crate) const HEADS: TableDefinition<&[u8], &[u8]> = TableDefinition::new("heads");
-pub(crate) const VERSIONS: TableDefinition<&[u8], &[u8]> = TableDefinition::new("versions");
-pub(crate) const EVENTS: TableDefinition<&[u8], &[u8]> = TableDefinition::new("events");
-pub(crate) const OPERATIONS: TableDefinition<&[u8], &[u8]> = TableDefinition::new("operations");
-pub(crate) const APPLY_RESULTS: TableDefinition<u64, &[u8]> = TableDefinition::new("apply_results");
-
-fn version_key(object: &ObjectIdentity, version: [u8; 16]) -> Result<Vec<u8>, StoreError> {
-    let mut b = object_bytes(object)?;
-    b.extend(version);
-    Ok(b)
-}
-fn decode_version(b: &[u8]) -> Result<CreatedVersion, StoreError> {
-    let mut d = Decoder::new(b);
-    let v = CreatedVersion {
-        prepared: d.prepared()?,
-        decision_log_id: d.id()?,
-        binding_revision: d.u64()?,
-    };
-    d.finish()?;
-    Ok(v)
-}
-fn decode_event(b: &[u8]) -> Result<CreatedEvent, StoreError> {
-    let mut d = Decoder::new(b);
-    let e = CreatedEvent {
-        object: d.object()?,
-        prepared: d.prepared()?,
-        decision_log_id: d.id()?,
-        binding: d.binding()?,
-    };
-    d.finish()?;
-    Ok(e)
-}
-fn decode_operation(b: &[u8]) -> Result<CreatedOperation, StoreError> {
-    let mut d = Decoder::new(b);
-    let len = usize::try_from(d.u64()?).map_err(|_| StoreError::CorruptRecord)?;
-    let command = b
-        .get(8..8usize.checked_add(len).ok_or(StoreError::CorruptRecord)?)
-        .ok_or(StoreError::CorruptRecord)?;
-    DecideCreated::decode(command)?;
-    let mut d = Decoder::new(&b[8 + len..]);
-    let result = d.result()?;
-    d.finish()?;
-    Ok(CreatedOperation {
-        canonical_command: command.to_vec(),
-        result,
-    })
-}
-fn decode_apply(b: &[u8]) -> Result<CreatedApplyResult, StoreError> {
-    let mut d = Decoder::new(b);
-    let r = CreatedApplyResult {
-        log_id: d.id()?,
-        result: d.result()?,
-    };
-    d.finish()?;
-    Ok(r)
-}
-fn read_binding(meta: &impl ReadableTable<&'static str, &'static [u8]>) -> Result<CaptureBinding, StoreError> {
-    let b = meta.get("capture_binding")?.ok_or(StoreError::BindingNotPinned)?;
-    let mut d = Decoder::new(b.value());
-    let binding = d.binding()?;
-    d.finish()?;
-    Ok(binding)
-}
-fn read_applied(meta: &impl ReadableTable<&'static str, &'static [u8]>) -> Result<Option<LogId>, StoreError> {
-    let b = meta.get("last_applied")?.ok_or(StoreError::CorruptRecord)?;
-    let mut d = Decoder::new(b.value());
-    let id = match d.take::<1>()?[0] {
-        0 => None,
-        1 => Some(d.id()?),
-        _ => return Err(StoreError::CorruptRecord),
-    };
-    d.finish()?;
-    Ok(id)
-}
-fn required<T>(value: Option<T>) -> Result<T, StoreError> {
-    value.ok_or(StoreError::CorruptRecord)
-}
-fn encode_native_entry(entry: &crate::CaptureRaftEntry) -> Result<Vec<u8>, StoreError> {
-    let id: LogId = entry.log_id.into();
-    let (kind, body) = match &entry.payload {
-        openraft::EntryPayload::Blank => (0, vec![]),
-        openraft::EntryPayload::Normal(command) => (1, command.created().encode()?),
-        openraft::EntryPayload::Membership(membership) => {
-            let stored = openraft::StoredMembership::new(Some(entry.log_id), membership.clone());
-            let checked = crate::CaptureMembership::try_from(&stored)?;
-            (2, checked.encode()?)
-        }
-    };
-    let mut payload = b"CENR\x01".to_vec();
-    payload.push(kind);
-    payload.extend(
-        u64::try_from(body.len())
-            .map_err(|_| StoreError::CorruptRecord)?
-            .to_be_bytes(),
-    );
-    payload.extend(body);
-    encode_log(&crate::LogEntry { id, payload })
-}
-
-pub(crate) fn decode_native_entry(entry: crate::LogEntry) -> Result<crate::CaptureRaftEntry, StoreError> {
-    let bytes = &entry.payload;
-    if bytes.len() < 14 || &bytes[..5] != b"CENR\x01" {
-        return Err(StoreError::CorruptRecord);
-    }
-    let length = usize::try_from(number(&bytes[6..14])?).map_err(|_| StoreError::CorruptRecord)?;
-    let body = &bytes[14..];
-    if length != body.len() {
-        return Err(StoreError::CorruptRecord);
-    }
-    let payload = match bytes[5] {
-        0 if body.is_empty() => openraft::EntryPayload::Blank,
-        1 => openraft::EntryPayload::Normal(crate::CaptureCommand::try_created(DecideCreated::decode(body)?)?),
-        2 => {
-            let membership = crate::CaptureMembership::decode(body).map_err(|_| StoreError::CorruptRecord)?;
-            if *membership.log_id() != Some(entry.id) {
-                return Err(StoreError::CorruptRecord);
-            }
-            let stored = openraft::StoredMembership::try_from(&membership).map_err(|_| StoreError::CorruptRecord)?;
-            openraft::EntryPayload::Membership(stored.membership().clone())
-        }
-        _ => return Err(StoreError::CorruptRecord),
-    };
-    Ok(crate::CaptureRaftEntry {
-        log_id: entry.id.into(),
-        payload,
-    })
-}
-
-fn apply_created(
-    tx: &redb::WriteTransaction,
-    id: LogId,
-    command: &DecideCreated,
-    canonical: &[u8],
-    binding: &CaptureBinding,
-) -> Result<CreatedResult, StoreError> {
-    let mut operations = tx.open_table(OPERATIONS)?;
-    let existing = operations
-        .get(command.operation_id.as_slice())?
-        .map(|v| decode_operation(v.value()))
-        .transpose()?;
-    let result = if let Some(existing) = existing {
-        if existing.canonical_command == canonical {
-            existing.result
-        } else {
-            CreatedResult::OperationConflict
-        }
-    } else {
-        let mut heads = tx.open_table(HEADS)?;
-        let mut versions = tx.open_table(VERSIONS)?;
-        let object = object_bytes(&command.object)?;
-        let key = version_key(&command.object, command.prepared.version_id)?;
-        let actual = heads
-            .get(object.as_slice())?
-            .map(|v| {
-                let mut d = Decoder::new(v.value());
-                let id = d.uuid()?;
-                d.finish()?;
-                Ok::<_, StoreError>(id)
-            })
-            .transpose()?;
-        let outcome = if command.binding_revision != binding.revision {
-            CreatedResult::BindingMismatch
-        } else if actual != command.expected_head {
-            CreatedResult::HeadMismatch { actual_head: actual }
-        } else if versions.get(key.as_slice())?.is_some() {
-            CreatedResult::VersionConflict
-        } else {
-            let mut version = Vec::new();
-            put_prepared(&mut version, &command.prepared);
-            version.extend(encode_id(id));
-            version.extend(binding.revision.to_be_bytes());
-            versions.insert(key.as_slice(), version.as_slice())?;
-            let mut event = object;
-            put_prepared(&mut event, &command.prepared);
-            event.extend(encode_id(id));
-            event.extend(binding_bytes(&binding)?);
-            tx.open_table(EVENTS)?
-                .insert(command.operation_id.as_slice(), event.as_slice())?;
-            heads.insert(object_bytes(&command.object)?.as_slice(), command.prepared.version_id.as_slice())?;
-            CreatedResult::Created {
-                version_id: command.prepared.version_id,
-                event_id: command.operation_id,
-                decision_log_id: id,
-            }
-        };
-        let mut op = u64::try_from(canonical.len())
-            .map_err(|_| StoreError::CorruptRecord)?
-            .to_be_bytes()
-            .to_vec();
-        op.extend(canonical);
-        op.extend(result_bytes(&outcome));
-        operations.insert(command.operation_id.as_slice(), op.as_slice())?;
-        outcome
-    };
-    Ok(result)
-}
-
-fn validate_response_kind(entry: &crate::CaptureRaftEntry, response: &crate::CaptureResponse) -> Result<(), StoreError> {
-    let kind = match entry.payload {
-        openraft::EntryPayload::Blank => crate::CaptureResponseKind::Blank,
-        openraft::EntryPayload::Normal(_) => crate::CaptureResponseKind::Created,
-        openraft::EntryPayload::Membership(_) => crate::CaptureResponseKind::Membership,
-    };
-    if response.kind() != kind || *response.log_id() != LogId::from(entry.log_id) {
-        return Err(StoreError::CorruptRecord);
-    }
-    Ok(())
-}
-fn encode_response(response: &crate::CaptureResponse) -> Result<Vec<u8>, StoreError> {
-    let (kind, body) = match response.kind() {
-        crate::CaptureResponseKind::Blank => (0, vec![]),
-        crate::CaptureResponseKind::Created => (1, result_bytes(response.created_result().ok_or(StoreError::CorruptRecord)?)),
-        crate::CaptureResponseKind::Membership => (2, vec![]),
-    };
-    let mut bytes = b"CRSP\x01".to_vec();
-    bytes.push(kind);
-    bytes.extend(encode_id(*response.log_id()));
-    bytes.extend(
-        u64::try_from(body.len())
-            .map_err(|_| StoreError::CorruptRecord)?
-            .to_be_bytes(),
-    );
-    bytes.extend(body);
-    Ok(bytes)
-}
-fn decode_response(bytes: &[u8]) -> Result<crate::CaptureResponse, StoreError> {
-    if bytes.len() < 38
-        || &bytes[..5] != b"CRSP\x01"
-        || number(&bytes[30..38])? != u64::try_from(bytes.len() - 38).map_err(|_| StoreError::CorruptRecord)?
-    {
-        return Err(StoreError::CorruptRecord);
-    }
-    let id = crate::decode_id(&bytes[6..30])?;
-    match bytes[5] {
-        0 if bytes.len() == 38 => crate::CaptureResponse::blank(id),
-        1 => {
-            let mut body = encode_id(id);
-            body.extend(&bytes[38..]);
-            crate::CaptureResponse::created(decode_apply(&body)?)
-        }
-        2 if bytes.len() == 38 => crate::CaptureResponse::membership(id),
-        _ => Err(StoreError::CorruptRecord),
-    }
-}
-fn entry_command(entry: &crate::LogEntry, native: bool) -> Result<DecideCreated, StoreError> {
-    if native {
-        match decode_native_entry(entry.clone())?.payload {
-            openraft::EntryPayload::Normal(command) => Ok(command.created().clone()),
-            _ => Err(StoreError::CorruptRecord),
-        }
-    } else {
-        DecideCreated::decode(&entry.payload)
-    }
-}
-
-pub(crate) fn native_application_state(tx: &ReadTransaction) -> Result<(Option<LogId>, crate::CaptureMembership), StoreError> {
-    let meta = tx.open_table(META)?;
-    if number(required(meta.get("schema")?)?.value())? != 4 {
-        return Err(StoreError::UnsupportedSchema);
-    }
-    let last = read_applied(&meta)?;
-    let membership = crate::CaptureMembership::decode(required(meta.get("native_membership")?)?.value())
-        .map_err(|_| StoreError::CorruptRecord)?;
-    if membership
-        .log_id()
-        .is_some_and(|id| last.is_none_or(|last| id.index > last.index))
-    {
-        return Err(StoreError::CorruptRecord);
-    }
-    Ok((last, membership))
-}
-
-impl Store {
-    /// Explicitly opt an empty, bound store into native entry persistence.
-    pub fn enable_native_empty(&mut self) -> Result<(), StoreError> {
-        let tx = self.write_transaction()?;
-        {
-            let mut meta = tx.open_table(META)?;
-            let schema = number(required(meta.get("schema")?)?.value())?;
-            if !matches!(schema, 3 | 4) {
-                return Err(StoreError::UnsupportedSchema);
-            }
-            read_binding(&meta)?;
-            if schema == 4 {
-                let membership = crate::CaptureMembership::decode(required(meta.get("native_membership")?)?.value())
-                    .map_err(|_| StoreError::CorruptRecord)?;
-                let last = read_applied(&meta)?;
-                if membership
-                    .log_id()
-                    .is_some_and(|id| last.is_none_or(|last| id.index > last.index))
-                    || (last.is_none() && (!membership.configs().is_empty() || !membership.nodes().is_empty()))
-                {
-                    return Err(StoreError::CorruptRecord);
-                }
-                return Ok(());
-            }
-            if read_applied(&meta)?.is_some() || !tx.open_table(LOGS)?.is_empty()? {
-                return Err(StoreError::UnsupportedSchema);
-            }
-            for table in [HEADS, VERSIONS, EVENTS, OPERATIONS] {
-                if !tx.open_table(table)?.is_empty()? {
-                    return Err(StoreError::UnsupportedSchema);
-                }
-            }
-            if !tx.open_table(APPLY_RESULTS)?.is_empty()? || meta.get("committed")?.is_some() || meta.get("tail")?.is_some() {
-                return Err(StoreError::UnsupportedSchema);
-            }
-            if meta.get("native_membership")?.is_some() {
-                return Err(StoreError::CorruptRecord);
-            }
-            let membership = crate::CaptureMembership::new(None, vec![], BTreeMap::new())?;
-            meta.insert("native_membership", membership.encode()?.as_slice())?;
-            meta.insert("schema", 4u64.to_be_bytes().as_slice())?;
-        }
-        self.commit(tx)
-    }
-    pub fn append_native(&mut self, entries: &[crate::CaptureRaftEntry]) -> Result<(), StoreError> {
-        let tx = self.write_transaction()?;
-        {
-            let mut meta = tx.open_table(META)?;
-            if number(required(meta.get("schema")?)?.value())? != 4 {
-                return Err(StoreError::UnsupportedSchema);
-            }
-            if entries.is_empty() {
-                return Ok(());
-            }
-            let tail = meta.get("tail")?.map(|value| crate::decode_id(value.value())).transpose()?;
-            let mut next = next_index(tail.map(|id| id.index))?;
-            let mut encoded = Vec::with_capacity(entries.len());
-            for (position, entry) in entries.iter().enumerate() {
-                if entry.log_id.index != next {
-                    return Err(StoreError::InvalidLogSequence);
-                }
-                encoded.push(encode_native_entry(entry)?);
-                if position + 1 < entries.len() {
-                    next = next_index(Some(next))?;
-                }
-            }
-            let mut logs = tx.open_table(LOGS)?;
-            for (entry, bytes) in entries.iter().zip(&encoded) {
-                logs.insert(entry.log_id.index, bytes.as_slice())?;
-            }
-            let last = entries.last().ok_or(StoreError::InvalidLogSequence)?;
-            meta.insert("tail", encode_id(last.log_id.into()).as_slice())?;
-        }
-        self.commit(tx)
-    }
-    pub fn read_native_entry(&self, index: u64) -> Result<Option<crate::CaptureRaftEntry>, StoreError> {
-        let tx = self.database.begin_read()?;
-        if number(required(tx.open_table(META)?.get("schema")?)?.value())? != 4 {
-            return Err(StoreError::UnsupportedSchema);
-        }
-        tx.open_table(LOGS)?
-            .get(index)?
-            .map(|value| {
-                let entry = decode_log(value.value())?;
-                if entry.id.index != index {
-                    return Err(StoreError::CorruptRecord);
-                }
-                decode_native_entry(entry)
-            })
-            .transpose()
-    }
-    /// Read typed native entries within the requested bounds.
-    pub fn read_native_range<R: std::ops::RangeBounds<u64>>(&self, range: R) -> Result<Vec<crate::CaptureRaftEntry>, StoreError> {
-        let tx = self.database.begin_read()?;
-        if number(required(tx.open_table(META)?.get("schema")?)?.value())? != 4 {
-            return Err(StoreError::UnsupportedSchema);
-        }
-        read_native_range_in(&tx, range)
-    }
-    pub fn read_native_entries(&self, start: u64, end: u64) -> Result<Vec<crate::CaptureRaftEntry>, StoreError> {
-        let tx = self.database.begin_read()?;
-        if number(required(tx.open_table(META)?.get("schema")?)?.value())? != 4 {
-            return Err(StoreError::UnsupportedSchema);
-        }
-        if start > end {
-            return Err(StoreError::InvalidLogSequence);
-        }
-        read_native_range_in(&tx, start..end)
-    }
-
-    pub fn apply_next_native(&mut self, id: LogId) -> Result<crate::CaptureResponse, StoreError> {
-        let tx = self.write_transaction()?;
-        let response;
-        {
-            let mut meta = tx.open_table(META)?;
-            if number(required(meta.get("schema")?)?.value())? != 4 {
-                return Err(StoreError::UnsupportedSchema);
-            }
-            let binding = read_binding(&meta)?;
-            let logs = tx.open_table(LOGS)?;
-            let committed = validate_committed(&meta, &logs)?.ok_or(StoreError::InvalidCommittedLog)?;
-            let entry = decode_log(required(logs.get(id.index)?)?.value())?;
-            if entry.id != id || id.index > committed.index {
-                return Err(StoreError::InvalidCommittedLog);
-            }
-            let native = decode_native_entry(entry)?;
-            let last = read_applied(&meta)?;
-            let mut results = tx.open_table(APPLY_RESULTS)?;
-            if last.is_some_and(|last| id.index <= last.index) {
-                let saved = decode_response(required(results.get(id.index)?)?.value())?;
-                if *saved.log_id() != id {
-                    return Err(StoreError::InvalidApplySequence);
-                }
-                validate_response_kind(&native, &saved)?;
-                return Ok(saved);
-            }
-            if id.index != next_index(last.map(|last| last.index))? {
-                return Err(StoreError::InvalidApplySequence);
-            }
-            response = match native.payload {
-                openraft::EntryPayload::Blank => crate::CaptureResponse::blank(id)?,
-                openraft::EntryPayload::Normal(command) => {
-                    let canonical = command.created().encode()?;
-                    let result = apply_created(&tx, id, command.created(), &canonical, &binding)?;
-                    crate::CaptureResponse::created(CreatedApplyResult { log_id: id, result })?
-                }
-                openraft::EntryPayload::Membership(membership) => {
-                    let stored = openraft::StoredMembership::new(Some(id.into()), membership);
-                    let checked = crate::CaptureMembership::try_from(&stored)?;
-                    meta.insert("native_membership", checked.encode()?.as_slice())?;
-                    crate::CaptureResponse::membership(id)?
-                }
-            };
-            results.insert(id.index, encode_response(&response)?.as_slice())?;
-            let mut last = vec![1];
-            last.extend(encode_id(id));
-            meta.insert("last_applied", last.as_slice())?;
-        }
-        self.commit(tx)?;
-        Ok(response)
-    }
-    pub fn native_apply_response(&self, index: u64) -> Result<Option<crate::CaptureResponse>, StoreError> {
-        let tx = self.database.begin_read()?;
-        if number(required(tx.open_table(META)?.get("schema")?)?.value())? != 4 {
-            return Err(StoreError::UnsupportedSchema);
-        }
-        let results = tx.open_table(APPLY_RESULTS)?;
-        results.get(index)?.map(|v| decode_response(v.value())).transpose()
-    }
-    pub fn native_application_state(&self) -> Result<(Option<LogId>, crate::CaptureMembership), StoreError> {
-        native_application_state(&self.database.begin_read()?)
-    }
-    /// Bootstrap the immutable capture policy. The caller must be the trusted
-    /// deployment provisioning owner. This crate does not authenticate callers;
-    /// do not expose this public method through an ordinary object request API.
-    /// Prepared identities assert identity only, not physical data durability.
-    pub fn pin_capture_binding(&mut self, binding: CaptureBinding) -> Result<(), StoreError> {
-        let encoded = binding_bytes(&binding)?;
-        let tx = self.write_transaction()?;
-        {
-            let mut meta = tx.open_table(META)?;
-            if matches!(number(required(meta.get("schema")?)?.value())?, 3 | 4) {
-                if read_binding(&meta)? == binding {
-                    return Ok(());
-                }
-                return Err(StoreError::BindingConflict);
-            }
-            if meta.get("capture_binding")?.is_some() || meta.get("last_applied")?.is_some() {
-                return Err(StoreError::CorruptRecord);
-            }
-            for table in [HEADS, VERSIONS, EVENTS, OPERATIONS] {
-                if !tx.open_table(table)?.is_empty()? {
-                    return Err(StoreError::CorruptRecord);
-                }
-            }
-            if !tx.open_table(APPLY_RESULTS)?.is_empty()? {
-                return Err(StoreError::CorruptRecord);
-            }
-            meta.insert("capture_binding", encoded.as_slice())?;
-            meta.insert("last_applied", &[0][..])?;
-            meta.insert("schema", 3u64.to_be_bytes().as_slice())?;
-        }
-        self.commit(tx)
-    }
-    /// Apply the next exact persisted committed Created command atomically.
-    pub fn apply_next_created(&mut self, id: LogId) -> Result<CreatedResult, StoreError> {
-        let tx = self.write_transaction()?;
-        let result;
-        {
-            let mut meta = tx.open_table(META)?;
-            if number(required(meta.get("schema")?)?.value())? != 3 {
-                return Err(StoreError::UnsupportedSchema);
-            }
-            let binding = read_binding(&meta)?;
-            let logs = tx.open_table(LOGS)?;
-            let committed = validate_committed(&meta, &logs)?.ok_or(StoreError::InvalidCommittedLog)?;
-            let entry = decode_log(required(logs.get(id.index)?)?.value())?;
-            if entry.id != id || id.index > committed.index {
-                return Err(StoreError::InvalidCommittedLog);
-            }
-            let last = read_applied(&meta)?;
-            let mut results = tx.open_table(APPLY_RESULTS)?;
-            if last.is_some_and(|last| id.index <= last.index) {
-                let saved = decode_apply(required(results.get(id.index)?)?.value())?;
-                if saved.log_id != id {
-                    return Err(StoreError::InvalidApplySequence);
-                }
-                return Ok(saved.result);
-            }
-            if id.index != next_index(last.map(|last| last.index))? {
-                return Err(StoreError::InvalidApplySequence);
-            }
-            let command = DecideCreated::decode(&entry.payload)?;
-            result = apply_created(&tx, id, &command, &entry.payload, &binding)?;
-            let mut applied = encode_id(id);
-            applied.extend(result_bytes(&result));
-            results.insert(id.index, applied.as_slice())?;
-            let mut last = vec![1];
-            last.extend(encode_id(id));
-            meta.insert("last_applied", last.as_slice())?;
-        }
-        self.commit(tx)?;
-        Ok(result)
-    }
-    pub fn capture_binding(&self) -> Result<Option<CaptureBinding>, StoreError> {
-        let tx = self.database.begin_read()?;
-        let meta = tx.open_table(META)?;
-        if !matches!(number(required(meta.get("schema")?)?.value())?, 3 | 4) {
-            return Ok(None);
-        }
-        Ok(Some(read_binding(&meta)?))
-    }
-    pub fn last_applied(&self) -> Result<Option<LogId>, StoreError> {
-        let tx = self.database.begin_read()?;
-        let meta = tx.open_table(META)?;
-        if !matches!(number(required(meta.get("schema")?)?.value())?, 3 | 4) {
-            return Ok(None);
-        }
-        read_applied(&meta)
-    }
-    pub fn created_head(&self, object: &ObjectIdentity) -> Result<Option<[u8; 16]>, StoreError> {
-        let tx = self.database.begin_read()?;
-        let t = tx.open_table(HEADS)?;
-        t.get(object_bytes(object)?.as_slice())?
-            .map(|v| {
-                let mut d = Decoder::new(v.value());
-                let id = d.uuid()?;
-                d.finish()?;
-                Ok(id)
-            })
-            .transpose()
-    }
-    pub fn created_version(&self, object: &ObjectIdentity, version: [u8; 16]) -> Result<Option<CreatedVersion>, StoreError> {
-        let tx = self.database.begin_read()?;
-        let t = tx.open_table(VERSIONS)?;
-        t.get(version_key(object, version)?.as_slice())?
-            .map(|v| decode_version(v.value()))
-            .transpose()
-    }
-    pub fn created_event(&self, operation: [u8; 16]) -> Result<Option<CreatedEvent>, StoreError> {
-        let tx = self.database.begin_read()?;
-        let t = tx.open_table(EVENTS)?;
-        t.get(operation.as_slice())?.map(|v| decode_event(v.value())).transpose()
-    }
-    pub fn created_operation(&self, operation: [u8; 16]) -> Result<Option<CreatedOperation>, StoreError> {
-        let tx = self.database.begin_read()?;
-        let t = tx.open_table(OPERATIONS)?;
-        t.get(operation.as_slice())?.map(|v| decode_operation(v.value())).transpose()
-    }
-    pub fn created_apply_result(&self, index: u64) -> Result<Option<CreatedApplyResult>, StoreError> {
-        let tx = self.database.begin_read()?;
-        let t = tx.open_table(APPLY_RESULTS)?;
-        if number(required(tx.open_table(META)?.get("schema")?)?.value())? == 4 {
-            t.get(index)?
-                .map(|v| {
-                    decode_response(v.value()).map(|r| {
-                        r.created_result().cloned().map(|result| CreatedApplyResult {
-                            log_id: *r.log_id(),
-                            result,
-                        })
-                    })
-                })
-                .transpose()
-                .map(Option::flatten)
-        } else {
-            t.get(index)?.map(|v| decode_apply(v.value())).transpose()
-        }
-    }
-}
-
-pub(crate) fn validate_application(tx: &ReadTransaction) -> Result<(), StoreError> {
-    let meta = tx.open_table(META)?;
-    let schema = number(required(meta.get("schema")?)?.value())?;
-    let native = schema == 4;
-    if !matches!(schema, 3 | 4) {
-        return Ok(());
-    }
-    let binding = read_binding(&meta).map_err(|error| match error {
-        StoreError::BindingNotPinned => StoreError::CorruptRecord,
-        other => other,
-    })?;
-    let last = read_applied(&meta)?;
-    let logs = tx.open_table(LOGS)?;
-    let committed = validate_committed(&meta, &logs)?;
-    let heads = tx.open_table(HEADS)?;
-    let versions = tx.open_table(VERSIONS)?;
-    let events = tx.open_table(EVENTS)?;
-    let operations = tx.open_table(OPERATIONS)?;
-    let results = tx.open_table(APPLY_RESULTS)?;
-    if last.is_some_and(|last| committed.is_none_or(|c| last.index > c.index)) {
-        return Err(StoreError::CorruptRecord);
-    }
-    let mut next = Some(0);
-    let mut observed = None;
-    let mut first_operations = BTreeMap::new();
-    let mut expected_membership = crate::CaptureMembership::new(None, vec![], BTreeMap::new())?;
-    let mut replay_operations = BTreeMap::<[u8; 16], (Vec<u8>, CreatedResult)>::new();
-    let mut replay_heads = BTreeMap::<Vec<u8>, [u8; 16]>::new();
-    let mut replay_versions = std::collections::BTreeSet::new();
-    for row in results.iter()? {
-        let (k, v) = row?;
-        let entry = decode_log(required(logs.get(k.value())?)?.value())?;
-        let r = if native {
-            let response = decode_response(v.value())?;
-            let native_entry = decode_native_entry(entry.clone())?;
-            validate_response_kind(&native_entry, &response)?;
-            if next != Some(k.value()) || response.log_id().index != k.value() {
-                return Err(StoreError::CorruptRecord);
-            }
-            observed = Some(*response.log_id());
-            next = k.value().checked_add(1);
-            match native_entry.payload {
-                openraft::EntryPayload::Membership(membership) => {
-                    expected_membership =
-                        crate::CaptureMembership::try_from(&openraft::StoredMembership::new(Some(entry.id.into()), membership))?;
-                    continue;
-                }
-                openraft::EntryPayload::Blank => continue,
-                openraft::EntryPayload::Normal(_) => CreatedApplyResult {
-                    log_id: *response.log_id(),
-                    result: response.created_result().cloned().ok_or(StoreError::CorruptRecord)?,
-                },
-            }
-        } else {
-            decode_apply(v.value())?
-        };
-        if !native && (next != Some(k.value()) || r.log_id.index != k.value()) {
-            return Err(StoreError::CorruptRecord);
-        }
-        if entry.id != r.log_id {
-            return Err(StoreError::CorruptRecord);
-        }
-        let c = entry_command(&entry, native)?;
-        let canonical = c.encode()?;
-        let op = decode_operation(required(operations.get(c.operation_id.as_slice())?)?.value())?;
-        let canonical_matches = op.canonical_command == canonical;
-        let expected = if canonical_matches {
-            op.result
-        } else {
-            CreatedResult::OperationConflict
-        };
-        if native {
-            let replay = if let Some((previous, result)) = replay_operations.get(&c.operation_id) {
-                if *previous == canonical {
-                    result.clone()
-                } else {
-                    CreatedResult::OperationConflict
-                }
-            } else {
-                let object = object_bytes(&c.object)?;
-                let key = version_key(&c.object, c.prepared.version_id)?;
-                let actual = replay_heads.get(&object).copied();
-                let result = if c.binding_revision != binding.revision {
-                    CreatedResult::BindingMismatch
-                } else if actual != c.expected_head {
-                    CreatedResult::HeadMismatch { actual_head: actual }
-                } else if replay_versions.contains(&key) {
-                    CreatedResult::VersionConflict
-                } else {
-                    replay_heads.insert(object, c.prepared.version_id);
-                    replay_versions.insert(key);
-                    CreatedResult::Created {
-                        version_id: c.prepared.version_id,
-                        event_id: c.operation_id,
-                        decision_log_id: r.log_id,
-                    }
-                };
-                replay_operations.insert(c.operation_id, (canonical.clone(), result.clone()));
-                result
-            };
-            if replay != r.result {
-                return Err(StoreError::CorruptRecord);
-            }
-        }
-        if expected != r.result {
-            return Err(StoreError::CorruptRecord);
-        }
-        if canonical_matches {
-            first_operations.entry(c.operation_id).or_insert(r.log_id);
-        }
-        observed = Some(r.log_id);
-        next = k.value().checked_add(1)
-    }
-    if observed != last {
-        return Err(StoreError::CorruptRecord);
-    }
-    if native {
-        if native_application_state(tx)?.1 != expected_membership
-            || heads.len()? != u64::try_from(replay_heads.len()).map_err(|_| StoreError::CorruptRecord)?
-            || versions.len()? != u64::try_from(replay_versions.len()).map_err(|_| StoreError::CorruptRecord)?
-        {
-            return Err(StoreError::CorruptRecord);
-        }
-        for (object, id) in replay_heads {
-            if required(heads.get(object.as_slice())?)?.value() != id {
-                return Err(StoreError::CorruptRecord);
-            }
-        }
-    }
-    for row in operations.iter()? {
-        let (k, v) = row?;
-        let op = decode_operation(v.value())?;
-        let c = DecideCreated::decode(&op.canonical_command)?;
-        if k.value() != c.operation_id {
-            return Err(StoreError::CorruptRecord);
-        }
-        let first = required(first_operations.get(&c.operation_id))?;
-        if let CreatedResult::Created {
-            version_id,
-            event_id,
-            decision_log_id,
-        } = op.result
-        {
-            if version_id != c.prepared.version_id
-                || event_id != c.operation_id
-                || c.binding_revision != binding.revision
-                || decision_log_id != *first
-            {
-                return Err(StoreError::CorruptRecord);
-            }
-            let event = decode_event(required(events.get(event_id.as_slice())?)?.value())?;
-            if event
-                != (CreatedEvent {
-                    object: c.object.clone(),
-                    prepared: c.prepared.clone(),
-                    decision_log_id,
-                    binding: binding.clone(),
-                })
-            {
-                return Err(StoreError::CorruptRecord);
-            }
-        } else if events.get(c.operation_id.as_slice())?.is_some() {
-            return Err(StoreError::CorruptRecord);
-        }
-    }
-    for row in events.iter()? {
-        let (k, v) = row?;
-        let e = decode_event(v.value())?;
-        if e.binding != binding {
-            return Err(StoreError::CorruptRecord);
-        }
-        let op = decode_operation(required(operations.get(k.value())?)?.value())?;
-        let c = DecideCreated::decode(&op.canonical_command)?;
-        if c.object != e.object
-            || c.prepared != e.prepared
-            || op.result
-                != (CreatedResult::Created {
-                    version_id: e.prepared.version_id,
-                    event_id: c.operation_id,
-                    decision_log_id: e.decision_log_id,
-                })
-        {
-            return Err(StoreError::CorruptRecord);
-        }
-        let version =
-            decode_version(required(versions.get(version_key(&e.object, e.prepared.version_id)?.as_slice())?)?.value())?;
-        if version
-            != (CreatedVersion {
-                prepared: e.prepared.clone(),
-                decision_log_id: e.decision_log_id,
-                binding_revision: binding.revision,
-            })
-        {
-            return Err(StoreError::CorruptRecord);
-        }
-    }
-    for row in versions.iter()? {
-        let (k, v) = row?;
-        let version = decode_version(v.value())?;
-        let mut d = Decoder::new(k.value());
-        let object = d.object()?;
-        let id = d.uuid()?;
-        d.finish()?;
-        if id != version.prepared.version_id || version.binding_revision != binding.revision {
-            return Err(StoreError::CorruptRecord);
-        }
-        let entry = decode_log(required(logs.get(version.decision_log_id.index)?)?.value())?;
-        let c = entry_command(&entry, native)?;
-        if entry.id != version.decision_log_id
-            || c.object != object
-            || c.prepared != version.prepared
-            || events.get(c.operation_id.as_slice())?.is_none()
-        {
-            return Err(StoreError::CorruptRecord);
-        }
-        let head = required(heads.get(object_bytes(&object)?.as_slice())?)?;
-        let mut d = Decoder::new(head.value());
-        let head_id = d.uuid()?;
-        d.finish()?;
-        let current = decode_version(required(versions.get(version_key(&object, head_id)?.as_slice())?)?.value())?;
-        if current.decision_log_id.index < version.decision_log_id.index {
-            return Err(StoreError::CorruptRecord);
-        }
-    }
-    for row in heads.iter()? {
-        let (k, v) = row?;
-        let mut d = Decoder::new(k.value());
-        let object = d.object()?;
-        d.finish()?;
-        let mut d = Decoder::new(v.value());
-        let id = d.uuid()?;
-        d.finish()?;
-        decode_version(required(versions.get(version_key(&object, id)?.as_slice())?)?.value())?;
-    }
-    Ok(())
-}

crates/capture-authority/src/application_tests.rs

@@ -1,531 +0,0 @@
-use super::*;
-use crate::application::{APPLY_RESULTS, EVENTS, OPERATIONS, VERSIONS};
-use crate::tests::CrashBackend;
-use redb::ReadableTableMetadata;
-use std::{
-    io::{BufRead, BufReader, Write},
-    process::{Command, Stdio},
-    sync::mpsc,
-    time::Duration,
-};
-fn identity() -> NodeIdentity {
-    NodeIdentity {
-        cluster_id: [7; 16],
-        node_id: 0,
-    }
-}
-fn binding() -> CaptureBinding {
-    CaptureBinding {
-        revision: 9,
-        binding_id: [8; 16],
-        target_id: [9; 16],
-    }
-}
-fn command() -> DecideCreated {
-    DecideCreated {
-        operation_id: [1; 16],
-        object: ObjectIdentity {
-            bucket_incarnation: [2; 16],
-            key: vec![0, 255, 3],
-        },
-        expected_head: None,
-        prepared: PreparedIdentity {
-            version_id: [3; 16],
-            preparation_id: [4; 16],
-            content_digest: [5; 32],
-            content_length: 123,
-        },
-        binding_revision: 9,
-    }
-}
-fn id(index: u64) -> LogId {
-    LogId {
-        term: 8,
-        leader_node: 0,
-        index,
-    }
-}
-fn append(store: &mut Store, index: u64, c: &DecideCreated) {
-    store
-        .append(&[LogEntry {
-            id: id(index),
-            payload: c.encode().expect("encode command"),
-        }])
-        .expect("append command");
-}
-fn apply(store: &mut Store, index: u64, c: &DecideCreated) -> CreatedResult {
-    append(store, index, c);
-    store.save_committed(id(index)).expect("commit log");
-    store.apply_next_created(id(index)).expect("apply command")
-}
-fn success(c: &DecideCreated, index: u64) -> CreatedResult {
-    CreatedResult::Created {
-        version_id: c.prepared.version_id,
-        event_id: c.operation_id,
-        decision_log_id: id(index),
-    }
-}
-fn complete(store: &Store) {
-    let c = command();
-    assert_eq!(store.capture_binding().expect("binding"), Some(binding()));
-    assert_eq!(
-        store.created_head(&c.object).expect("head"),
-        Some(c.prepared.version_id),
-        "U11 acknowledged head lost"
-    );
-    assert_eq!(
-        store.created_version(&c.object, c.prepared.version_id).expect("version"),
-        Some(CreatedVersion {
-            prepared: c.prepared.clone(),
-            decision_log_id: id(0),
-            binding_revision: 9
-        })
-    );
-    assert_eq!(
-        store.created_event(c.operation_id).expect("event"),
-        Some(CreatedEvent {
-            object: c.object.clone(),
-            prepared: c.prepared.clone(),
-            decision_log_id: id(0),
-            binding: binding()
-        }),
-        "U11 complete event missing"
-    );
-    assert_eq!(
-        store.created_operation(c.operation_id).expect("operation"),
-        Some(CreatedOperation {
-            canonical_command: c.encode().expect("encode"),
-            result: success(&c, 0)
-        })
-    );
-    assert_eq!(
-        store.created_apply_result(0).expect("result"),
-        Some(CreatedApplyResult {
-            log_id: id(0),
-            result: success(&c, 0)
-        })
-    );
-    assert_eq!(store.last_applied().expect("last applied"), Some(id(0)));
-}
-fn fixture() -> (tempfile::TempDir, Store) {
-    let dir = tempfile::tempdir().expect("tempdir");
-    let mut store = Store::initialize(dir.path().join("db"), identity()).expect("initialize");
-    store.pin_capture_binding(binding()).expect("pin");
-    (dir, store)
-}
-fn reopen(path: &std::path::Path) -> Store {
-    let result = Store::open(path, identity());
-    assert!(result.is_ok(), "U11 complete reopen: {:?}", result.as_ref().err());
-    result.expect("asserted successful reopen")
-}
-#[test]
-fn created_codec_is_canonical_and_strict() {
-    let c = command();
-    let b = c.encode().expect("encode");
-    let mut golden = b"CRTD\x01".to_vec();
-    golden.extend([1; 16]);
-    golden.extend([2; 16]);
-    golden.extend(3u32.to_be_bytes());
-    golden.extend([0, 255, 3]);
-    golden.push(0);
-    golden.extend([3; 16]);
-    golden.extend([4; 16]);
-    golden.extend([5; 32]);
-    golden.extend(123u64.to_be_bytes());
-    golden.extend(9u64.to_be_bytes());
-    assert_eq!(b, golden);
-    assert_eq!(DecideCreated::decode(&b).expect("decode"), c);
-    let mut variants = Vec::new();
-    let mut v = c.clone();
-    v.operation_id = [6; 16];
-    variants.push(v);
-    let mut v = c.clone();
-    v.object.bucket_incarnation = [6; 16];
-    variants.push(v);
-    let mut v = c.clone();
-    v.object.key.push(4);
-    variants.push(v);
-    let mut v = c.clone();
-    v.expected_head = Some([6; 16]);
-    variants.push(v);
-    let mut v = c.clone();
-    v.prepared.version_id = [6; 16];
-    variants.push(v);
-    let mut v = c.clone();
-    v.prepared.preparation_id = [6; 16];
-    variants.push(v);
-    let mut v = c.clone();
-    v.prepared.content_digest = [6; 32];
-    variants.push(v);
-    let mut v = c.clone();
-    v.prepared.content_length += 1;
-    variants.push(v);
-    let mut v = c.clone();
-    v.binding_revision += 1;
-    variants.push(v);
-    for v in variants {
-        assert_ne!(v.encode().expect("variant"), b)
-    }
-    for n in 0..b.len() {
-        assert!(DecideCreated::decode(&b[..n]).is_err())
-    }
-    let mut bad = b.clone();
-    bad[4] = 2;
-    assert!(DecideCreated::decode(&bad).is_err());
-    let mut bad = b.clone();
-    bad[44] = 2;
-    assert!(DecideCreated::decode(&bad).is_err());
-    let mut bad = b.clone();
-    bad.push(0);
-    assert!(DecideCreated::decode(&bad).is_err());
-    let mut bad = b;
-    bad[37..41].copy_from_slice(&4097u32.to_be_bytes());
-    assert!(DecideCreated::decode(&bad).is_err());
-    let mut bad = c.clone();
-    bad.operation_id = [0; 16];
-    assert!(bad.encode().is_err());
-    bad = c;
-    bad.object.key = vec![0; MAX_OBJECT_KEY_LENGTH + 1];
-    assert!(bad.encode().is_err());
-}
-#[test]
-fn created_committed_order_and_identity_are_enforced() {
-    let (_dir, mut s) = fixture();
-    let c = command();
-    append(&mut s, 0, &c);
-    append(&mut s, 1, &c);
-    assert!(s.apply_next_created(id(0)).is_err());
-    assert_eq!(s.last_applied().expect("no apply"), None);
-    s.save_committed(id(1)).expect("committed");
-    assert!(matches!(s.apply_next_created(id(1)), Err(StoreError::InvalidApplySequence)));
-    let mut wrong = id(0);
-    wrong.term += 1;
-    assert!(s.apply_next_created(wrong).is_err());
-    wrong = id(0);
-    wrong.leader_node += 1;
-    assert!(s.apply_next_created(wrong).is_err());
-    assert_eq!(s.apply_next_created(id(0)).expect("first"), success(&c, 0));
-    complete(&s);
-    assert_eq!(s.apply_next_created(id(0)).expect("retry"), success(&c, 0));
-    complete(&s);
-    assert!(s.apply_next_created(wrong).is_err());
-    assert_eq!(s.apply_next_created(id(1)).expect("duplicate log"), success(&c, 0));
-    assert_eq!(s.last_applied().expect("last"), Some(id(1)));
-    assert_eq!(
-        s.created_apply_result(1).expect("result"),
-        Some(CreatedApplyResult {
-            log_id: id(1),
-            result: success(&c, 0)
-        })
-    );
-    let tx = s.database.begin_read().expect("read");
-    assert_eq!(tx.open_table(EVENTS).expect("events").len().expect("len"), 1);
-}
-#[test]
-fn created_head_version_event_and_dedup_reopen_together() {
-    let (dir, mut s) = fixture();
-    let c = command();
-    assert_eq!(apply(&mut s, 0, &c), success(&c, 0));
-    drop(s);
-    let mut s = reopen(&dir.path().join("db"));
-    complete(&s);
-    assert_eq!(apply(&mut s, 1, &c), success(&c, 0));
-    let mut changed = c.clone();
-    changed.prepared.content_digest = [6; 32];
-    assert_eq!(
-        apply(&mut s, 2, &changed),
-        CreatedResult::OperationConflict,
-        "U11 full canonical operation conflict"
-    );
-    assert_eq!(
-        s.created_operation(c.operation_id)
-            .expect("original")
-            .expect("operation")
-            .result,
-        success(&c, 0)
-    );
-    let mut stale = c.clone();
-    stale.operation_id = [6; 16];
-    stale.prepared.version_id = [6; 16];
-    assert_eq!(
-        apply(&mut s, 3, &stale),
-        CreatedResult::HeadMismatch {
-            actual_head: Some(c.prepared.version_id)
-        }
-    );
-    let mut reuse = c.clone();
-    reuse.operation_id = [7; 16];
-    reuse.expected_head = Some(c.prepared.version_id);
-    assert_eq!(apply(&mut s, 4, &reuse), CreatedResult::VersionConflict);
-    let mut replacement = stale.clone();
-    replacement.operation_id = [10; 16];
-    replacement.expected_head = Some(c.prepared.version_id);
-    assert_eq!(apply(&mut s, 5, &replacement), success(&replacement, 5));
-    assert_eq!(s.created_head(&c.object).expect("head"), Some(replacement.prepared.version_id));
-    assert_eq!(
-        s.created_version(&c.object, c.prepared.version_id)
-            .expect("old version")
-            .expect("version")
-            .decision_log_id,
-        id(0)
-    );
-    assert!(s.created_event(c.operation_id).expect("old event").is_some());
-    assert!(s.created_event(replacement.operation_id).expect("new event").is_some());
-    assert!(s.created_event(stale.operation_id).expect("rejection event").is_none());
-    assert_eq!(
-        apply(&mut s, 6, &stale),
-        CreatedResult::HeadMismatch {
-            actual_head: Some(c.prepared.version_id)
-        }
-    );
-    drop(s);
-    let s = reopen(&dir.path().join("db"));
-    assert_eq!(s.last_applied().expect("last"), Some(id(6)));
-    let tx = s.database.begin_read().expect("read");
-    assert_eq!(tx.open_table(EVENTS).expect("events").len().expect("len"), 2);
-}
-#[test]
-fn created_binding_is_pinned_and_not_caller_selected() {
-    let (dir, mut s) = fixture();
-    let mut bad = command();
-    bad.binding_revision = 10;
-    assert_eq!(apply(&mut s, 0, &bad), CreatedResult::BindingMismatch);
-    assert!(s.created_head(&bad.object).expect("head").is_none());
-    assert!(s.created_event(bad.operation_id).expect("event").is_none());
-    assert_eq!(
-        s.created_operation(bad.operation_id)
-            .expect("operation")
-            .expect("saved")
-            .result,
-        CreatedResult::BindingMismatch
-    );
-    s.pin_capture_binding(binding()).expect("idempotent pin");
-    let mut changed = binding();
-    changed.target_id = [10; 16];
-    assert!(matches!(s.pin_capture_binding(changed), Err(StoreError::BindingConflict)));
-    drop(s);
-    let s = reopen(&dir.path().join("db"));
-    assert_eq!(s.capture_binding().expect("binding"), Some(binding()));
-}
-#[test]
-fn created_schema_compatibility_and_corruption_are_checked() {
-    let dir = tempfile::tempdir().expect("dir");
-    let path = dir.path().join("db");
-    let s = Store::initialize(&path, identity()).expect("schema1");
-    assert_eq!(s.capture_binding().expect("unmigrated"), None);
-    drop(s);
-    let mut s = reopen(&path);
-    assert_eq!(s.last_applied().expect("no apply"), None);
-    append(&mut s, 0, &command());
-    s.save_committed(id(0)).expect("schema2");
-    drop(s);
-    let mut s = reopen(&path);
-    assert_eq!(s.capture_binding().expect("schema2 unchanged"), None);
-    s.pin_capture_binding(binding()).expect("explicit upgrade");
-    drop(s);
-    let mut s = reopen(&path);
-    assert_eq!(s.capture_binding().expect("schema3"), Some(binding()));
-    s.apply_next_created(id(0)).expect("apply");
-    drop(s);
-    let db = Database::open(&path).expect("raw db");
-    let tx = db.begin_write().expect("write corruption");
-    tx.open_table(EVENTS)
-        .expect("events")
-        .remove(command().operation_id.as_slice())
-        .expect("remove reference");
-    tx.commit().expect("commit corruption");
-    drop(db);
-    let result = Store::open(&path, identity());
-    assert!(
-        matches!(result, Err(StoreError::CorruptRecord)),
-        "U11 missing event rejected before expect"
-    );
-    let (dir, mut s) = fixture();
-    assert_eq!(s.read_committed().expect("empty committed"), None);
-    append(&mut s, 0, &command());
-    s.save_committed(id(0)).expect("committed schema3");
-    assert_eq!(s.capture_binding().expect("no downgrade"), Some(binding()));
-    drop(s);
-    let path = dir.path().join("db");
-    let db = Database::open(&path).expect("raw db");
-    let tx = db.begin_write().expect("write");
-    let mut marker = vec![1];
-    marker.extend(encode_id(id(0)));
-    tx.open_table(META)
-        .expect("meta")
-        .insert("last_applied", marker.as_slice())
-        .expect("corrupt applied");
-    tx.commit().expect("commit");
-    drop(db);
-    assert!(
-        matches!(Store::open(&path, identity()), Err(StoreError::CorruptRecord)),
-        "U11 last applied missing result rejected"
-    );
-    let (dir, s) = fixture();
-    drop(s);
-    let path = dir.path().join("db");
-    let db = Database::open(&path).expect("raw pinned db");
-    let tx = db.begin_write().expect("missing binding write");
-    tx.open_table(META)
-        .expect("meta")
-        .remove("capture_binding")
-        .expect("remove binding");
-    tx.commit().expect("binding corruption commit");
-    drop(db);
-    assert!(
-        matches!(Store::open(&path, identity()), Err(StoreError::CorruptRecord)),
-        "U11 missing pinned binding rejected"
-    );
-
-    let (dir, mut s) = fixture();
-    let c = command();
-    assert_eq!(apply(&mut s, 0, &c), success(&c, 0));
-    assert_eq!(apply(&mut s, 1, &c), success(&c, 0));
-    drop(s);
-    let path = dir.path().join("db");
-    let db = Database::open(&path).expect("raw duplicate db");
-    let tx = db.begin_write().expect("retarget decision write");
-    let mut key = created::object_bytes(&c.object).expect("object key");
-    key.extend(c.prepared.version_id);
-    let mut version = Vec::new();
-    created::put_prepared(&mut version, &c.prepared);
-    version.extend(encode_id(id(1)));
-    version.extend(binding().revision.to_be_bytes());
-    tx.open_table(VERSIONS)
-        .expect("versions")
-        .insert(key.as_slice(), version.as_slice())
-        .expect("retarget version");
-    let mut event = created::object_bytes(&c.object).expect("event object");
-    created::put_prepared(&mut event, &c.prepared);
-    event.extend(encode_id(id(1)));
-    event.extend(created::binding_bytes(&binding()).expect("binding bytes"));
-    tx.open_table(EVENTS)
-        .expect("events")
-        .insert(c.operation_id.as_slice(), event.as_slice())
-        .expect("retarget event");
-    let canonical = c.encode().expect("canonical");
-    let mut operation = u64::try_from(canonical.len()).expect("length").to_be_bytes().to_vec();
-    operation.extend(canonical);
-    operation.extend(created::result_bytes(&success(&c, 1)));
-    tx.open_table(OPERATIONS)
-        .expect("operations")
-        .insert(c.operation_id.as_slice(), operation.as_slice())
-        .expect("retarget original response");
-    for index in 0..=1 {
-        let mut result = encode_id(id(index));
-        result.extend(created::result_bytes(&success(&c, 1)));
-        tx.open_table(APPLY_RESULTS)
-            .expect("results")
-            .insert(index, result.as_slice())
-            .expect("retarget duplicate response");
-    }
-    tx.commit().expect("coherent retarget commit");
-    drop(db);
-    assert!(
-        matches!(Store::open(&path, identity()), Err(StoreError::CorruptRecord)),
-        "U11 original decision cannot retarget duplicate log"
-    );
-}
-#[test]
-fn created_process_child() {
-    if let Some(path) = std::env::var_os("U11_CHILD_DB") {
-        let mut s = Store::initialize(path, identity()).expect("child init");
-        s.pin_capture_binding(binding()).expect("child pin");
-        assert_eq!(apply(&mut s, 0, &command()), success(&command(), 0));
-        complete(&s);
-        println!("U11_ACK");
-        std::io::stdout().flush().expect("flush ACK");
-        loop {
-            std::thread::park()
-        }
-    }
-    let (_dir, mut s) = fixture();
-    assert_eq!(apply(&mut s, 0, &command()), success(&command(), 0));
-    complete(&s);
-}
-#[test]
-fn created_acknowledged_apply_survives_child_kill() {
-    let dir = tempfile::tempdir().expect("dir");
-    let path = dir.path().join("db");
-    let mut child = Command::new(std::env::current_exe().expect("test executable"))
-        .args(["--exact", "application_tests::created_process_child", "--nocapture"])
-        .env("U11_CHILD_DB", &path)
-        .stdout(Stdio::piped())
-        .stderr(Stdio::inherit())
-        .spawn()
-        .expect("spawn");
-    let stdout = child.stdout.take().expect("stdout");
-    let (tx, rx) = mpsc::channel();
-    let reader = std::thread::spawn(move || {
-        for line in BufReader::new(stdout).lines() {
-            let line = line.expect("child line");
-            if line.ends_with("U11_ACK") {
-                let _ = tx.send(());
-                break;
-            }
-        }
-    });
-    let acknowledged = rx.recv_timeout(Duration::from_secs(20));
-    child.kill().expect("kill child");
-    child.wait().expect("wait child");
-    reader.join().expect("reader cleanup");
-    assert!(acknowledged.is_ok(), "U11 bounded child ACK missing");
-    complete(&reopen(&path));
-}
-#[test]
-fn created_power_cut_before_and_after_apply_flush() {
-    let (dir, mut s) = fixture();
-    append(&mut s, 0, &command());
-    s.save_committed(id(0)).expect("committed");
-    drop(s);
-    let path = dir.path().join("db");
-    let backend = CrashBackend::new(&path);
-    let mut s = Store::from_database(backend.database(), identity()).expect("backend");
-    assert_eq!(s.apply_next_created(id(0)).expect("apply ACK"), success(&command(), 0));
-    backend.cut();
-    drop(s);
-    complete(&reopen(&path));
-    for sync in 1..=2 {
-        for after in [false, true] {
-            let (dir, mut s) = fixture();
-            append(&mut s, 0, &command());
-            s.save_committed(id(0)).expect("committed");
-            drop(s);
-            let path = dir.path().join("db");
-            let backend = CrashBackend::new(&path);
-            let mut s = Store::from_database(backend.database(), identity()).expect("backend");
-            let (entered, resume) = backend.pause_sync(sync, after);
-            let writer = std::thread::spawn(move || {
-                let result = s.apply_next_created(id(0));
-                (s, result)
-            });
-            entered.wait();
-            backend.cut();
-            resume.wait();
-            let (s, result) = writer.join().expect("writer");
-            drop(s);
-            let reopened = Store::open(&path, identity());
-            assert!(
-                reopened.is_ok(),
-                "U11 apply cut must reopen whole old/new state sync={sync} after={after}: {:?}",
-                reopened.as_ref().err()
-            );
-            let s = reopened.expect("asserted reopen");
-            if s.last_applied().expect("last").is_some() {
-                complete(&s)
-            } else {
-                assert!(result.is_err());
-                assert_eq!(s.created_head(&command().object).expect("old head"), None);
-                assert_eq!(
-                    s.created_version(&command().object, command().prepared.version_id)
-                        .expect("old version"),
-                    None
-                );
-                assert_eq!(s.created_event(command().operation_id).expect("old event"), None);
-                assert_eq!(s.created_operation(command().operation_id).expect("old op"), None);
-                assert_eq!(s.created_apply_result(0).expect("old result"), None)
-            }
-            eprintln!("U11 cut boundary sync={sync} after={after}");
-        }
-    }
-}

crates/capture-authority/src/committed.rs

@@ -1,63 +0,0 @@
-use crate::{LOGS, LogId, META, Store, StoreError, decode_id, decode_log, encode_id, number};
-use redb::{ReadableDatabase, ReadableTable};
-
-pub(crate) fn validate_committed(
-    meta: &impl ReadableTable<&'static str, &'static [u8]>,
-    logs: &impl ReadableTable<u64, &'static [u8]>,
-) -> Result<Option<LogId>, StoreError> {
-    let schema = number(meta.get("schema")?.ok_or(StoreError::CorruptRecord)?.value())?;
-    let marker = meta.get("committed")?;
-    match (schema, marker) {
-        (1 | 3 | 4, None) => {
-            if matches!(schema, 3 | 4) && meta.get("last_applied")?.is_none_or(|v| v.value() != [0]) {
-                return Err(StoreError::CorruptRecord);
-            }
-            Ok(None)
-        }
-        (2 | 3 | 4, Some(marker)) => {
-            let id = decode_id(marker.value())?;
-            let record = logs.get(id.index)?.ok_or(StoreError::CorruptRecord)?;
-            if decode_log(record.value())?.id != id {
-                return Err(StoreError::CorruptRecord);
-            }
-            Ok(Some(id))
-        }
-        (1 | 2, _) => Err(StoreError::CorruptRecord),
-        _ => Err(StoreError::UnsupportedSchema),
-    }
-}
-
-impl Store {
-    /// Read the exact committed boundary, validating it against the stored log.
-    pub fn read_committed(&self) -> Result<Option<LogId>, StoreError> {
-        let tx = self.database.begin_read()?;
-        validate_committed(&tx.open_table(META)?, &tx.open_table(LOGS)?)
-    }
-
-    /// Persist a monotonic committed boundary supplied by the authority owner.
-    pub fn save_committed(&mut self, id: LogId) -> Result<(), StoreError> {
-        let tx = self.write_transaction()?;
-        {
-            let mut meta = tx.open_table(META)?;
-            let logs = tx.open_table(LOGS)?;
-            let previous = validate_committed(&meta, &logs)?;
-            let record = logs.get(id.index)?.ok_or(StoreError::InvalidCommittedLog)?;
-            if decode_log(record.value())?.id != id {
-                return Err(StoreError::InvalidCommittedLog);
-            }
-            if let Some(previous) = previous {
-                if id.index < previous.index {
-                    return Err(StoreError::CommittedRegression);
-                }
-                if id == previous {
-                    return Ok(());
-                }
-            }
-            meta.insert("committed", encode_id(id).as_slice())?;
-            if previous.is_none() && !matches!(number(meta.get("schema")?.ok_or(StoreError::CorruptRecord)?.value())?, 3 | 4) {
-                meta.insert("schema", 2u64.to_be_bytes().as_slice())?;
-            }
-        }
-        self.commit(tx)
-    }
-}

crates/capture-authority/src/created.rs

@@ -1,242 +0,0 @@
-use crate::{LogId, StoreError, decode_id, encode_id};
-
-pub const MAX_OBJECT_KEY_LENGTH: usize = 4096;
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct ObjectIdentity {
-    pub bucket_incarnation: [u8; 16],
-    pub key: Vec<u8>,
-}
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct PreparedIdentity {
-    pub version_id: [u8; 16],
-    pub preparation_id: [u8; 16],
-    pub content_digest: [u8; 32],
-    pub content_length: u64,
-}
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct DecideCreated {
-    pub operation_id: [u8; 16],
-    pub object: ObjectIdentity,
-    pub expected_head: Option<[u8; 16]>,
-    pub prepared: PreparedIdentity,
-    pub binding_revision: u64,
-}
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct CaptureBinding {
-    pub revision: u64,
-    pub binding_id: [u8; 16],
-    pub target_id: [u8; 16],
-}
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub enum CreatedResult {
-    Created {
-        version_id: [u8; 16],
-        event_id: [u8; 16],
-        decision_log_id: LogId,
-    },
-    HeadMismatch {
-        actual_head: Option<[u8; 16]>,
-    },
-    OperationConflict,
-    BindingMismatch,
-    VersionConflict,
-}
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct CreatedVersion {
-    pub prepared: PreparedIdentity,
-    pub decision_log_id: LogId,
-    pub binding_revision: u64,
-}
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct CreatedEvent {
-    pub object: ObjectIdentity,
-    pub prepared: PreparedIdentity,
-    pub decision_log_id: LogId,
-    pub binding: CaptureBinding,
-}
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct CreatedOperation {
-    pub canonical_command: Vec<u8>,
-    pub result: CreatedResult,
-}
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct CreatedApplyResult {
-    pub log_id: LogId,
-    pub result: CreatedResult,
-}
-
-pub(crate) struct Decoder<'a> {
-    bytes: &'a [u8],
-}
-impl<'a> Decoder<'a> {
-    pub(crate) fn new(bytes: &'a [u8]) -> Self {
-        Self { bytes }
-    }
-    pub(crate) fn take<const N: usize>(&mut self) -> Result<[u8; N], StoreError> {
-        let a = self
-            .bytes
-            .get(..N)
-            .ok_or(StoreError::CorruptRecord)?
-            .try_into()
-            .map_err(|_| StoreError::CorruptRecord)?;
-        self.bytes = &self.bytes[N..];
-        Ok(a)
-    }
-    pub(crate) fn u64(&mut self) -> Result<u64, StoreError> {
-        Ok(u64::from_be_bytes(self.take()?))
-    }
-    pub(crate) fn uuid(&mut self) -> Result<[u8; 16], StoreError> {
-        let id = self.take()?;
-        if id == [0; 16] {
-            return Err(StoreError::CorruptRecord);
-        }
-        Ok(id)
-    }
-    pub(crate) fn option(&mut self) -> Result<Option<[u8; 16]>, StoreError> {
-        match self.take::<1>()?[0] {
-            0 => Ok(None),
-            1 => Ok(Some(self.uuid()?)),
-            _ => Err(StoreError::CorruptRecord),
-        }
-    }
-    pub(crate) fn object(&mut self) -> Result<ObjectIdentity, StoreError> {
-        let bucket_incarnation = self.uuid()?;
-        let len = usize::try_from(u32::from_be_bytes(self.take()?)).map_err(|_| StoreError::CorruptRecord)?;
-        if len > MAX_OBJECT_KEY_LENGTH {
-            return Err(StoreError::CorruptRecord);
-        }
-        let key = self.bytes.get(..len).ok_or(StoreError::CorruptRecord)?.to_vec();
-        self.bytes = &self.bytes[len..];
-        Ok(ObjectIdentity { bucket_incarnation, key })
-    }
-    pub(crate) fn prepared(&mut self) -> Result<PreparedIdentity, StoreError> {
-        Ok(PreparedIdentity {
-            version_id: self.uuid()?,
-            preparation_id: self.uuid()?,
-            content_digest: self.take()?,
-            content_length: self.u64()?,
-        })
-    }
-    pub(crate) fn id(&mut self) -> Result<LogId, StoreError> {
-        decode_id(&self.take::<24>()?)
-    }
-    pub(crate) fn binding(&mut self) -> Result<CaptureBinding, StoreError> {
-        Ok(CaptureBinding {
-            revision: self.u64()?,
-            binding_id: self.uuid()?,
-            target_id: self.uuid()?,
-        })
-    }
-    pub(crate) fn finish(self) -> Result<(), StoreError> {
-        if self.bytes.is_empty() {
-            Ok(())
-        } else {
-            Err(StoreError::CorruptRecord)
-        }
-    }
-    pub(crate) fn result(&mut self) -> Result<CreatedResult, StoreError> {
-        match self.take::<1>()?[0] {
-            0 => Ok(CreatedResult::Created {
-                version_id: self.uuid()?,
-                event_id: self.uuid()?,
-                decision_log_id: self.id()?,
-            }),
-            1 => Ok(CreatedResult::HeadMismatch {
-                actual_head: self.option()?,
-            }),
-            2 => Ok(CreatedResult::OperationConflict),
-            3 => Ok(CreatedResult::BindingMismatch),
-            4 => Ok(CreatedResult::VersionConflict),
-            _ => Err(StoreError::CorruptRecord),
-        }
-    }
-}
-pub(crate) fn put_option(b: &mut Vec<u8>, id: Option<[u8; 16]>) {
-    b.push(u8::from(id.is_some()));
-    if let Some(id) = id {
-        b.extend(id)
-    }
-}
-pub(crate) fn object_bytes(o: &ObjectIdentity) -> Result<Vec<u8>, StoreError> {
-    if o.bucket_incarnation == [0; 16] || o.key.len() > MAX_OBJECT_KEY_LENGTH {
-        return Err(StoreError::CorruptRecord);
-    }
-    let capacity = 20usize.checked_add(o.key.len()).ok_or(StoreError::CorruptRecord)?;
-    let mut b = Vec::with_capacity(capacity);
-    b.extend(o.bucket_incarnation);
-    b.extend(
-        u32::try_from(o.key.len())
-            .map_err(|_| StoreError::CorruptRecord)?
-            .to_be_bytes(),
-    );
-    b.extend(&o.key);
-    Ok(b)
-}
-pub(crate) fn put_prepared(b: &mut Vec<u8>, p: &PreparedIdentity) {
-    b.extend(p.version_id);
-    b.extend(p.preparation_id);
-    b.extend(p.content_digest);
-    b.extend(p.content_length.to_be_bytes())
-}
-pub(crate) fn binding_bytes(v: &CaptureBinding) -> Result<Vec<u8>, StoreError> {
-    let mut b = v.revision.to_be_bytes().to_vec();
-    b.extend(v.binding_id);
-    b.extend(v.target_id);
-    let mut d = Decoder::new(&b);
-    d.binding()?;
-    d.finish()?;
-    Ok(b)
-}
-pub(crate) fn result_bytes(r: &CreatedResult) -> Vec<u8> {
-    let mut b = Vec::new();
-    match r {
-        CreatedResult::Created {
-            version_id,
-            event_id,
-            decision_log_id,
-        } => {
-            b.push(0);
-            b.extend(version_id);
-            b.extend(event_id);
-            b.extend(encode_id(*decision_log_id))
-        }
-        CreatedResult::HeadMismatch { actual_head } => {
-            b.push(1);
-            put_option(&mut b, *actual_head)
-        }
-        CreatedResult::OperationConflict => b.push(2),
-        CreatedResult::BindingMismatch => b.push(3),
-        CreatedResult::VersionConflict => b.push(4),
-    }
-    b
-}
-impl DecideCreated {
-    pub fn encode(&self) -> Result<Vec<u8>, StoreError> {
-        let object = object_bytes(&self.object)?;
-        let capacity = object.len().checked_add(134).ok_or(StoreError::CorruptRecord)?;
-        let mut b = Vec::with_capacity(capacity);
-        b.extend(b"CRTD\x01");
-        b.extend(self.operation_id);
-        b.extend(object);
-        put_option(&mut b, self.expected_head);
-        put_prepared(&mut b, &self.prepared);
-        b.extend(self.binding_revision.to_be_bytes());
-        Self::decode(&b)?;
-        Ok(b)
-    }
-    pub fn decode(bytes: &[u8]) -> Result<Self, StoreError> {
-        let mut d = Decoder::new(bytes);
-        if d.take::<5>()? != *b"CRTD\x01" {
-            return Err(StoreError::CorruptRecord);
-        }
-        let c = Self {
-            operation_id: d.uuid()?,
-            object: d.object()?,
-            expected_head: d.option()?,
-            prepared: d.prepared()?,
-            binding_revision: d.u64()?,
-        };
-        d.finish()?;
-        Ok(c)
-    }
-}

crates/capture-authority/src/node.rs

@@ -1,371 +0,0 @@
-//! Validated capture peer metadata and its independent CNOD v1 codec.
-use serde::{Deserialize, Deserializer, Serialize, Serializer, ser::SerializeStruct};
-
-#[derive(Clone, Debug, Default, PartialEq, Eq)]
-pub struct CaptureNode {
-    rpc_uri: String,
-}
-
-#[derive(Debug, thiserror::Error)]
-pub enum CaptureNodeError {
-    #[error("invalid RPC URI")]
-    InvalidRpcUri,
-    #[error("invalid RPC URI")]
-    InvalidRpcUriParse(#[source] url::ParseError),
-    #[error("RPC URI too long")]
-    RpcUriTooLong,
-    #[error("invalid node magic")]
-    InvalidNodeMagic,
-    #[error("unsupported node version")]
-    UnsupportedNodeVersion,
-    #[error("invalid node length")]
-    InvalidNodeLength,
-    #[error("truncated node record")]
-    TruncatedNodeRecord,
-    #[error("invalid node UTF-8")]
-    InvalidNodeUtf8(#[source] std::str::Utf8Error),
-    #[error("trailing node bytes")]
-    TrailingNodeBytes,
-}
-impl CaptureNodeError {
-    pub fn code(&self) -> &'static str {
-        match self {
-            Self::InvalidRpcUri | Self::InvalidRpcUriParse(_) => "invalid_rpc_uri",
-            Self::RpcUriTooLong => "rpc_uri_too_long",
-            Self::InvalidNodeMagic => "invalid_node_magic",
-            Self::UnsupportedNodeVersion => "unsupported_node_version",
-            Self::InvalidNodeLength => "invalid_node_length",
-            Self::TruncatedNodeRecord => "truncated_node_record",
-            Self::InvalidNodeUtf8(_) => "invalid_node_utf8",
-            Self::TrailingNodeBytes => "trailing_node_bytes",
-        }
-    }
-}
-impl CaptureNode {
-    pub fn new(rpc_uri: String) -> Result<Self, CaptureNodeError> {
-        let node = Self { rpc_uri };
-        node.validate()?;
-        Ok(node)
-    }
-
-    pub fn rpc_uri(&self) -> &str {
-        &self.rpc_uri
-    }
-
-    pub fn validate(&self) -> Result<(), CaptureNodeError> {
-        let uri = &self.rpc_uri;
-        if uri.len() > 4096 {
-            return Err(CaptureNodeError::RpcUriTooLong);
-        }
-        if uri.is_empty() || uri.chars().any(|c| c.is_whitespace() || c.is_control() || c == '\\') {
-            return Err(CaptureNodeError::InvalidRpcUri);
-        }
-        let rest = uri
-            .strip_prefix("http://")
-            .or_else(|| uri.strip_prefix("https://"))
-            .ok_or(CaptureNodeError::InvalidRpcUri)?;
-        if uri.contains('?') || uri.contains('#') {
-            return Err(CaptureNodeError::InvalidRpcUri);
-        }
-        let (authority, path) = rest.split_once('/').map_or((rest, ""), |(a, p)| (a, p));
-        // An empty suffix after the first slash is the only allowed path.
-        if !path.is_empty() || authority.contains('@') {
-            return Err(CaptureNodeError::InvalidRpcUri);
-        }
-        let (host, port) = if authority.starts_with('[') {
-            let end = authority.find(']').ok_or(CaptureNodeError::InvalidRpcUri)?;
-            let port = authority
-                .get(end + 1..)
-                .and_then(|suffix| suffix.strip_prefix(':'))
-                .ok_or(CaptureNodeError::InvalidRpcUri)?;
-            (&authority[..=end], port)
-        } else {
-            authority.split_once(':').ok_or(CaptureNodeError::InvalidRpcUri)?
-        };
-        if host.is_empty() || port.is_empty() || !port.bytes().all(|b| b.is_ascii_digit()) {
-            return Err(CaptureNodeError::InvalidRpcUri);
-        }
-        let port = port.parse::<u16>().map_err(|_| CaptureNodeError::InvalidRpcUri)?;
-        if port == 0 {
-            return Err(CaptureNodeError::InvalidRpcUri);
-        }
-        let parsed_url = url::Url::parse(uri).map_err(CaptureNodeError::InvalidRpcUriParse)?;
-        if parsed_url.host().is_none() {
-            return Err(CaptureNodeError::InvalidRpcUri);
-        }
-        Ok(())
-    }
-
-    pub fn encode(&self) -> Result<Vec<u8>, CaptureNodeError> {
-        self.validate()?;
-        let length = u32::try_from(self.rpc_uri.len()).map_err(|_| CaptureNodeError::InvalidNodeLength)?;
-        let mut bytes = b"CNOD\x01".to_vec();
-        bytes.extend_from_slice(&length.to_be_bytes());
-        bytes.extend_from_slice(self.rpc_uri.as_bytes());
-        Ok(bytes)
-    }
-
-    pub fn decode(bytes: &[u8]) -> Result<Self, CaptureNodeError> {
-        if bytes.len() < 9 {
-            return Err(CaptureNodeError::TruncatedNodeRecord);
-        }
-        if &bytes[..4] != b"CNOD" {
-            return Err(CaptureNodeError::InvalidNodeMagic);
-        }
-        if bytes[4] != 1 {
-            return Err(CaptureNodeError::UnsupportedNodeVersion);
-        }
-        let length = u32::from_be_bytes(bytes[5..9].try_into().map_err(|_| CaptureNodeError::InvalidNodeLength)?);
-        if !(1..=4096).contains(&length) {
-            return Err(CaptureNodeError::InvalidNodeLength);
-        }
-        let length = usize::try_from(length).map_err(|_| CaptureNodeError::InvalidNodeLength)?;
-        let end = 9_usize.checked_add(length).ok_or(CaptureNodeError::InvalidNodeLength)?;
-        if bytes.len() < end {
-            return Err(CaptureNodeError::TruncatedNodeRecord);
-        }
-        if bytes.len() != end {
-            return Err(CaptureNodeError::TrailingNodeBytes);
-        }
-        let uri = std::str::from_utf8(&bytes[9..end]).map_err(CaptureNodeError::InvalidNodeUtf8)?;
-        Self::new(uri.to_owned())
-    }
-}
-impl Serialize for CaptureNode {
-    fn serialize<S: Serializer>(&self, serializer: S) -> Result<S::Ok, S::Error> {
-        self.validate().map_err(serde::ser::Error::custom)?;
-        let mut state = serializer.serialize_struct("CaptureNode", 1)?;
-        state.serialize_field("rpc_uri", &self.rpc_uri)?;
-        state.end()
-    }
-}
-impl<'de> Deserialize<'de> for CaptureNode {
-    fn deserialize<D: Deserializer<'de>>(deserializer: D) -> Result<Self, D::Error> {
-        #[derive(Deserialize)]
-        #[serde(deny_unknown_fields)]
-        struct Fields {
-            rpc_uri: String,
-        }
-        let fields = Fields::deserialize(deserializer).map_err(|_| serde::de::Error::custom("invalid capture node fields"))?;
-        Self::new(fields.rpc_uri).map_err(serde::de::Error::custom)
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-    use std::error::Error;
-
-    fn record(payload: &[u8]) -> Vec<u8> {
-        let mut bytes = b"CNOD\x01".to_vec();
-        bytes.extend_from_slice(&u32::try_from(payload.len()).expect("test payload length").to_be_bytes());
-        bytes.extend_from_slice(payload);
-        bytes
-    }
-    fn reject_uri(uri: &str) {
-        let result = CaptureNode::new(uri.to_owned());
-        assert!(result.is_err(), "URI should reject: {uri:?}");
-        assert_eq!(result.expect_err("rejected URI").code(), "invalid_rpc_uri");
-    }
-    fn reject_record(bytes: &[u8], code: &str) {
-        let result = CaptureNode::decode(bytes);
-        assert!(result.is_err(), "record should reject with {code}");
-        assert_eq!(result.expect_err("rejected record").code(), code);
-    }
-    #[test]
-    fn actual_openraft_node_bound() {
-        fn require_node<T: openraft::Node>() {}
-        require_node::<CaptureNode>();
-        assert_eq!(CaptureNode::default().rpc_uri(), "");
-    }
-    #[test]
-    fn default_and_empty_rejected_at_every_boundary() {
-        let node = CaptureNode::default();
-        assert_eq!(node.rpc_uri(), "");
-        reject_uri("");
-        assert_eq!(node.validate().expect_err("empty validation").code(), "invalid_rpc_uri");
-        assert!(serde_json::to_string(&node).is_err());
-        assert!(serde_json::from_str::<CaptureNode>(r#"{"rpc_uri":""}"#).is_err());
-        assert_eq!(node.encode().expect_err("empty encoding").code(), "invalid_rpc_uri");
-        reject_record(&record(b""), "invalid_node_length");
-    }
-    #[test]
-    fn accepted_uris_preserve_exact_bytes() {
-        for uri in [
-            "http://node:80",
-            "https://node:443",
-            "http://node:1",
-            "https://node:65535/",
-            "http://127.0.0.1:9000",
-            "https://[::1]:443/",
-            "http://[2001:db8::1]:1",
-            "http://NoDe:00080/",
-        ] {
-            let result = CaptureNode::new(uri.to_owned());
-            assert!(result.is_ok(), "valid URI construction: {uri}");
-            let node = result.expect("valid URI");
-            assert_eq!(node.rpc_uri(), uri, "preserve explicit default port and raw URI");
-            assert_eq!(node.encode().expect("encode URI"), record(uri.as_bytes()));
-            assert_eq!(CaptureNode::decode(&record(uri.as_bytes())).expect("decode URI"), node);
-        }
-    }
-    #[test]
-    fn golden_cnod_bytes() {
-        let node = CaptureNode {
-            rpc_uri: "http://node:80".to_owned(),
-        };
-        let golden = b"CNOD\x01\x00\x00\x00\x0ehttp://node:80";
-        assert_eq!(node.encode().ok().as_deref(), Some(golden.as_slice()), "golden CNOD bytes");
-    }
-    #[test]
-    fn valid_cnod_decode() {
-        let golden = b"CNOD\x01\x00\x00\x00\x0ehttp://node:80";
-        let decoded = CaptureNode::decode(golden);
-        assert!(decoded.is_ok(), "golden decode succeeds");
-        assert_eq!(decoded.expect("golden decoded node").rpc_uri(), "http://node:80");
-    }
-    #[test]
-    fn serde_shape_roundtrip_and_strict_fields() {
-        let node = CaptureNode::new("https://node:443/".to_owned()).expect("serde node");
-        let json = r#"{"rpc_uri":"https://node:443/"}"#;
-        assert_eq!(serde_json::to_string(&node).expect("serialize node"), json);
-        assert_eq!(serde_json::from_str::<CaptureNode>(json).expect("deserialize node"), node);
-        for json in [
-            r#"{}"#,
-            r#"{"rpc_uri":1}"#,
-            r#"{"rpc_uri":null}"#,
-            r#"{"rpc_uri":"http://node:1","extra":0}"#,
-            r#"{"rpc_uri":"http://node:1","rpc_uri":"http://node:2"}"#,
-        ] {
-            assert!(serde_json::from_str::<CaptureNode>(json).is_err(), "strict fields: {json}");
-        }
-    }
-    #[test]
-    fn credentials_and_raw_uri_shapes_reject() {
-        for uri in [
-            "http://user@node:1",
-            "http://user:password@node:1",
-            "http://@node:1",
-            "HTTP://node:1",
-            "Https://node:1",
-            "http://node",
-            "http://node:",
-            "http://node:0",
-            "http://node:-1",
-            "http://node:+1",
-            "http://node:65536",
-            "http://node:999999999999999999999",
-            "http://[::1:1",
-            "http://[::1]extra:1",
-            "http://[invalid]:1",
-            "http://::1:1",
-            "http://:1",
-            "/disk/path",
-            "file:///disk/path",
-            "http://node:1/./",
-            "http://node:1/../",
-            "http://node:1/%2e/",
-            "http://node:1/%2E%2e/",
-            "http://node:1/path",
-            "http://node:1\\",
-            " http://node:1",
-            "http://node:1\n",
-            "http://no\tde:1",
-            "http://node:1\u{00a0}",
-            "http://node:1\u{0085}",
-            "http://node:1\0",
-        ] {
-            reject_uri(uri);
-        }
-    }
-    #[test]
-    fn query_and_fragment_markers_reject() {
-        for uri in [
-            "http://node:1?",
-            "http://node:1?x=y",
-            "http://node:1/?",
-            "http://node:1#",
-            "http://node:1#fragment",
-            "http://node:1/#",
-        ] {
-            reject_uri(uri);
-        }
-    }
-    #[test]
-    fn utf8_byte_length_boundaries() {
-        for length in [4095, 4096] {
-            let prefix = "http://node:";
-            let uri = format!("{prefix}{}1", "0".repeat(length - prefix.len() - 1));
-            assert_eq!(uri.len(), length);
-            assert!(url::Url::parse(&uri).is_ok(), "boundary fixture accepted by actual URL parser");
-            let node = CaptureNode::new(uri.clone()).expect("boundary node");
-            assert_eq!(node.rpc_uri(), uri);
-            assert_eq!(
-                CaptureNode::decode(&node.encode().expect("boundary encoding")).expect("boundary decoding"),
-                node
-            );
-        }
-        let uri = format!("http://node:{}1", "0".repeat(4097 - "http://node:".len() - 1));
-        assert_eq!(CaptureNode::new(uri).expect_err("oversize URI").code(), "rpc_uri_too_long");
-        let unicode = format!("http://{}:1", "é".repeat(2044));
-        assert!(unicode.chars().count() < 4096);
-        assert!(unicode.len() > 4096);
-        assert_eq!(CaptureNode::new(unicode).expect_err("UTF-8 byte limit").code(), "rpc_uri_too_long");
-    }
-    #[test]
-    fn cnod_headers_and_lengths_reject_exact_codes() {
-        let valid = record(b"http://node:80");
-        for end in 0..9 {
-            reject_record(&valid[..end], "truncated_node_record");
-        }
-        let mut bad = valid.clone();
-        bad[0] = b'X';
-        reject_record(&bad, "invalid_node_magic");
-        bad = valid.clone();
-        bad[4] = 2;
-        reject_record(&bad, "unsupported_node_version");
-        for length in [0_u32, 4097, u32::MAX] {
-            bad = b"CNOD\x01".to_vec();
-            bad.extend_from_slice(&length.to_be_bytes());
-            reject_record(&bad, "invalid_node_length");
-        }
-        for end in 9..valid.len() {
-            reject_record(&valid[..end], "truncated_node_record");
-        }
-        reject_record(&record(&[0xff]), "invalid_node_utf8");
-        reject_record(&record(&[0xc3, 0x28]), "invalid_node_utf8");
-        reject_record(&record(b"invalid URI"), "invalid_rpc_uri");
-    }
-    #[test]
-    fn trailing_bytes_reject() {
-        for tail in [&b"x"[..], &b"xyz"[..]] {
-            let mut bytes = record(b"http://node:80");
-            bytes.extend_from_slice(tail);
-            reject_record(&bytes, "trailing_node_bytes");
-        }
-    }
-    #[test]
-    fn errors_and_sources_redact_input() {
-        let sentinel = "http://SECRET_USER:SECRET_PASSWORD@node:1";
-        let error = CaptureNode::new(sentinel.to_owned()).expect_err("credentials reject");
-        let mut current: &dyn Error = &error;
-        loop {
-            let rendered = format!("{current} {current:?}");
-            assert!(!rendered.contains("SECRET"));
-            match current.source() {
-                Some(source) => current = source,
-                None => break,
-            }
-        }
-        let json = format!(r#"{{"rpc_uri":"{sentinel}"}}"#);
-        let error = serde_json::from_str::<CaptureNode>(&json).expect_err("serde credentials reject");
-        assert!(!format!("{error} {error:?}").contains("SECRET"));
-        let json = format!(r#"{{"{sentinel}":0}}"#);
-        let error = serde_json::from_str::<CaptureNode>(&json).expect_err("unknown fields reject");
-        assert!(!format!("{error} {error:?}").contains("SECRET"));
-        let error = CaptureNode::new("http://[bad]:1".to_owned()).expect_err("parser rejection");
-        assert!(error.source().is_some(), "retain URL parser source");
-        let error = CaptureNode::decode(&record(&[0xff])).expect_err("UTF-8 rejection");
-        assert!(error.source().is_some(), "retain UTF-8 source");
-    }
-}

crates/capture-authority/src/truncate.rs

@@ -1,43 +0,0 @@
-use crate::{LOGS, META, Store, StoreError, decode_id, decode_log, encode_id};
-use redb::ReadableTable;
-
-impl Store {
-    /// Atomically remove every log entry at or after `from`.
-    pub fn truncate_suffix(&mut self, from: u64) -> Result<(), StoreError> {
-        let tx = self.write_transaction()?;
-        {
-            let mut meta = tx.open_table(META)?;
-            let logs = tx.open_table(LOGS)?;
-            if crate::committed::validate_committed(&meta, &logs)?.is_some_and(|id| from <= id.index) {
-                return Err(StoreError::CommittedLogProtected);
-            }
-            drop(logs);
-            let tail = meta.get("tail")?.map(|v| decode_id(v.value())).transpose()?;
-            if tail.is_none_or(|id| from > id.index) {
-                return Ok(());
-            }
-            let mut logs = tx.open_table(LOGS)?;
-            let new_tail = if from == 0 {
-                None
-            } else {
-                let index = from - 1;
-                let record = logs.get(index)?.ok_or(StoreError::InvalidLogSequence)?;
-                let id = decode_log(record.value())?.id;
-                if id.index != index {
-                    return Err(StoreError::CorruptRecord);
-                }
-                Some(id)
-            };
-            logs.retain(|index, _| index < from)?;
-            match new_tail {
-                Some(id) => {
-                    meta.insert("tail", encode_id(id).as_slice())?;
-                }
-                None => {
-                    meta.remove("tail")?;
-                }
-            }
-        }
-        self.commit(tx)
-    }
-}

crates/common/Cargo.toml

@@ -34,6 +34,7 @@ hotpath-alloc = ["hotpath", "hotpath/hotpath-alloc"]
 hotpath-cpu = ["hotpath", "hotpath/hotpath-cpu"]
 
 [dependencies]
+arc-swap.workspace = true
 hotpath.workspace = true
 tokio = { workspace = true, features = ["fs", "rt-multi-thread"] }
 tonic = { workspace = true, features = ["gzip", "deflate"] }

crates/common/src/globals.rs

@@ -12,9 +12,10 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use arc_swap::ArcSwap;
 use std::collections::HashMap;
-use std::sync::LazyLock;
 use std::sync::atomic::{AtomicU64, Ordering};
+use std::sync::{Arc, LazyLock};
 use tokio::sync::RwLock;
 use tonic::transport::Channel;
 
@@ -27,6 +28,42 @@ pub static GLOBAL_ROOT_CERT: LazyLock<RwLock<Option<Vec<u8>>>> = LazyLock::new(|
 pub static GLOBAL_MTLS_IDENTITY: LazyLock<RwLock<Option<MtlsIdentityPem>>> = LazyLock::new(|| RwLock::new(None));
 pub static GLOBAL_OUTBOUND_TLS_GENERATION: LazyLock<AtomicU64> = LazyLock::new(|| AtomicU64::new(0));
 
+#[derive(Clone, Default)]
+pub struct OutboundTlsSnapshot {
+    pub generation: u64,
+    pub root_ca_pem: Option<Vec<u8>>,
+    pub mtls_identity: Option<MtlsIdentityPem>,
+}
+
+static GLOBAL_OUTBOUND_TLS_STATE: LazyLock<ArcSwap<OutboundTlsSnapshot>> =
+    LazyLock::new(|| ArcSwap::from_pointee(OutboundTlsSnapshot::default()));
+
+pub fn get_global_outbound_tls_snapshot() -> Arc<OutboundTlsSnapshot> {
+    GLOBAL_OUTBOUND_TLS_STATE.load_full()
+}
+
+/// Publish identity, trust and generation together; the individual globals are legacy views.
+pub async fn publish_global_outbound_tls_snapshot(snapshot: OutboundTlsSnapshot) -> u64 {
+    GLOBAL_OUTBOUND_TLS_STATE.rcu(|current| {
+        if current.generation != 0
+            && current.root_ca_pem == snapshot.root_ca_pem
+            && current.mtls_identity == snapshot.mtls_identity
+        {
+            return Arc::clone(current);
+        }
+        Arc::new(OutboundTlsSnapshot {
+            generation: snapshot.generation.max(current.generation.saturating_add(1)),
+            root_ca_pem: snapshot.root_ca_pem.clone(),
+            mtls_identity: snapshot.mtls_identity.clone(),
+        })
+    });
+    let accepted = get_global_outbound_tls_snapshot();
+    *GLOBAL_ROOT_CERT.write().await = accepted.root_ca_pem.clone();
+    *GLOBAL_MTLS_IDENTITY.write().await = accepted.mtls_identity.clone();
+    GLOBAL_OUTBOUND_TLS_GENERATION.store(accepted.generation, Ordering::Relaxed);
+    accepted.generation
+}
+
 /// Log level to use when reporting cached gRPC connection eviction.
 #[derive(Clone, Copy, Debug, Eq, PartialEq)]
 pub enum ConnectionEvictionLogLevel {
@@ -85,17 +122,27 @@ pub async fn get_global_addr() -> String {
 /// # Arguments
 /// * `cert` - A vector of bytes representing the PEM-encoded root CA certificate.
 pub async fn set_global_root_cert(cert: Vec<u8>) {
+    GLOBAL_OUTBOUND_TLS_STATE.rcu(|current| {
+        let mut next = (**current).clone();
+        next.root_ca_pem = Some(cert.clone());
+        Arc::new(next)
+    });
     *GLOBAL_ROOT_CERT.write().await = Some(cert);
 }
 
 /// Clear the global root CA certificate for outbound gRPC clients.
 pub async fn clear_global_root_cert() {
+    GLOBAL_OUTBOUND_TLS_STATE.rcu(|current| {
+        let mut next = (**current).clone();
+        next.root_ca_pem = None;
+        Arc::new(next)
+    });
     *GLOBAL_ROOT_CERT.write().await = None;
 }
 
 /// Get the global root CA certificate for outbound gRPC clients.
 pub async fn get_global_root_cert() -> Option<Vec<u8>> {
-    GLOBAL_ROOT_CERT.read().await.clone()
+    get_global_outbound_tls_snapshot().root_ca_pem.clone()
 }
 
 /// Set the global mTLS identity (cert+key PEM) for outbound gRPC clients.
@@ -105,22 +152,32 @@ pub async fn get_global_root_cert() -> Option<Vec<u8>> {
 /// # Arguments
 /// * `identity` - An optional MtlsIdentityPem struct containing the cert and key PEM.
 pub async fn set_global_mtls_identity(identity: Option<MtlsIdentityPem>) {
+    GLOBAL_OUTBOUND_TLS_STATE.rcu(|current| {
+        let mut next = (**current).clone();
+        next.mtls_identity = identity.clone();
+        Arc::new(next)
+    });
     *GLOBAL_MTLS_IDENTITY.write().await = identity;
 }
 
 /// Get the global mTLS identity for outbound gRPC clients.
 pub async fn get_global_mtls_identity() -> Option<MtlsIdentityPem> {
-    GLOBAL_MTLS_IDENTITY.read().await.clone()
+    get_global_outbound_tls_snapshot().mtls_identity.clone()
 }
 
 /// Set the global outbound TLS generation.
 pub fn set_global_outbound_tls_generation(generation: u64) {
+    GLOBAL_OUTBOUND_TLS_STATE.rcu(|current| {
+        let mut next = (**current).clone();
+        next.generation = generation;
+        Arc::new(next)
+    });
     GLOBAL_OUTBOUND_TLS_GENERATION.store(generation, Ordering::Relaxed);
 }
 
 /// Get the global outbound TLS generation.
 pub fn get_global_outbound_tls_generation() -> u64 {
-    GLOBAL_OUTBOUND_TLS_GENERATION.load(Ordering::Relaxed)
+    get_global_outbound_tls_snapshot().generation
 }
 
 /// Evict a stale/dead connection from the global connection cache.
@@ -194,8 +251,77 @@ pub async fn clear_all_connections() {
 ///
 /// When present, gRPC clients will present this identity to servers requesting/requiring mTLS.
 /// When absent, clients proceed with standard server-authenticated TLS.
-#[derive(Clone, Debug)]
+#[derive(Clone, PartialEq, Eq)]
 pub struct MtlsIdentityPem {
     pub cert_pem: Vec<u8>,
     pub key_pem: Vec<u8>,
 }
+
+impl std::fmt::Debug for MtlsIdentityPem {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("MtlsIdentityPem")
+            .field("cert_bytes", &self.cert_pem.len())
+            .field("key_pem", &"[REDACTED]")
+            .finish()
+    }
+}
+
+#[cfg(test)]
+mod tls_snapshot_tests {
+    use super::*;
+
+    #[tokio::test]
+    async fn outbound_snapshot_publication_is_coherent_and_skips_unchanged_material() {
+        let start = get_global_outbound_tls_generation().saturating_add(1);
+        let snapshot = OutboundTlsSnapshot {
+            generation: start,
+            root_ca_pem: Some(vec![1]),
+            mtls_identity: Some(MtlsIdentityPem {
+                cert_pem: vec![1],
+                key_pem: vec![1],
+            }),
+        };
+        let first = publish_global_outbound_tls_snapshot(snapshot.clone()).await;
+        assert_eq!(
+            publish_global_outbound_tls_snapshot(OutboundTlsSnapshot {
+                generation: first.saturating_add(1),
+                ..snapshot
+            })
+            .await,
+            first
+        );
+        let writer = tokio::spawn(async move {
+            for tag in 2_u8..=255 {
+                publish_global_outbound_tls_snapshot(OutboundTlsSnapshot {
+                    generation: start.saturating_add(u64::from(tag)),
+                    root_ca_pem: Some(vec![tag]),
+                    mtls_identity: Some(MtlsIdentityPem {
+                        cert_pem: vec![tag],
+                        key_pem: vec![tag],
+                    }),
+                })
+                .await;
+                tokio::task::yield_now().await;
+            }
+        });
+        while !writer.is_finished() {
+            let state = get_global_outbound_tls_snapshot();
+            let identity = state.mtls_identity.as_ref().expect("identity in complete snapshot");
+            assert_eq!(state.root_ca_pem.as_ref().expect("roots in complete snapshot"), &identity.cert_pem);
+            assert_eq!(identity.cert_pem, identity.key_pem);
+            tokio::task::yield_now().await;
+        }
+        writer.await.expect("writer completes");
+    }
+
+    #[test]
+    fn mtls_identity_debug_redacts_private_key() {
+        let identity = MtlsIdentityPem {
+            cert_pem: b"public certificate".to_vec(),
+            key_pem: b"PRIVATE-KEY-MARKER".to_vec(),
+        };
+        let debug = format!("{identity:?}");
+        assert!(!debug.contains("PRIVATE-KEY-MARKER"));
+        assert!(debug.contains("[REDACTED]"));
+    }
+}

crates/config/src/observability/mod.rs

@@ -51,6 +51,11 @@ pub const ENV_OBS_PROFILING_EXPORT_ENABLED: &str = "RUSTFS_OBS_PROFILING_EXPORT_
 /// PUT records multiple timers and histograms when attribution is active.
 pub const ENV_OBS_PUT_STAGE_METRICS_ENABLED: &str = "RUSTFS_OBS_PUT_STAGE_METRICS_ENABLED";
 
+/// Enables sampled in-process GetObject stage summaries without OTLP histograms.
+pub const ENV_OBS_GET_STAGE_LOCAL_SUMMARY_ENABLED: &str = "RUSTFS_OBS_GET_STAGE_LOCAL_SUMMARY_ENABLED";
+pub const ENV_OBS_GET_STAGE_LOCAL_SUMMARY_SAMPLE_RATE: &str = "RUSTFS_OBS_GET_STAGE_LOCAL_SUMMARY_SAMPLE_RATE";
+pub const ENV_OBS_GET_STAGE_LOCAL_SUMMARY_INTERVAL_SECS: &str = "RUSTFS_OBS_GET_STAGE_LOCAL_SUMMARY_INTERVAL_SECS";
+
 pub const ENV_OBS_LOGGER_LEVEL: &str = "RUSTFS_OBS_LOGGER_LEVEL";
 pub const ENV_OBS_LOG_STDOUT_ENABLED: &str = "RUSTFS_OBS_LOG_STDOUT_ENABLED";
 pub const ENV_OBS_LOG_DIRECTORY: &str = "RUSTFS_OBS_LOG_DIRECTORY";
@@ -159,6 +164,15 @@ mod tests {
         assert_eq!(ENV_OBS_LOGS_EXPORT_ENABLED, "RUSTFS_OBS_LOGS_EXPORT_ENABLED");
         assert_eq!(ENV_OBS_PROFILING_EXPORT_ENABLED, "RUSTFS_OBS_PROFILING_EXPORT_ENABLED");
         assert_eq!(ENV_OBS_PUT_STAGE_METRICS_ENABLED, "RUSTFS_OBS_PUT_STAGE_METRICS_ENABLED");
+        assert_eq!(ENV_OBS_GET_STAGE_LOCAL_SUMMARY_ENABLED, "RUSTFS_OBS_GET_STAGE_LOCAL_SUMMARY_ENABLED");
+        assert_eq!(
+            ENV_OBS_GET_STAGE_LOCAL_SUMMARY_SAMPLE_RATE,
+            "RUSTFS_OBS_GET_STAGE_LOCAL_SUMMARY_SAMPLE_RATE"
+        );
+        assert_eq!(
+            ENV_OBS_GET_STAGE_LOCAL_SUMMARY_INTERVAL_SECS,
+            "RUSTFS_OBS_GET_STAGE_LOCAL_SUMMARY_INTERVAL_SECS"
+        );
         // Test log cleanup related env keys
         assert_eq!(ENV_OBS_LOG_MAX_TOTAL_SIZE_BYTES, "RUSTFS_OBS_LOG_MAX_TOTAL_SIZE_BYTES");
         assert_eq!(ENV_OBS_LOG_MAX_SINGLE_FILE_SIZE_BYTES, "RUSTFS_OBS_LOG_MAX_SINGLE_FILE_SIZE_BYTES");

crates/e2e_test/src/common.rs

@@ -1542,6 +1542,15 @@ impl RustFSTestClusterEnvironment {
         self.start_node_from_binary(node_idx, &binary_path).await
     }
 
+    /// Start a process and wait only for TCP, allowing tests to bring up a
+    /// quorum one node at a time before the S3 service can become ready.
+    pub async fn start_node_process(&mut self, node_idx: usize) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+        let binary_path = rustfs_binary_path();
+        let volumes_arg = self.build_volumes_arg();
+        self.spawn_node(node_idx, &binary_path, &volumes_arg)?;
+        self.wait_for_node_ready(&self.nodes[node_idx].address, node_idx).await
+    }
+
     /// Start one stopped cluster node with a specific RustFS binary while
     /// preserving the cluster's volume layout and that node's data directory.
     pub async fn start_node_from_binary(

crates/e2e_test/src/heal_erasure_disk_rebuild_test.rs

@@ -1191,11 +1191,59 @@ mod tests {
         .into())
     }
 
+    fn scanner_cycle_completion_budget(
+        cycle_end: u64,
+        interval: Option<u64>,
+        now: u64,
+        active: bool,
+    ) -> Result<Duration, &'static str> {
+        if cycle_end > now {
+            return Err("scanner cycle timestamp is in the future");
+        }
+        if active {
+            return Ok(Duration::from_secs(60));
+        }
+        let interval = interval
+            .filter(|interval| *interval != 0)
+            .ok_or("scanner schedule has no positive interval")?;
+        let next_cycle = cycle_end
+            .checked_add(interval)
+            .ok_or("scanner schedule timestamp overflows")?;
+        let budget = next_cycle
+            .saturating_sub(now)
+            .checked_add(60)
+            .ok_or("scanner completion budget overflows")?;
+        Ok(Duration::from_secs(budget))
+    }
+
+    #[test]
+    fn scanner_cycle_completion_budget_accounts_for_captured_retry_schedule() {
+        assert_eq!(
+            scanner_cycle_completion_budget(1_791_815_390, Some(256), 1_791_815_447, false),
+            Ok(Duration::from_secs(259)),
+        );
+        assert_eq!(scanner_cycle_completion_budget(100, Some(1), 110, false), Ok(Duration::from_secs(60)));
+    }
+
+    #[test]
+    fn scanner_cycle_completion_budget_does_not_grant_planned_sleep_to_active_cycle() {
+        assert_eq!(scanner_cycle_completion_budget(100, Some(256), 110, true), Ok(Duration::from_secs(60)));
+        assert_eq!(scanner_cycle_completion_budget(0, None, 110, true), Ok(Duration::from_secs(60)));
+    }
+
+    #[test]
+    fn scanner_cycle_completion_budget_rejects_invalid_schedule() {
+        for (end, interval, now) in [(101, 1, 100), (100, 0, 100), (u64::MAX, 1, u64::MAX), (1, u64::MAX - 1, 1)] {
+            assert!(scanner_cycle_completion_budget(end, Some(interval), now, false).is_err());
+        }
+    }
+
     async fn wait_for_scanner_cycle_after(
         cluster: &RustFSTestClusterEnvironment,
         previous_cycle_end: u64,
     ) -> Result<u64, Box<dyn Error + Send + Sync>> {
-        let deadline = Instant::now() + Duration::from_secs(60);
+        let mut deadline = Instant::now() + Duration::from_secs(60);
+        let mut schedule_accounted = false;
         loop {
             let mut latest_cycle_end = 0;
             let mut versions_observed = false;
@@ -1237,14 +1285,36 @@ mod tests {
                 let versions_scanned = metrics["versions_scanned"]
                     .as_u64()
                     .ok_or("scanner status is missing its version-coverage counter")?;
-                latest_cycle_end = latest_cycle_end.max(cycle_end);
+                let now = std::time::SystemTime::now().duration_since(std::time::UNIX_EPOCH)?.as_secs();
+                if cycle_end > now {
+                    return Err("scanner cycle timestamp is in the future".into());
+                }
+                let schedule = &status["cycle_schedule"];
+                if !schedule_accounted && schedule["execution_role"].as_str() == Some("leader") {
+                    let active = metrics["current_cycle_active"]
+                        .as_bool()
+                        .ok_or("scanner status is missing its active-cycle flag")?;
+                    if active || (cycle_end != 0 && schedule["effective_interval_available"].as_bool() == Some(true)) {
+                        let interval = schedule["effective_interval_seconds"].as_u64();
+                        let budget = scanner_cycle_completion_budget(cycle_end, interval, now, active)?;
+                        deadline = Instant::now().checked_add(budget).ok_or("scanner deadline overflows")?;
+                        schedule_accounted = true;
+                    }
+                }
+                // Superseded walks complete all sets while withholding quota authority.
+                // Deferred and partial attempts do not certify completed scanning.
+                if matches!(metrics["last_cycle_result"].as_str(), Some("success" | "superseded")) {
+                    latest_cycle_end = latest_cycle_end.max(cycle_end);
+                }
                 versions_observed |= versions_scanned > 0;
                 observations.push(format!(
-                    "node{node_index}: end={cycle_end}, versions={versions_scanned}, cycle={}, active={}, leader={}, result={}",
+                    "node{node_index}: end={cycle_end}, versions={versions_scanned}, cycle={}, active={}, leader={}, result={}, partial_reason={}, schedule={}",
                     metrics["current_cycle"],
                     metrics["current_cycle_active"],
                     metrics["leader_lock_state"],
                     metrics["last_cycle_result"],
+                    metrics["last_cycle_partial_reason"],
+                    schedule,
                 ));
             }
             // The coordinator records cycle completion, but remote workers
@@ -1555,7 +1625,8 @@ mod tests {
         }
 
         if background_enabled {
-            wait_for_scanner_cycle_after(&cluster, 0).await?;
+            let baseline_cycle_floor = std::time::SystemTime::now().duration_since(std::time::UNIX_EPOCH)?.as_secs();
+            wait_for_scanner_cycle_after(&cluster, baseline_cycle_floor).await?;
         }
 
         cluster.stop_node(1)?;
@@ -1898,11 +1969,16 @@ mod tests {
             let status_body = signed_admin_post(&status_url, None, &cluster.access_key, &cluster.secret_key).await?;
             let status: serde_json::Value = serde_json::from_str(&status_body)
                 .map_err(|err| format!("pre-interrupt background heal status is not JSON ({err}): {status_body}"))?;
-            if background_enabled || only_admin_heal_is_active(&status) {
+            let admin_active = background_rejoin_heal_evidence
+                || (status["state"].as_str() == Some("active")
+                    && status["healOperations"]["activeBySource"]["admin"].as_u64() == Some(1));
+            if admin_active && (background_enabled || only_admin_heal_is_active(&status)) {
                 break status;
             }
             if Instant::now() >= partial_deadline {
-                return Err(format!("non-admin heal did not drain before the interruption deadline: {status}").into());
+                return Err(
+                    format!("heal tasks did not reach the required interruption state before the deadline: {status}").into(),
+                );
             }
             sleep(Duration::from_millis(50)).await;
         };
@@ -2562,4 +2638,157 @@ mod tests {
         };
         assert!(!outage_candidate_targets_replacement(&insufficient_peer_shards, 12, replacement_set_slot));
     }
+    async fn two_node_heal_status(
+        cluster: &RustFSTestClusterEnvironment,
+        bucket: Option<&str>,
+        recreate: bool,
+        scan_mode: u8,
+    ) -> Result<serde_json::Value, Box<dyn Error + Send + Sync>> {
+        let url = format!("{}/rustfs/admin/v3/heal/{}", cluster.nodes[0].url, bucket.unwrap_or_default());
+        let options = serde_json::json!({
+            "recursive": true, "dryRun": false, "remove": false,
+            "recreate": recreate, "scanMode": scan_mode, "updateParity": false, "nolock": false,
+        });
+        let started =
+            start_root_heal_when_control_ready(&url, &options.to_string(), &cluster.access_key, &cluster.secret_key).await?;
+        let started: serde_json::Value = serde_json::from_str(&started)?;
+        let token = started["clientToken"]
+            .as_str()
+            .ok_or_else(|| format!("missing heal token: {started}"))?;
+        let status_url = format!("{url}?clientToken={token}");
+        let deadline = Instant::now() + Duration::from_secs(240);
+        loop {
+            let body = timeout(
+                Duration::from_secs(20),
+                signed_admin_post(&status_url, None, &cluster.access_key, &cluster.secret_key),
+            )
+            .await??;
+            let status: serde_json::Value = serde_json::from_str(&body)?;
+            match status["summary"].as_str() {
+                Some("finished" | "stopped") => return Ok(status),
+                Some("running") if Instant::now() < deadline => sleep(Duration::from_millis(250)).await,
+                _ => return Err(format!("heal did not reach a terminal state: {status}").into()),
+            }
+        }
+    }
+
+    async fn verify_two_node_directory_recovery(
+        remove_bucket: bool,
+        scan_mode: u8,
+        bucket_scoped: bool,
+        recreate: bool,
+    ) -> Result<(), Box<dyn Error + Send + Sync>> {
+        init_logging();
+        let mut cluster = RustFSTestClusterEnvironment::new(2).await?;
+        // Isolate explicit administrator recovery from scanner, MRF, and disk healing.
+        for key in [
+            "RUSTFS_HEAL_AUTO_HEAL_ENABLE",
+            "RUSTFS_HEAL_MRF_ENABLE",
+            "RUSTFS_SCANNER_ENABLED",
+            "RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE",
+        ] {
+            cluster.set_env(key, "false");
+        }
+        cluster.start().await?;
+        let source = cluster.create_s3_client(0)?;
+        let target = cluster.create_s3_client(1)?;
+        let bucket = "two-node-directory-recovery";
+        source.create_bucket().bucket(bucket).send().await?;
+        let objects = [
+            ("empty", Vec::new()),
+            ("nested/inline", vec![0x3c; 1024]),
+            ("nested/external", (0_u8..=255).cycle().take(1024 * 1024).collect::<Vec<_>>()),
+        ];
+        for (key, bytes) in &objects {
+            source
+                .put_object()
+                .bucket(bucket)
+                .key(*key)
+                .body(ByteStream::from(bytes.clone()))
+                .send()
+                .await?;
+        }
+        let target_bucket = Path::new(&cluster.nodes[1].data_dir).join(bucket);
+        let original = objects
+            .iter()
+            .map(|(key, _)| census_object_version_on_disk(Path::new(&cluster.nodes[1].data_dir), bucket, key, None))
+            .collect::<Result<Vec<_>, _>>()?;
+        for census in &original {
+            assert!(census.has_xl_meta, "fixture must have a target metadata copy: {census:?}");
+        }
+        assert!(
+            !original[2].present_part_fingerprints.is_empty(),
+            "large fixture must exercise external parts"
+        );
+        let removed = Path::new(&cluster.temp_dir).join("removed");
+        tokio::fs::create_dir(&removed).await?;
+        if remove_bucket {
+            tokio::fs::rename(&target_bucket, removed.join(bucket)).await?;
+        } else {
+            for (index, (key, _)) in objects.iter().enumerate() {
+                tokio::fs::rename(target_bucket.join(key), removed.join(index.to_string())).await?;
+            }
+        }
+        let status = two_node_heal_status(&cluster, bucket_scoped.then_some(bucket), recreate, scan_mode).await?;
+        assert_eq!(status["summary"], "finished", "administrator recovery should complete: {status}");
+        assert_eq!(status["outcome"]["execution"]["state"], "completed");
+        assert_eq!(status["outcome"]["coverage"], "complete");
+        for ((key, _), before) in objects.iter().zip(&original) {
+            let after = census_object_version_on_disk(Path::new(&cluster.nodes[1].data_dir), bucket, key, None)?;
+            assert!(after.has_xl_meta, "metadata must be restored before GET: {key}");
+            assert_eq!(
+                after.present_part_fingerprints, before.present_part_fingerprints,
+                "external shard bytes must match: {key}"
+            );
+            assert_eq!(
+                after.inline_data_fingerprint, before.inline_data_fingerprint,
+                "inline shard bytes must match: {key}"
+            );
+        }
+        cluster.stop_node(0)?;
+        for (key, expected) in &objects {
+            let actual = timeout(Duration::from_secs(30), async {
+                target
+                    .get_object()
+                    .bucket(bucket)
+                    .key(*key)
+                    .send()
+                    .await?
+                    .body
+                    .collect()
+                    .await
+                    .map(|body| body.into_bytes())
+                    .map_err(Box::<dyn Error + Send + Sync>::from)
+            })
+            .await??;
+            assert_eq!(actual.len(), expected.len(), "target-only read length: {key}");
+            assert_eq!(sha256_hex(&actual), sha256_hex(expected), "target-only content: {key}");
+        }
+        Ok(())
+    }
+
+    #[tokio::test]
+    async fn test_two_node_bucket_discovery_and_explicit_recreate() -> Result<(), Box<dyn Error + Send + Sync>> {
+        verify_two_node_directory_recovery(true, 2, true, true).await
+    }
+
+    #[tokio::test]
+    async fn test_two_node_missing_bucket_default_cluster_recovery() -> Result<(), Box<dyn Error + Send + Sync>> {
+        verify_two_node_directory_recovery(true, 2, false, false).await
+    }
+
+    #[tokio::test]
+    async fn test_two_node_missing_bucket_default_scoped_recovery() -> Result<(), Box<dyn Error + Send + Sync>> {
+        verify_two_node_directory_recovery(true, 2, true, false).await
+    }
+
+    #[tokio::test]
+    async fn test_two_node_object_directory_recovery_normal() -> Result<(), Box<dyn Error + Send + Sync>> {
+        verify_two_node_directory_recovery(false, 1, false, false).await
+    }
+
+    #[tokio::test]
+    async fn test_two_node_object_directory_recovery_deep() -> Result<(), Box<dyn Error + Send + Sync>> {
+        verify_two_node_directory_recovery(false, 2, false, false).await
+    }
 }

crates/e2e_test/src/internode_rpc_signature_e2e_test.rs

@@ -264,7 +264,7 @@ async fn call_make_volume_response(
     request: MakeVolumeRequest,
     headers: HeaderMap,
 ) -> Result<Response<MakeVolumeResponse>, Status> {
-    let mut client = node_service_time_out_client_no_auth(&url.to_string())
+    let mut client = node_service_time_out_client_no_auth(url)
         .await
         .map_err(|err| Status::unavailable(format!("cannot reach the node service: {err}")))?;
     let mut rpc_request = Request::new(request);
@@ -273,7 +273,7 @@ async fn call_make_volume_response(
 }
 
 async fn call_ping_response(url: &str, headers: HeaderMap) -> Result<Response<PingResponse>, Status> {
-    let mut client = node_service_time_out_client_no_auth(&url.to_string())
+    let mut client = node_service_time_out_client_no_auth(url)
         .await
         .map_err(|err| Status::unavailable(format!("cannot reach the node service: {err}")))?;
     let mut rpc_request = Request::new(PingRequest {

crates/e2e_test/src/namespace_lock_quorum_test.rs

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use crate::common::RustFSTestClusterEnvironment;
+use crate::common::{ClusterTopology, RustFSTestClusterEnvironment};
 use aws_sdk_s3::Client;
 use aws_sdk_s3::error::SdkError;
 use bytes::Bytes;
@@ -185,21 +185,113 @@ async fn wait_for_quorum_read_admission(clients: &[Client], bucket: &str) -> Tes
             return Ok(());
         }
         if tokio::time::Instant::now() >= deadline {
-            return Err(format!("read quorum did not become available after lease convergence for {bucket}").into());
+            let diagnostics = clients[0].get_object().bucket(bucket).key("warm-small").send().await;
+            return Err(format!(
+                "read quorum did not become available after lease convergence for {bucket}: GET={diagnostics:?}"
+            )
+            .into());
         }
         tokio::time::sleep(std::time::Duration::from_millis(100)).await;
     }
 }
 
+#[tokio::test]
+async fn test_cold_degraded_object_mutations_require_only_storage_quorum() -> TestResult {
+    crate::common::init_logging();
+    let mut cluster = RustFSTestClusterEnvironment::new(4).await?;
+    cluster.set_env("RUSTFS_STORAGE_CLASS_STANDARD", "EC:2");
+    cluster.set_env("RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE", "false");
+    cluster.set_env("RUSTFS_OBS_METRICS_EXPORT_ENABLED", "false");
+    cluster.set_env("RUSTFS_GET_OBJECT_METADATA_CACHE_DISTRIBUTED_ENABLE", "false");
+    cluster.start().await?;
+    let clients = cluster
+        .create_all_clients()?
+        .into_iter()
+        .map(|client| {
+            Client::from_conf(
+                client
+                    .config()
+                    .to_builder()
+                    .retry_config(aws_sdk_s3::config::retry::RetryConfig::standard().with_max_attempts(1))
+                    .build(),
+            )
+        })
+        .collect::<Vec<_>>();
+    let bucket = "cold-degraded-mutations";
+    clients[0].create_bucket().bucket(bucket).send().await?;
+    // No object mutation may warm a process-wide configuration probe before
+    // the missing peer is introduced. Three shards still satisfy EC:2 writes.
+    cluster.stop_node(3)?;
+    let admission_deadline = Instant::now() + Duration::from_secs(30);
+    for client in &clients[..3] {
+        loop {
+            match client
+                .get_object()
+                .bucket(bucket)
+                .key("never-created-readiness-probe")
+                .send()
+                .await
+            {
+                Err(error) if error.as_service_error().and_then(|error| error.meta().code()) == Some("NoSuchKey") => break,
+                Err(error)
+                    if error.raw_response().is_some_and(|response| response.status().as_u16() == 503)
+                        && Instant::now() < admission_deadline =>
+                {
+                    tokio::time::sleep(Duration::from_millis(100)).await;
+                }
+                Err(error) => return Err(error.into()),
+                Ok(_) => panic!("a cold readiness probe must not create an object"),
+            }
+        }
+    }
+    let body = b"first write while one peer is offline";
+    clients[0]
+        .put_object()
+        .bucket(bucket)
+        .key("object")
+        .body(Bytes::from_static(body).into())
+        .send()
+        .await?;
+    for client in &clients[..3] {
+        assert_quorum_object_body(client, bucket, "object", body).await?;
+    }
+    // This coordinator has not previously committed an object mutation either.
+    clients[1].delete_object().bucket(bucket).key("object").send().await?;
+    let error = clients[2]
+        .get_object()
+        .bucket(bucket)
+        .key("object")
+        .send()
+        .await
+        .expect_err("acknowledged deletion must be visible at another entrypoint");
+    assert_eq!(error.as_service_error().and_then(|error| error.meta().code()), Some("NoSuchKey"));
+
+    cluster.stop_node(2)?;
+    let error = clients[0]
+        .put_object()
+        .bucket(bucket)
+        .key("below-quorum")
+        .body(Bytes::from_static(body).into())
+        .send()
+        .await
+        .expect_err("removing the cache gate must not weaken storage write quorum");
+    assert_eq!(error.as_service_error().and_then(|error| error.meta().code()), Some("ServiceUnavailable"));
+    cluster.stop();
+    Ok(())
+}
+
 #[tokio::test]
 async fn test_degraded_cluster_read_quorum_follows_erasure_layout() -> TestResult {
     crate::common::init_logging();
 
-    for (node_count, parity) in [(4, 2), (6, 3), (6, 2)] {
+    for (node_count, parity) in [(3, 1), (4, 2), (6, 3), (6, 2), (4, 0)] {
         let read_quorum = node_count - parity;
         let write_quorum = read_quorum + usize::from(read_quorum == parity);
         let mut cluster = RustFSTestClusterEnvironment::new(node_count).await?;
         cluster.set_env("RUSTFS_STORAGE_CLASS_STANDARD", format!("EC:{parity}"));
+        if parity == 0 {
+            cluster.set_env("RUSTFS_STORAGE_CLASS_RRS", "EC:0");
+        }
         // Wait for every seed fanout before removing any physical shard.
         cluster.set_env("RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE", "false");
         cluster.set_env("RUSTFS_OBS_METRICS_EXPORT_ENABLED", "false");
@@ -338,6 +430,277 @@ async fn test_degraded_cluster_read_quorum_follows_erasure_layout() -> TestResul
     Ok(())
 }
 
+#[tokio::test]
+async fn test_cold_start_read_quorum_follows_erasure_layout() -> TestResult {
+    crate::common::init_logging();
+    for (node_count, parity) in [(3, 1), (4, 2), (6, 3), (6, 2), (4, 0)] {
+        let read_quorum = node_count - parity;
+        let write_quorum = read_quorum + usize::from(read_quorum == parity);
+        let mut cluster = RustFSTestClusterEnvironment::new(node_count).await?;
+        cluster.set_env("RUSTFS_STORAGE_CLASS_STANDARD", format!("EC:{parity}"));
+        if parity == 0 {
+            cluster.set_env("RUSTFS_STORAGE_CLASS_RRS", "EC:0");
+        }
+        cluster.set_env("RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE", "false");
+        cluster.set_env("RUSTFS_OBS_METRICS_EXPORT_ENABLED", "false");
+        cluster.set_env("RUST_LOG", "warn");
+        cluster.start().await?;
+        let clients = cluster
+            .create_all_clients()?
+            .into_iter()
+            .map(|client| {
+                Client::from_conf(
+                    client
+                        .config()
+                        .to_builder()
+                        .retry_config(aws_sdk_s3::config::retry::RetryConfig::standard().with_max_attempts(1))
+                        .build(),
+                )
+            })
+            .collect::<Vec<_>>();
+        let bucket = format!("cold-read-quorum-{node_count}-{parity}");
+        clients[0].create_bucket().bucket(&bucket).send().await?;
+        let small = b"cold startup follows the stored EC layout".to_vec();
+        let large = vec![73; 256 * 1024];
+        for (key, body) in [
+            ("warm-small", &small),
+            ("warm-large", &large),
+            ("cold-large", &large),
+            ("below-quorum", &large),
+        ] {
+            clients[0]
+                .put_object()
+                .bucket(&bucket)
+                .key(key)
+                .body(Bytes::copy_from_slice(body).into())
+                .send()
+                .await?;
+            for node in &cluster.nodes {
+                let census =
+                    crate::chaos::census_object_version_on_disk(std::path::Path::new(&node.data_dir), &bucket, key, None)?;
+                assert!(census.is_complete(), "seed shard must exist before cold startup: {census:?}");
+                assert_eq!((census.data_blocks, census.parity_blocks), (Some(read_quorum), Some(parity)));
+            }
+        }
+        for node in 0..node_count {
+            cluster.stop_node(node)?;
+        }
+        // Rebuild every process's in-memory state below read quorum, then
+        // cross the EC boundary. A format bootstrap must not admit S3 reads
+        // merely because half the peers are online (e.g. six disks, EC:2).
+        for node in 0..read_quorum - 1 {
+            cluster.start_node_process(node).await?;
+        }
+        let error = clients[0]
+            .get_object()
+            .bucket(&bucket)
+            .key("below-quorum")
+            .send()
+            .await
+            .expect_err("cold startup below the configured EC read quorum must stay unavailable");
+        assert_eq!(error.raw_response().map(|response| response.status().as_u16()), Some(503));
+        cluster.start_node_process(read_quorum - 1).await?;
+        wait_for_quorum_read_admission(&clients[..read_quorum], &bucket).await?;
+        for client in clients.iter().take(read_quorum) {
+            assert_quorum_object_body(client, &bucket, "warm-small", &small).await?;
+            assert_quorum_object_body(client, &bucket, "cold-large", &large).await?;
+            let write = client
+                .put_object()
+                .bucket(&bucket)
+                .key("cold-quorum-write")
+                .body(Bytes::copy_from_slice(&small).into())
+                .send()
+                .await;
+            if read_quorum >= write_quorum {
+                write?;
+                assert_quorum_object_body(client, &bucket, "cold-quorum-write", &small).await?;
+            } else {
+                let error = write.expect_err("cold startup at read quorum must preserve the EC write quorum");
+                assert_eq!(error.raw_response().map(|response| response.status().as_u16()), Some(503));
+            }
+        }
+        for node in read_quorum..node_count {
+            cluster.start_node(node).await?;
+        }
+        for client in &clients {
+            assert_quorum_object_body(client, &bucket, "cold-large", &large).await?;
+        }
+    }
+    Ok(())
+}
+
+#[tokio::test]
+async fn test_four_by_six_cold_start_at_read_quorum() -> TestResult {
+    crate::common::init_logging();
+    let mut cluster = RustFSTestClusterEnvironment::with_topology(ClusterTopology::single_pool_multidrive(4, 6)).await?;
+    // Explicit URLs are consumed in order. Place one drive from each node in
+    // every set, matching a four-host, six-drive-per-host deployment.
+    let volumes = (0..6)
+        .flat_map(|drive| {
+            cluster
+                .nodes
+                .iter()
+                .map(move |node| format!("http://{}{}", node.address, node.data_dirs[drive]))
+        })
+        .collect::<Vec<_>>()
+        .join(" ");
+    cluster.set_env("RUSTFS_VOLUMES", volumes);
+    cluster.set_env("RUSTFS_ERASURE_SET_DRIVE_COUNT", "4");
+    cluster.set_env("RUSTFS_STORAGE_CLASS_STANDARD", "EC:2");
+    cluster.set_env("RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE", "false");
+    cluster.set_env("RUSTFS_OBS_METRICS_EXPORT_ENABLED", "false");
+    cluster.set_env("RUST_LOG", "error");
+    cluster.start().await?;
+    let clients = cluster
+        .create_all_clients()?
+        .into_iter()
+        .map(|client| {
+            Client::from_conf(
+                client
+                    .config()
+                    .to_builder()
+                    .retry_config(aws_sdk_s3::config::retry::RetryConfig::standard().with_max_attempts(1))
+                    .build(),
+            )
+        })
+        .collect::<Vec<_>>();
+    let bucket = "cold-start-read-quorum";
+    clients[0].create_bucket().bucket(bucket).send().await?;
+    let mut fixtures = Vec::new();
+    let mut covered = [[false; 2]; 6];
+    for index in 0..128 {
+        let key = match index {
+            0 => "warm-small".to_string(),
+            1 => "warm-large".to_string(),
+            _ => format!("cold-{index}"),
+        };
+        let size_class = index % 2;
+        let size = if size_class == 0 { 71 } else { 256 * 1024 };
+        let body = (0..size)
+            .map(|offset| u8::try_from(offset % 251).expect("bounded fixture byte"))
+            .collect::<Vec<_>>();
+        clients[0]
+            .put_object()
+            .bucket(bucket)
+            .key(&key)
+            .body(Bytes::copy_from_slice(&body).into())
+            .send()
+            .await?;
+        for node in &cluster.nodes {
+            let mut complete_shards = 0;
+            for (drive, path) in node.data_dirs.iter().enumerate() {
+                let census = crate::chaos::census_object_version_on_disk(std::path::Path::new(path), bucket, &key, None)?;
+                if census.has_xl_meta {
+                    assert!(census.is_complete(), "seed shard must be complete before shutdown: {census:?}");
+                    assert_eq!((census.data_blocks, census.parity_blocks), (Some(2), Some(2)));
+                    covered[drive][size_class] = true;
+                    complete_shards += 1;
+                }
+            }
+            assert_eq!(complete_shards, 1, "each node must own one shard of every seed object");
+        }
+        fixtures.push((key, body));
+        if covered.iter().flatten().all(|covered| *covered) {
+            break;
+        }
+    }
+    assert!(
+        covered.iter().flatten().all(|covered| *covered),
+        "fixtures must cover all six sets with inline and large objects"
+    );
+
+    for first in 0..4 {
+        for second in first + 1..4 {
+            for node in 0..4 {
+                cluster.stop_node(node)?;
+            }
+            cluster.start_node_process(first).await?;
+            let error = clients[first]
+                .get_object()
+                .bucket(bucket)
+                .key("warm-large")
+                .send()
+                .await
+                .expect_err("one node cannot satisfy read quorum after a cold start");
+            assert_eq!(error.raw_response().map(|response| response.status().as_u16()), Some(503));
+            cluster.start_node_process(second).await?;
+            wait_for_quorum_read_admission(&[clients[first].clone(), clients[second].clone()], bucket).await?;
+            for node in [first, second] {
+                for (key, body) in &fixtures {
+                    assert_quorum_object_body(&clients[node], bucket, key, body).await?;
+                }
+                let range = clients[node]
+                    .get_object()
+                    .bucket(bucket)
+                    .key("warm-large")
+                    .range("bytes=17-1023")
+                    .send()
+                    .await?;
+                assert_eq!(range.body.collect().await?.into_bytes().as_ref(), &fixtures[1].1[17..1024]);
+                let listing = clients[node].list_objects_v2().bucket(bucket).send().await?;
+                for (key, _) in &fixtures {
+                    assert!(
+                        listing.contents().iter().any(|object| object.key() == Some(key.as_str())),
+                        "listing omitted {key}"
+                    );
+                }
+                let error = clients[node]
+                    .put_object()
+                    .bucket(bucket)
+                    .key("recovery-write")
+                    .body(Bytes::from_static(b"after recovery").into())
+                    .send()
+                    .await
+                    .expect_err("two nodes must not authorize a write");
+                assert_eq!(error.raw_response().map(|response| response.status().as_u16()), Some(503));
+            }
+            let error = clients[first]
+                .delete_object()
+                .bucket(bucket)
+                .key("warm-large")
+                .send()
+                .await
+                .expect_err("two nodes must not authorize a delete");
+            assert_eq!(error.raw_response().map(|response| response.status().as_u16()), Some(503));
+            assert_quorum_object_body(&clients[first], bucket, "warm-large", &fixtures[1].1).await?;
+
+            let remaining = (0..4).filter(|node| *node != first && *node != second).collect::<Vec<_>>();
+            cluster.start_node(remaining[0]).await?;
+            let deadline = tokio::time::Instant::now() + Duration::from_secs(90);
+            loop {
+                match clients[first]
+                    .put_object()
+                    .bucket(bucket)
+                    .key("recovery-write")
+                    .body(Bytes::from_static(b"after recovery").into())
+                    .send()
+                    .await
+                {
+                    Ok(_) => break,
+                    Err(error)
+                        if error.raw_response().is_some_and(|response| response.status().as_u16() == 503)
+                            && tokio::time::Instant::now() < deadline =>
+                    {
+                        tokio::time::sleep(Duration::from_millis(200)).await
+                    }
+                    Err(error) => return Err(error.into()),
+                }
+            }
+            for node in [first, second, remaining[0]] {
+                assert_quorum_object_body(&clients[node], bucket, "recovery-write", b"after recovery").await?;
+            }
+            cluster.start_node(remaining[1]).await?;
+            for client in &clients {
+                for (key, body) in &fixtures {
+                    assert_quorum_object_body(client, bucket, key, body).await?;
+                }
+                assert_quorum_object_body(client, bucket, "recovery-write", b"after recovery").await?;
+            }
+        }
+    }
+    Ok(())
+}
+
 async fn put_object(client: Client, payload: Vec<u8>, writer_id: usize) -> Result<(), String> {
     client
         .put_object()

crates/e2e_test/src/reliant/grpc_lock_server.rs

@@ -465,6 +465,13 @@ impl NodeService for MinimalLockNodeService {
         Err(Status::unimplemented("lock-only test server"))
     }
 
+    async fn recover_delete_transaction(
+        &self,
+        _request: Request<rustfs_protos::proto_gen::node_service::RecoverDeleteTransactionRequest>,
+    ) -> Result<Response<rustfs_protos::proto_gen::node_service::RecoverDeleteTransactionResponse>, Status> {
+        Err(Status::unimplemented("delete recovery is not supported by the lock fixture"))
+    }
+
     async fn settle_part_transaction(
         &self,
         _request: Request<rustfs_protos::proto_gen::node_service::SettlePartTransactionRequest>,

crates/e2e_test/src/replication_extension_test.rs

@@ -3945,7 +3945,15 @@ async fn test_bucket_replication_version_purge_of_non_inline_object_releases_sou
         (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?;
+    // Managed-SSE listings may omit ETags. Check the replicated version's body
+    // before deleting it rather than comparing keyless listing fingerprints.
+    for (key, version_id) in &purged {
+        let source_body = get_version_body(&source_client, &source_bucket, key, version_id).await?;
+        let expected_sha256: [u8; 32] = Sha256::digest(&source_body).into();
+        wait_for_replicated_sha256(&target_client, &target_bucket, key, expected_sha256).await?;
+        let target_body = get_version_body(&target_client, &target_bucket, key, version_id).await?;
+        assert_eq!(target_body, source_body, "replicated version body differs for {key}/{version_id}");
+    }
 
     for (key, version_id) in &purged {
         source_client
@@ -3956,10 +3964,6 @@ async fn test_bucket_replication_version_purge_of_non_inline_object_releases_sou
             .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
@@ -3984,6 +3988,9 @@ async fn test_bucket_replication_version_purge_of_non_inline_object_releases_sou
         }
     }
 
+    let target_state = list_replication_state(&target_client, &target_bucket).await?;
+    assert!(target_state.is_empty(), "target retained an explicitly purged version: {target_state:?}");
+
     Ok(())
 }
 

crates/e2e_test/src/sts_query_compat_test.rs

@@ -740,7 +740,7 @@ async fn test_bucket_tags_opa_contract() -> TestResult {
         .version_id(version)
         .send()
         .await?;
-    opa.expect_bucket_tags("s3:GetObject", source, Some("finance")).await?;
+    opa.expect_bucket_tags("s3:GetObjectVersion", source, Some("finance")).await?;
     client.list_objects_v2().bucket(source).send().await?;
     opa.expect_bucket_tags("s3:ListBucket", source, Some("finance")).await?;
     client.list_object_versions().bucket(source).send().await?;

crates/e2e_test/src/upgrade_compatibility_test.rs

@@ -2345,7 +2345,7 @@ fn layout_reported_parts(attributes: &aws_sdk_s3::operation::get_object_attribut
     attributes.object_parts().map(|parts| parts.parts().len())
 }
 
-async fn assert_layout_readable(client: &Client, case: &LayoutCase, context: &str) -> TestResult {
+async fn assert_layout_full_body(client: &Client, case: &LayoutCase, context: &str) -> TestResult {
     let label = case.label();
     let head = layout_head(client, case).await?;
     assert_eq!(
@@ -2362,7 +2362,12 @@ async fn assert_layout_readable(client: &Client, case: &LayoutCase, context: &st
     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");
+    Ok(())
+}
 
+async fn assert_layout_readable(client: &Client, case: &LayoutCase, context: &str) -> TestResult {
+    assert_layout_full_body(client, case, context).await?;
+    let label = case.label();
     if case.is_multipart_layout() {
         let first = case.part_sizes[0];
         let range = format!("bytes={}-{}", first - 32, first + 31);
@@ -2641,10 +2646,11 @@ async fn direct_upgrade_from_rc5_preserves_multipart_layouts() -> TestResult {
             "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.
+    // Prove the rc.5 writer persisted every original byte before upgrading.
+    // Its compressed cross-part range reader has a legacy limitation; the
+    // upgraded reader must still satisfy every range and partNumber assertion.
     for case in &cases {
-        assert_layout_readable(&old_client, case, "rc.5").await?;
+        assert_layout_full_body(&old_client, case, "rc.5").await?;
     }
 
     // Upgrade in place.

crates/ecstore/Cargo.toml

@@ -183,6 +183,7 @@ sha2 = { workspace = true }
 hex-simd = { workspace = true }
 tempfile.workspace = true
 hyper = { workspace = true, features = ["http2", "http1", "server"] }
+h2 = { workspace = true }
 hyper-util = { workspace = true, features = ["tokio", "server-auto", "server-graceful", "tracing"] }
 hyper-rustls = { workspace = true, default-features = false, features = ["native-tokio", "http1", "tls12", "logging", "http2", "aws-lc-rs"] }
 hostname.workspace = true
@@ -200,8 +201,8 @@ md-5.workspace = true
 memmap2 = { workspace = true }
 libc.workspace = true
 # "process" adds getrlimit for the io_uring fd-cache RLIMIT_NOFILE gate
-# (backlog#1178); "fs" comes from the workspace default.
-rustix = { workspace = true, features = ["process", "fs"] }
+# (backlog#1178); "thread" provides cleanup affinity; "fs" is a workspace default.
+rustix = { workspace = true, features = ["process", "fs", "thread"] }
 rustfs-madmin.workspace = true
 reqwest = { workspace = true }
 aws-sdk-s3 = { workspace = true, default-features = false, features = ["sigv4a", "default-https-client", "rt-tokio"] }

crates/ecstore/src/api/mod.rs

@@ -199,10 +199,10 @@ pub mod bucket {
             get_object_lock_config, get_object_lock_config_state, get_on_demand_migration_config,
             get_on_demand_migration_config_in, get_public_access_block_config, get_quota_config, get_replication_config,
             get_request_payment_config, get_sse_config, get_tagging_config, get_versioning_config, get_website_config,
-            init_bucket_metadata_sys, list_bucket_targets, reload_bucket_metadata, remove_bucket_metadata, set_bucket_metadata,
-            update, update_bucket_targets_under_transaction_lock, update_config_with, update_if_incarnation,
-            update_if_incarnation_at, update_quota_if_incarnation, update_quota_if_incarnation_at, update_under_transaction_lock,
-            update_under_transaction_lock_at,
+            init_bucket_metadata_sys, list_bucket_targets, reload_bucket_metadata, remove_bucket_metadata,
+            remove_bucket_metadata_if_absent, set_bucket_metadata, update, update_bucket_targets_under_transaction_lock,
+            update_config_with, update_if_incarnation, update_if_incarnation_at, update_quota_if_incarnation,
+            update_quota_if_incarnation_at, update_under_transaction_lock, update_under_transaction_lock_at,
         };
         #[cfg(feature = "test-util")]
         pub use crate::bucket::metadata_sys::{ConfigWriteLockProbe, test_support};
@@ -210,7 +210,8 @@ pub mod bucket {
 
     pub mod migration {
         pub use crate::bucket::migration::{
-            LegacyBlobDecryptFn, migration_startup_error, try_migrate_bucket_metadata, try_migrate_iam_config,
+            LegacyBlobDecryptFn, migration_startup_error, try_migrate_bucket_metadata, try_migrate_bucket_metadata_for_buckets,
+            try_migrate_iam_config,
         };
     }
 
@@ -371,24 +372,6 @@ pub mod config {
             with_config_object_read_lock, with_config_object_write_lock, with_server_config_read_lock,
             with_server_config_write_lock,
         };
-
-        /// Read config bytes and metadata without treating an empty payload as absent.
-        pub async fn read_config_preserve_empty_with_metadata_opts(
-            api: std::sync::Arc<crate::store::ECStore>,
-            file: &str,
-            opts: &crate::object_api::ObjectOptions,
-        ) -> crate::error::Result<(Vec<u8>, crate::object_api::ObjectInfo)> {
-            crate::config::com::read_config_preserve_empty_with_metadata_opts(api, file, opts).await
-        }
-
-        pub async fn read_config_limited_preserve_empty_with_metadata_opts(
-            api: std::sync::Arc<crate::store::ECStore>,
-            file: &str,
-            opts: &crate::object_api::ObjectOptions,
-            max_bytes: usize,
-        ) -> crate::error::Result<(Vec<u8>, crate::object_api::ObjectInfo)> {
-            crate::config::com::read_config_limited_preserve_empty_with_metadata_opts(api, file, opts, max_bytes).await
-        }
     }
 
     pub mod storageclass {
@@ -440,7 +423,9 @@ pub mod disk {
     pub use crate::disk::os::{LocalPublicationPause, LocalPublicationStage};
     pub use crate::disk::{
         BATCH_READ_VERSION_MAX_ITEMS, BUCKET_META_PREFIX, BatchReadVersionItem, BatchReadVersionReq, BatchReadVersionResp,
-        CheckPartsResp, ConditionalFileUpdate, DeleteOptions, Disk, DiskAPI, DiskInfo, DiskInfoOptions, DiskLocation, DiskOption,
+        CheckPartsResp, ConditionalFileUpdate, DELETE_RECOVERY_PROTOCOL_VERSION, DeleteOptions, DeleteRecoveryAction,
+        DeleteRecoveryCursor, DeleteRecoveryDigest, DeleteRecoveryInspection, DeleteRecoveryPage, DeleteRecoveryRpcOperation,
+        DeleteRecoveryRpcReply, DeleteRecoveryState, Disk, DiskAPI, DiskInfo, DiskInfoOptions, DiskLocation, DiskOption,
         DiskStore, FileInfoVersions, FileReader, FileWriter, HEALING_MARKER_PATH, NsScannerOpenRequest, OldCurrentSize,
         PartTransactionAction, RUSTFS_META_BUCKET, ReadMultipleReq, ReadMultipleResp, ReadOptions, RenameDataResp,
         STORAGE_FORMAT_FILE, SnapshotLeaseToken, UpdateMetadataOpts, VolumeInfo, WalkDirOptions, new_disk,
@@ -537,13 +522,14 @@ pub mod integrity {
 pub mod object {
     pub use crate::object_api::{
         BLOCK_SIZE_V2, ERASURE_ALGORITHM, EncryptionResolutionError, EncryptionResolutionErrorKind, GetObjectBodyCacheHook,
-        GetObjectBodyCacheHookLookup, GetObjectBodySource, GetObjectReader, NamespaceLockFence, ObjectEncryptionResolver,
-        ObjectInfo, ObjectLockConfigSnapshot, ObjectMutationHook, ObjectOptions, PutObjReader, QuotaAdmission,
-        RangedDecompressReader, ReadEncryptionMaterial, ReadEncryptionMode, ReadEncryptionRequest,
+        GetObjectBodyCacheHookLookup, GetObjectBodySource, GetObjectReadCondition, GetObjectReadMetadata, GetObjectReader,
+        NamespaceLockFence, ObjectEncryptionResolver, ObjectInfo, ObjectLockConfigSnapshot, ObjectMutationHook, ObjectOptions,
+        PutObjReader, QuotaAdmission, RangedDecompressReader, ReadEncryptionMaterial, ReadEncryptionMode, ReadEncryptionRequest,
         SCANNER_PUBLICATION_LEASE_FENCE_METADATA_KEY, ScannerPublicationCommitScope, ScannerPublicationCommitStartError,
         ScannerPublicationCommitState, ShardIntegrityWriteMode, StreamConsumer, WriteCommitGuard, WriteCompletion,
         get_object_body_cache_plaintext_len, lookup_get_object_body_cache_hook, register_get_object_body_cache_hook,
         register_object_mutation_hook, unregister_get_object_body_cache_hook, unregister_object_mutation_hook,
+        with_get_object_read_condition,
     };
     pub use crate::store::{
         PrepareSelectObjectSnapshotError, PreparedGetObjectReader, SelectObjectSnapshot, SelectObjectSnapshotReadError,
@@ -608,6 +594,7 @@ pub mod set_disk {
     #[cfg(feature = "test-util")]
     pub mod test_util {
         pub use crate::bucket::quota::reservation::fail_next_quota_ledger_save_for_test;
+        pub use crate::set_disk::ConditionalReadBenchmarkMetadataGuard;
         pub use crate::set_disk::{MultipartCommitBarrier, MultipartCommitPause, PutObjectCommitBarrier, PutObjectCommitPause};
 
         /// Keep a namespace commit pending until the returned owner is dropped.

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

@@ -2417,6 +2417,16 @@ fn is_slow_down(err: &Error) -> bool {
     matches!(err, Error::SlowDown)
 }
 
+/// Lifecycle expiry is idempotent when the exact object or version has already
+/// disappeared between the scanner listing and the delete admission.
+///
+/// Keep this predicate narrower than bucket/volume "not found" helpers: an
+/// unavailable bucket or disk is an operational failure and must not be
+/// reported as a completed expiry.
+fn is_lifecycle_expiry_already_gone(err: &Error) -> bool {
+    is_err_object_not_found(err) || is_err_version_not_found(err)
+}
+
 /// Resolves the expiry worker count from the single documented knob,
 /// `RUSTFS_MAX_EXPIRY_WORKERS`: a set, parsable, non-zero value wins;
 /// anything else falls back to `min(cpus, 16)`. The historical
@@ -4920,6 +4930,20 @@ async fn expire_transitioned_object_with_lock_lost_signal(
                 //audit_log_lifecycle(*oi, ILMExpiry, tags, traceFn);
                 Ok(dobj)
             }
+            Err(err) if is_lifecycle_expiry_already_gone(&err) => {
+                debug!(
+                    event = EVENT_LIFECYCLE_DELETE_COMPLETED,
+                    component = LOG_COMPONENT_ECSTORE,
+                    subsystem = LOG_SUBSYSTEM_LIFECYCLE,
+                    bucket = %oi.bucket,
+                    object = %oi.name,
+                    operation = "delete_restored_object",
+                    reason = "already_gone",
+                    "Lifecycle delete target was already absent"
+                );
+                crate::object_api::notify_object_mutation(&oi.bucket, &oi.name).await;
+                Ok(oi.clone())
+            }
             Err(err) => Err(std::io::Error::other(err)),
         };
     }
@@ -4930,6 +4954,8 @@ async fn expire_transitioned_object_with_lock_lost_signal(
     // pointing to a remote tier version that has already been removed. If this
     // only creates a delete marker, remote cleanup must be driven by persisted
     // free-version recovery rather than the visible delete result.
+    // Keep remote transitioned not-found errors visible: local metadata
+    // absence alone does not prove that the tier object was cleaned up.
     let dobj = match api.delete_object(&oi.bucket, &oi.name, opts).await {
         Ok(obj) => obj,
         Err(e) => {
@@ -5802,6 +5828,20 @@ async fn apply_expiry_on_non_transitioned_objects_with_lock_lost_signal(
         .await
     {
         Ok(dobj) => dobj,
+        Err(err) if is_lifecycle_expiry_already_gone(&err) => {
+            debug!(
+                event = EVENT_LIFECYCLE_DELETE_COMPLETED,
+                component = LOG_COMPONENT_ECSTORE,
+                subsystem = LOG_SUBSYSTEM_LIFECYCLE,
+                bucket = %oi.bucket,
+                object = %oi.name,
+                operation = "delete_object",
+                reason = "already_gone",
+                "Lifecycle delete target was already absent"
+            );
+            crate::object_api::notify_object_mutation(&oi.bucket, &oi.name).await;
+            return true;
+        }
         Err(e) => {
             error!(
                 event = EVENT_LIFECYCLE_DELETE_FAILED,
@@ -6041,18 +6081,19 @@ mod tests {
         cleanup_empty_multipart_sha_dirs_on_local_disks, cleanup_stale_multipart_uploads_once_at,
         enqueue_recovered_free_version_with_state, enqueue_transition_for_existing_objects_scoped,
         enqueue_transition_with_lifecycle, enqueue_transition_with_lifecycle_report, eval_action_from_lifecycle,
-        get_lock_acquire_timeout, jitter_tier_free_version_recovery_delay, lifecycle_action_blocked_by_replication,
-        lifecycle_delete_all_versions_replication_scan, lifecycle_deleted_object, lifecycle_replication_blocks_action,
-        lifecycle_rule_has_date_expiration, lifecycle_version_delete_target, manual_transition_duration_elapsed,
-        manual_transition_has_more_after_limit, manual_transition_recovery_progress_sink, manual_transition_version_marker,
-        manual_transition_worker_failure_reason, mark_delete_opts_skip_decommissioned_on_remote_success,
-        merge_stale_multipart_candidate, persist_manual_transition_job_progress_if_owned,
-        persist_manual_transition_page_checkpoint, recover_manual_transition_job, recover_manual_transition_jobs,
-        resolve_tier_free_version_recovery_enabled, resolve_transition_queue_capacity, resolve_transition_queue_send_timeout,
-        resolve_transition_worker_count, resolve_transition_workers_absolute_max, run_tier_free_version_recovery_coordinator,
-        run_tier_free_version_recovery_loop, select_restore_s3_location, set_lifecycle_observability_observer,
-        set_recovered_free_version_enqueue_observer, should_defer_date_expiry_for_recent_config_update,
-        transitioned_cleanup_tuple, transitioned_object_delete_opts, wait_for_tier_free_version_recovery,
+        get_lock_acquire_timeout, is_lifecycle_expiry_already_gone, jitter_tier_free_version_recovery_delay,
+        lifecycle_action_blocked_by_replication, lifecycle_delete_all_versions_replication_scan, lifecycle_deleted_object,
+        lifecycle_replication_blocks_action, lifecycle_rule_has_date_expiration, lifecycle_version_delete_target,
+        manual_transition_duration_elapsed, manual_transition_has_more_after_limit, manual_transition_recovery_progress_sink,
+        manual_transition_version_marker, manual_transition_worker_failure_reason,
+        mark_delete_opts_skip_decommissioned_on_remote_success, merge_stale_multipart_candidate,
+        persist_manual_transition_job_progress_if_owned, persist_manual_transition_page_checkpoint,
+        recover_manual_transition_job, recover_manual_transition_jobs, resolve_tier_free_version_recovery_enabled,
+        resolve_transition_queue_capacity, resolve_transition_queue_send_timeout, resolve_transition_worker_count,
+        resolve_transition_workers_absolute_max, run_tier_free_version_recovery_coordinator, run_tier_free_version_recovery_loop,
+        select_restore_s3_location, set_lifecycle_observability_observer, set_recovered_free_version_enqueue_observer,
+        should_defer_date_expiry_for_recent_config_update, transitioned_cleanup_tuple, transitioned_object_delete_opts,
+        wait_for_tier_free_version_recovery,
     };
     #[cfg(feature = "test-util")]
     use super::{delete_free_version_remote_object_then, encode_dir_object, get_transitioned_object_reader_with_tier_manager};
@@ -10511,6 +10552,26 @@ mod tests {
         );
     }
 
+    #[test]
+    fn lifecycle_expiry_already_gone_only_accepts_object_or_version_absence() {
+        assert!(is_lifecycle_expiry_already_gone(&Error::FileNotFound));
+        assert!(is_lifecycle_expiry_already_gone(&Error::ObjectNotFound(
+            "bucket".to_string(),
+            "object".to_string(),
+        )));
+        assert!(is_lifecycle_expiry_already_gone(&Error::FileVersionNotFound));
+        assert!(is_lifecycle_expiry_already_gone(&Error::VersionNotFound(
+            "bucket".to_string(),
+            "object".to_string(),
+            "version".to_string(),
+        )));
+
+        assert!(!is_lifecycle_expiry_already_gone(&Error::VolumeNotFound));
+        assert!(!is_lifecycle_expiry_already_gone(&Error::DiskNotFound));
+        assert!(!is_lifecycle_expiry_already_gone(&Error::ErasureReadQuorum));
+        assert!(!is_lifecycle_expiry_already_gone(&Error::MethodNotAllowed));
+    }
+
     #[tokio::test]
     async fn manual_transition_counts_already_transitioned_object() {
         let lc = latest_transition_lifecycle();
@@ -13385,7 +13446,9 @@ mod tests {
                 .bucket_incarnation_id_from_disk(&bucket)
                 .await
                 .expect("bucket incarnation");
-            for (attempt_incarnation, expected) in [(Uuid::new_v4(), false), (incarnation, true), (incarnation, false)] {
+            // Replaying the same scanner snapshot after the delete is an
+            // already-gone success and must not be reported as a failure.
+            for (attempt_incarnation, expected) in [(Uuid::new_v4(), false), (incarnation, true), (incarnation, true)] {
                 assert_eq!(
                     super::apply_expiry_on_non_transitioned_objects(
                         ecstore.clone(),

crates/ecstore/src/bucket/metadata_sys.rs

@@ -178,6 +178,13 @@ pub(crate) fn object_lock_config_state_from_authoritative_metadata(bm: &BucketMe
     Ok(ObjectLockConfigState::ConfirmedAbsent)
 }
 
+#[derive(Debug, Clone, PartialEq)]
+pub(crate) struct CachedQuotaMetadataSnapshot {
+    pub(crate) quota: Option<BucketQuota>,
+    pub(crate) bucket_incarnation: Uuid,
+    pub(crate) quota_revision: OffsetDateTime,
+}
+
 /// Convert the persisted serving-layer configuration into the storage-level
 /// [`DefaultRetention`](crate::bucket::object_lock::types::DefaultRetention)
 /// the WORM evaluation code consumes (rustfs/backlog#1842). A rule without a
@@ -314,6 +321,33 @@ pub async fn remove_bucket_metadata(bucket: &str) -> Result<bool> {
     Ok(lock.remove(bucket).await)
 }
 
+/// Remove a peer's cached metadata only when the bucket is absent under the
+/// namespace read fence. A delayed delete notification must not withdraw the
+/// metadata of a same-name bucket that has already been recreated.
+pub async fn remove_bucket_metadata_if_absent(api: Arc<ECStore>, bucket: &str) -> Result<bool> {
+    if is_meta_bucketname(bucket) {
+        return Ok(false);
+    }
+    // Lock order: bucket namespace read -> metadata system read (released)
+    // -> metadata publish, matching peer reloads.
+    let namespace_lock = api.new_ns_lock(bucket, bucket).await?;
+    let namespace_guard = namespace_lock
+        .get_read_lock(crate::set_disk::get_lock_acquire_timeout())
+        .await?;
+    let sys = bucket_metadata_sys_of(&api.ctx)?;
+    let sys = sys.read().await.clone();
+    if sys
+        .bucket_exists(bucket, &namespace_guard, "peer bucket metadata removal existence check")
+        .await?
+    {
+        return Ok(false);
+    }
+    await_bucket_namespace_operation(Some(&namespace_guard), bucket, "peer bucket metadata removal", async {
+        Ok(sys.remove(bucket).await)
+    })
+    .await
+}
+
 fn start_refresh_buckets_metadata_loop(sys: Arc<RwLock<BucketMetadataSys>>) {
     let Some(cancel_token) = runtime_sources::background_services_cancel_token() else {
         warn!("bucket metadata refresh loop skipped because background cancellation token is not initialized");
@@ -491,6 +525,36 @@ pub(crate) async fn get_bucket_incarnation_id_in(ctx: &crate::runtime::instance:
     sys.get_bucket_incarnation_id_from_disk(bucket).await
 }
 
+/// Directory loss is recoverable only for a committed, non-retired generation.
+/// The caller retains the bucket lifecycle fence through the ensuing reconstruction.
+pub(crate) async fn validate_bucket_reconstruction_in(
+    ctx: &crate::runtime::instance::InstanceContext,
+    bucket: &str,
+    expected: Uuid,
+) -> Result<()> {
+    let api = bucket_metadata_sys_of(ctx)?.read().await.object_store();
+    let lock = api
+        .new_ns_lock(RUSTFS_META_BUCKET, &bucket_metadata_transaction_lock_key(bucket))
+        .await?;
+    let guard = lock.get_read_lock(crate::set_disk::get_lock_acquire_timeout()).await?;
+    let (metadata, persisted) = load_bucket_metadata_parse_with_presence(api.clone(), bucket, true).await?;
+    if !persisted
+        || !metadata.bucket_incarnation_sidecar
+        || expected.is_nil()
+        || metadata.bucket_incarnation_id != expected
+        || !metadata.bucket_creation_committed
+    {
+        return Err(Error::PreconditionFailed);
+    }
+    if crate::bucket::retirement::is_retired(api, bucket, expected).await? {
+        return Err(Error::BucketNotFound(bucket.to_owned()));
+    }
+    if guard.is_lock_lost() {
+        return Err(Error::PreconditionFailed);
+    }
+    Ok(())
+}
+
 /// Validate against disk while retaining an already-held Object Lock fence.
 pub(crate) async fn get_bucket_incarnation_id_for_options_in(
     ctx: &crate::runtime::instance::InstanceContext,
@@ -1423,6 +1487,9 @@ fn quota_config_and_incarnation_from_authority(
         BucketMetadataAuthority::Authoritative(metadata)
             if metadata.bucket_incarnation_sidecar && !metadata.bucket_incarnation_id.is_nil() =>
         {
+            if !metadata.quota_config_json.is_empty() && metadata.quota_config.is_none() {
+                return Err(Error::other("persisted bucket quota configuration is invalid"));
+            }
             Ok((
                 metadata.quota_config.clone(),
                 metadata.bucket_incarnation_id,
@@ -1437,6 +1504,15 @@ fn quota_config_and_incarnation_from_authority(
     }
 }
 
+pub(crate) async fn get_cached_quota_config_and_incarnation_in(
+    ctx: &crate::runtime::instance::InstanceContext,
+    bucket: &str,
+) -> Result<Option<CachedQuotaMetadataSnapshot>> {
+    let bucket_meta_sys_lock = bucket_metadata_sys_of(ctx)?;
+    let bucket_meta_sys = bucket_meta_sys_lock.read().await.clone();
+    bucket_meta_sys.cached_quota_config_and_incarnation(bucket).await
+}
+
 pub async fn get_replication_config(bucket: &str) -> Result<(ReplicationConfiguration, OffsetDateTime)> {
     let bucket_meta_sys_lock = get_bucket_metadata_sys()?;
     let bucket_meta_sys = bucket_meta_sys_lock.read().await;
@@ -2952,6 +3028,27 @@ impl BucketMetadataSys {
         }
     }
 
+    async fn cached_quota_config_and_incarnation(&self, bucket: &str) -> Result<Option<CachedQuotaMetadataSnapshot>> {
+        let metadata = {
+            let metadata_map = self.metadata_map.read().await;
+            metadata_map.get(bucket).cloned()
+        };
+        let Some(metadata) = metadata else {
+            return Ok(None);
+        };
+        if !metadata.bucket_incarnation_sidecar || metadata.bucket_incarnation_id.is_nil() {
+            return Ok(None);
+        }
+        if !metadata.quota_config_json.is_empty() && metadata.quota_config.is_none() {
+            return Err(Error::other("persisted bucket quota configuration is invalid"));
+        }
+        Ok(Some(CachedQuotaMetadataSnapshot {
+            quota: metadata.quota_config.clone(),
+            bucket_incarnation: metadata.bucket_incarnation_id,
+            quota_revision: metadata.quota_config_updated_at,
+        }))
+    }
+
     pub async fn get_replication_config(&self, bucket: &str) -> Result<(ReplicationConfiguration, OffsetDateTime)> {
         let (bm, _) = self.get_config(bucket).await?;
 
@@ -4856,6 +4953,154 @@ mod tests {
         );
     }
 
+    #[tokio::test]
+    #[serial]
+    async fn quota_admission_rejects_stale_cached_no_quota() {
+        let (_dirs, store) = isolated_store_over_temp_disks().await;
+        init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+        let sys = bucket_metadata_sys_of(&store.ctx).expect("scoped metadata system");
+        let bucket = "stale-no-quota-admission";
+        store
+            .make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("create bucket");
+        let (stale, _) = sys.read().await.get_config(bucket).await.expect("cache initial metadata");
+        store
+            .update_bucket_metadata_config(bucket, "quota.json", br#"{"quota":1}"#.to_vec())
+            .await
+            .expect("publish quota to authoritative storage");
+        // Model a peer that has not received the newer configuration snapshot.
+        sys.read().await.set(bucket.to_owned(), stale).await;
+        assert!(
+            get_cached_quota_config_and_incarnation_in(&store.ctx, bucket)
+                .await
+                .expect("read stale cache")
+                .expect("cached bucket identity")
+                .quota
+                .is_none()
+        );
+
+        temp_env::async_with_vars([("RUSTFS_QUOTA_BEGIN_SAFE_NO_QUOTA_CACHE_FAST_PATH", Some("true"))], async {
+            let error = crate::bucket::quota::reservation::begin(&store.ctx, bucket, "object", &ObjectOptions::default(), 0, 0)
+                .await
+                .err()
+                .expect("persisted quota must reject admission without its snapshot");
+            assert!(matches!(error, Error::PartMissingOrCorrupt), "{error:?}");
+        })
+        .await;
+    }
+
+    #[tokio::test]
+    #[serial]
+    async fn peer_reload_clears_negative_cache_after_remote_bucket_creation() {
+        let (dirs, node_a, node_b) = two_nodes_over_one_store().await;
+        let bucket = "cross-node-negative-cache-refresh";
+        let node_b_sys = node_b.read().await.clone();
+
+        assert_eq!(
+            node_b_sys.get_bucket_incarnation_id(bucket).await.unwrap_err(),
+            Error::BucketNotFound(bucket.to_string()),
+            "the probe must populate node B's missing-bucket cache"
+        );
+
+        for dir in &dirs {
+            std::fs::create_dir_all(dir.path().join(bucket)).expect("physical bucket should exist before creation metadata");
+        }
+        let created = BucketMetadata::new(bucket);
+        let created_incarnation = created.bucket_incarnation_id;
+        node_a
+            .read()
+            .await
+            .persist_new_and_set(created)
+            .await
+            .expect("node A should publish the created bucket metadata");
+
+        node_b_sys
+            .reload_from_store(bucket)
+            .await
+            .expect("peer reload should clear the stale negative cache");
+        assert_eq!(
+            node_b_sys.get_bucket_incarnation_id(bucket).await.unwrap(),
+            created_incarnation,
+            "node B should resolve the newly created bucket immediately"
+        );
+    }
+
+    #[tokio::test(flavor = "multi_thread", worker_threads = 4)]
+    #[serial]
+    async fn peer_delete_clears_metadata_for_absent_bucket() {
+        let (_dirs, store) = isolated_store_over_temp_disks().await;
+        init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+        let sys = bucket_metadata_sys_of(&store.ctx).expect("metadata owner");
+        let bucket = "peer-delete-absent-bucket";
+
+        store
+            .make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("initial bucket creation");
+        let cached = sys.read().await.get(bucket).await.expect("initial cached metadata");
+        store
+            .delete_bucket(bucket, &DeleteBucketOptions::default())
+            .await
+            .expect("bucket deletion");
+
+        // Model a peer that still caches the metadata after the physical delete.
+        sys.read().await.set(bucket.to_string(), cached).await;
+        assert!(
+            remove_bucket_metadata_if_absent(store.clone(), bucket)
+                .await
+                .expect("conditional peer delete"),
+            "the peer must evict the deleted bucket's cached metadata"
+        );
+        let sys = sys.read().await;
+        assert!(matches!(sys.get(bucket).await, Err(Error::ConfigNotFound)));
+        assert_eq!(
+            sys.get_bucket_incarnation_id(bucket)
+                .await
+                .expect_err("deleted bucket lookup"),
+            Error::BucketNotFound(bucket.to_string()),
+        );
+    }
+
+    #[tokio::test(flavor = "multi_thread", worker_threads = 4)]
+    #[serial]
+    async fn delayed_peer_delete_preserves_recreated_bucket_metadata() {
+        let (_dirs, store) = isolated_store_over_temp_disks().await;
+        init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+        let sys = bucket_metadata_sys_of(&store.ctx).expect("metadata owner");
+        let bucket = "delayed-peer-delete-recreated-bucket";
+
+        store
+            .make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("initial bucket creation");
+        let old_incarnation = store.bucket_incarnation_id_from_disk(bucket).await.expect("old incarnation");
+        store
+            .delete_bucket(bucket, &DeleteBucketOptions::default())
+            .await
+            .expect("bucket deletion");
+        store
+            .make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("bucket recreation");
+        let new_incarnation = store.bucket_incarnation_id_from_disk(bucket).await.expect("new incarnation");
+        assert_ne!(old_incarnation, new_incarnation);
+
+        let removed = remove_bucket_metadata_if_absent(store.clone(), bucket)
+            .await
+            .expect("conditional peer delete");
+        assert!(!removed, "a delayed delete must not remove metadata for a live recreated bucket");
+        assert_eq!(
+            sys.read()
+                .await
+                .get(bucket)
+                .await
+                .expect("recreated metadata should remain cached")
+                .bucket_incarnation_id,
+            new_incarnation,
+        );
+    }
+
     #[tokio::test(flavor = "multi_thread", worker_threads = 4)]
     #[serial]
     async fn stale_config_request_cannot_mutate_a_recreated_bucket() {

crates/ecstore/src/bucket/migration.rs

@@ -273,6 +273,30 @@ where
 
     let buckets: Vec<String> = buckets_list.into_iter().map(|b| b.name).collect();
 
+    try_migrate_bucket_metadata_for_buckets(store, &buckets).await
+}
+
+/// Migrate the already discovered startup candidates without repeating a
+/// majority-only namespace listing. Every migration write retains its quorum.
+pub async fn try_migrate_bucket_metadata_for_buckets<S>(store: Arc<S>, buckets: &[String]) -> Result<()>
+where
+    S: ObjectIO<
+            Error = Error,
+            RangeSpec = HTTPRangeSpec,
+            HeaderMap = HeaderMap,
+            ObjectOptions = ObjectOptions,
+            ObjectInfo = ObjectInfo,
+            GetObjectReader = GetObjectReader,
+            PutObjectReader = PutObjReader,
+        > + ObjectOperations<
+            Error = Error,
+            ObjectInfo = ObjectInfo,
+            ObjectOptions = ObjectOptions,
+            FileInfo = FileInfo,
+            ObjectToDelete = ObjectToDelete,
+            DeletedObject = DeletedObject,
+        >,
+{
     if buckets.is_empty() {
         debug!("No migrating bucket metadata found");
         return Ok(());

crates/ecstore/src/bucket/quota/reservation.rs

@@ -850,6 +850,11 @@ pub(crate) async fn begin(
     let snapshot_admission = opts.quota_admission;
     let data_movement = opts.data_movement;
     if crate::bucket::utils::is_meta_bucketname(bucket) {
+        let meta_bucket_fast_path_started = rustfs_io_metrics::put_stage_timer();
+        rustfs_io_metrics::record_put_object_stage_duration_from(
+            rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_META_BUCKET_FAST_PATH,
+            meta_bucket_fast_path_started,
+        );
         return Ok(QuotaContext {
             store: None,
             bucket: bucket.to_string(),
@@ -906,9 +911,63 @@ pub(crate) async fn begin(
         });
     }
 
+    let metadata_lock_started = rustfs_io_metrics::put_stage_timer();
     let metadata_guard = metadata_sys::acquire_bucket_metadata_transaction_read_lock_for_options_in(ctx, bucket, opts).await?;
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_METADATA_LOCK,
+        metadata_lock_started,
+    );
+    let cache_lookup_started = rustfs_io_metrics::put_stage_timer();
+    let cached_quota_snapshot = if rustfs_io_metrics::put_stage_metrics_enabled() {
+        match metadata_sys::get_cached_quota_config_and_incarnation_in(ctx, bucket).await {
+            Ok(Some(snapshot)) => {
+                rustfs_io_metrics::record_put_object_quota_cache_decision(
+                    "candidate",
+                    quota_cache_reason(snapshot.quota.as_ref()),
+                );
+                Some(snapshot)
+            }
+            Ok(None) => {
+                rustfs_io_metrics::record_put_object_quota_cache_decision("miss", "no_authoritative_cache");
+                None
+            }
+            Err(_) => {
+                rustfs_io_metrics::record_put_object_quota_cache_decision("miss", "invalid_cached_quota");
+                None
+            }
+        }
+    } else {
+        None
+    };
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_CACHE_LOOKUP,
+        cache_lookup_started,
+    );
+    let config_read_started = rustfs_io_metrics::put_stage_timer();
     let (quota, bucket_incarnation, quota_revision) =
         metadata_sys::get_quota_config_and_incarnation_from_disk_for_options_in(ctx, bucket, opts, &metadata_guard).await?;
+    if let Some(cached) = cached_quota_snapshot.as_ref() {
+        let reason = quota_cache_reason(quota.as_ref());
+        if cached.quota == quota && cached.bucket_incarnation == bucket_incarnation && cached.quota_revision == quota_revision {
+            rustfs_io_metrics::record_put_object_quota_cache_decision("authoritative_match", reason);
+        } else {
+            rustfs_io_metrics::record_put_object_quota_cache_decision("authoritative_mismatch", reason);
+            if cached.quota != quota {
+                rustfs_io_metrics::record_put_object_quota_cache_decision("authoritative_mismatch_field", "quota");
+            }
+            if cached.bucket_incarnation != bucket_incarnation {
+                rustfs_io_metrics::record_put_object_quota_cache_decision("authoritative_mismatch_field", "bucket_incarnation");
+            }
+            if cached.quota_revision != quota_revision {
+                rustfs_io_metrics::record_put_object_quota_cache_decision("authoritative_mismatch_field", "quota_revision");
+            }
+        }
+    }
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_CONFIG_READ,
+        config_read_started,
+    );
+    let lock_lost_check_started = rustfs_io_metrics::put_stage_timer();
     if metadata_guard.is_lock_lost() {
         return Err(StorageError::NamespaceLockQuorumUnavailable {
             mode: "quota_config",
@@ -918,6 +977,11 @@ pub(crate) async fn begin(
             achieved: 0,
         });
     }
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_LOCK_LOST_CHECK,
+        lock_lost_check_started,
+    );
+    let policy_evaluate_started = rustfs_io_metrics::put_stage_timer();
     if quota
         .as_ref()
         .is_some_and(|quota| quota.has_unsupported_reservation_protocol())
@@ -926,6 +990,11 @@ pub(crate) async fn begin(
         return Err(StorageError::PartMissingOrCorrupt);
     }
     let durable_quota = quota.as_ref().filter(|quota| quota.uses_durable_reservations());
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_POLICY_EVALUATE,
+        policy_evaluate_started,
+    );
+    let capability_proof_started = rustfs_io_metrics::put_stage_timer();
     let capability_proof = if durable_quota.is_some() {
         Some(
             crate::services::notification_sys::acquire_cross_pool_fence_fleet_proof()
@@ -934,7 +1003,12 @@ pub(crate) async fn begin(
     } else {
         None
     };
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_CAPABILITY_PROOF,
+        capability_proof_started,
+    );
     let durable_quota_limit = durable_quota.and_then(|quota| quota.quota);
+    let snapshot_admission_started = rustfs_io_metrics::put_stage_timer();
     let reservation_protocol = durable_quota.and_then(|quota| quota.reservation_protocol);
     let snapshot_admission = match quota.as_ref().filter(|quota| !quota.uses_durable_reservations()) {
         Some(quota) => match (quota.quota, snapshot_admission) {
@@ -954,6 +1028,10 @@ pub(crate) async fn begin(
         },
         None => None,
     };
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_SNAPSHOT_ADMISSION,
+        snapshot_admission_started,
+    );
     let legacy_data_movement = durable_quota_limit.is_none()
         && quota.as_ref().and_then(|quota| quota.quota).is_some()
         && snapshot_admission.is_none()
@@ -965,11 +1043,21 @@ pub(crate) async fn begin(
                 .then(|| quota.as_ref().and_then(|quota| quota.quota))
                 .flatten()
         });
+    let object_store_lookup_started = rustfs_io_metrics::put_stage_timer();
     let store = if durable_quota_limit.is_some() {
         Some(metadata_sys::object_store_in(ctx).await?)
     } else {
         None
     };
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_OBJECT_STORE_LOOKUP,
+        object_store_lookup_started,
+    );
+    let context_build_started = rustfs_io_metrics::put_stage_timer();
+    rustfs_io_metrics::record_put_object_stage_duration_from(
+        rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_QUOTA_BEGIN_CONTEXT_BUILD,
+        context_build_started,
+    );
     Ok(QuotaContext {
         store,
         bucket: bucket.to_string(),
@@ -988,6 +1076,15 @@ pub(crate) async fn begin(
     })
 }
 
+fn quota_cache_reason(quota: Option<&crate::bucket::quota::BucketQuota>) -> &'static str {
+    match quota {
+        None => "no_quota",
+        Some(quota) if quota.uses_durable_reservations() => "durable_quota",
+        Some(quota) if quota.quota.is_some() => "snapshot_quota",
+        Some(_) => "quota_without_limit",
+    }
+}
+
 fn quota_capability_error(bucket: &str, object: &str) -> StorageError {
     StorageError::NamespaceLockQuorumUnavailable {
         mode: "quota_capability",

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

@@ -448,6 +448,26 @@ fn is_version_id_format_mismatch(err: &SdkError<HeadObjectError>) -> bool {
     is_version_id_mismatch(code, raw_status)
 }
 
+/// A version-addressed HEAD rejected because the target cannot address the
+/// source version id (AWS S3: bodiless 400). Narrower than
+/// [`is_version_id_format_mismatch`]: a bodiless 403 may be a genuine
+/// `s3:GetObjectVersion` denial, so it is not accepted where a mutation or a
+/// completion verdict follows instead of another non-destructive HEAD.
+fn rejects_source_version_id(err: &SdkError<HeadObjectError>) -> bool {
+    is_version_id_format_mismatch(err) && !has_raw_status(err, 403)
+}
+
+/// Whether the target's latest version of `object` is a delete marker. For
+/// targets that mint their own version ids, which cannot be asked about the
+/// source marker id: S3 answers a versionless HEAD on such a key with 404
+/// (405 on some targets) plus `x-amz-delete-marker: true`.
+async fn latest_target_version_is_delete_marker(tgt_client: &TargetClient, object: &str) -> bool {
+    match head_object_for_worker(tgt_client, &tgt_client.bucket, object, None).await {
+        Ok(head) => head.delete_marker == Some(true),
+        Err(e) => raw_header_value(e.as_ref(), "x-amz-delete-marker").as_deref() == Some("true"),
+    }
+}
+
 async fn mark_replication_target_offline_if_needed(target_client: &Arc<TargetClient>, err: &(impl Display + ?Sized)) {
     if is_replication_target_offline_error(err) {
         ReplicationTargetStore::mark_target_offline(target_client).await;
@@ -1846,7 +1866,17 @@ async fn verify_resync_head_result(
             // only absence proves completion. All other outcomes remain
             // failed (backlog#862 / #799 B13).
             let retryable = if roi.version_purge_status.is_empty() {
-                matches!(classify_delete_marker_head_error(&err), DeleteMarkerHeadOutcome::Ambiguous)
+                match classify_delete_marker_head_error(&err) {
+                    // A target that mints its own version ids never holds the
+                    // source marker id; verify the marker on the latest version.
+                    DeleteMarkerHeadOutcome::Ambiguous => {
+                        let mints_own = rejects_source_version_id(&err)
+                            || ReplicationTargetStore::version_identity_capability(&target_client.arn)
+                                .version_addressing_unreliable();
+                        !(mints_own && latest_target_version_is_delete_marker(target_client, &roi.name).await)
+                    }
+                    DeleteMarkerHeadOutcome::Exists | DeleteMarkerHeadOutcome::Missing => false,
+                }
             } else {
                 !matches!(classify_delete_marker_head_error(&err), DeleteMarkerHeadOutcome::Missing)
             };
@@ -3567,7 +3597,16 @@ async fn replicate_delete_to_target(
         }
     }
 
-    if !is_version_purge && dobj.delete_object.delete_marker && dobj.delete_object.delete_marker_version_id.is_some() {
+    // A target that mints its own version ids never holds the source marker
+    // id, so a version-addressed pre-check cannot succeed there; the marker
+    // is created by the versionless DELETE below.
+    let version_addressable =
+        !ReplicationTargetStore::version_identity_capability(&tgt_client.arn).version_addressing_unreliable();
+    if version_addressable
+        && !is_version_purge
+        && dobj.delete_object.delete_marker
+        && dobj.delete_object.delete_marker_version_id.is_some()
+    {
         match head_object_for_worker(
             tgt_client.as_ref(),
             &tgt_client.bucket,
@@ -3576,6 +3615,11 @@ async fn replicate_delete_to_target(
         )
         .await
         {
+            // The verdict is not known yet (e.g. no PUT since restart) and the
+            // target rejected the source id: learn it as the PUT path does.
+            Err(e) if rejects_source_version_id(e.as_ref()) => {
+                ReplicationTargetStore::record_version_identity_capability(&tgt_client.arn, VersionIdentityCapability::MintsOwn);
+            }
             Ok(head) if head.delete_marker == Some(true) => {
                 // Some targets expose an existing marker as a successful HEAD
                 // with the marker bit instead of the usual 405. Treat that as
@@ -6675,6 +6719,93 @@ mod tests {
         assert!(seen[0].starts_with("HEAD "), "{seen:?}");
     }
 
+    /// AWS S3 answers a HEAD for a version id it did not mint with a bodiless
+    /// 400. On a target with no verdict yet, that teaches `MintsOwn` and the
+    /// marker is still created with one versionless DELETE.
+    #[tokio::test]
+    async fn live_delete_marker_worker_learns_mints_own_from_bodiless_400() {
+        let (endpoint, server) = spawn_scripted_target_server(2, |line| {
+            if line.starts_with("HEAD ") {
+                empty_response("400 Bad Request")
+            } else if line.starts_with("DELETE ") {
+                empty_response("204 No Content")
+            } else {
+                empty_response("500 Unexpected")
+            }
+        });
+        let target = test_target_client(endpoint);
+        register_test_target(&target).await;
+        let dobj = marker_creation_dobj(&target.arn);
+
+        let rinfo = replicate_delete_to_target(&dobj, target.clone(), None).await;
+
+        assert_eq!(rinfo.replication_status, ReplicationStatusType::Completed, "{rinfo:?}");
+        assert_eq!(
+            ReplicationTargetStore::version_identity_capability(&target.arn),
+            VersionIdentityCapability::MintsOwn
+        );
+        let seen = server.join().expect("test HTTP server should finish");
+        assert_eq!(seen.len(), 2, "{seen:?}");
+        assert!(seen[0].starts_with("HEAD "), "{seen:?}");
+        assert!(seen[1].starts_with("DELETE "), "{seen:?}");
+        assert!(!seen[1].contains("versionId="), "marker creation must remain versionless: {seen:?}");
+    }
+
+    /// A bodiless 403 on the version-addressed HEAD may be a genuine
+    /// GetObjectVersion denial on an id-adopting target. It must fail closed:
+    /// no DELETE, and no `MintsOwn` verdict cached for the ARN.
+    #[tokio::test]
+    async fn live_delete_marker_worker_keeps_bodiless_403_failure_closed() {
+        let (endpoint, server) = spawn_scripted_target_server(1, |line| {
+            if line.starts_with("HEAD ") {
+                empty_response("403 Forbidden")
+            } else if line.starts_with("DELETE ") {
+                empty_response("204 No Content")
+            } else {
+                empty_response("500 Unexpected")
+            }
+        });
+        let target = test_target_client(endpoint);
+        register_test_target(&target).await;
+        let dobj = marker_creation_dobj(&target.arn);
+
+        let rinfo = replicate_delete_to_target(&dobj, target.clone(), None).await;
+
+        assert_eq!(rinfo.replication_status, ReplicationStatusType::Failed, "{rinfo:?}");
+        assert_ne!(
+            ReplicationTargetStore::version_identity_capability(&target.arn),
+            VersionIdentityCapability::MintsOwn
+        );
+        let seen = server.join().expect("test HTTP server should finish");
+        assert_eq!(seen.len(), 1, "a denied HEAD must not trigger DELETE: {seen:?}");
+        assert!(seen[0].starts_with("HEAD "), "{seen:?}");
+    }
+
+    /// A target already known to mint its own version ids is never asked for
+    /// the source marker id: the marker is created by the versionless DELETE.
+    #[tokio::test]
+    async fn live_delete_marker_worker_skips_version_head_on_mints_own_target() {
+        let (endpoint, server) = spawn_scripted_target_server(1, |line| {
+            if line.starts_with("DELETE ") {
+                empty_response("204 No Content")
+            } else {
+                empty_response("500 Unexpected")
+            }
+        });
+        let target = test_target_client(endpoint);
+        register_test_target(&target).await;
+        ReplicationTargetStore::record_version_identity_capability(&target.arn, VersionIdentityCapability::MintsOwn);
+        let dobj = marker_creation_dobj(&target.arn);
+
+        let rinfo = replicate_delete_to_target(&dobj, target.clone(), None).await;
+
+        assert_eq!(rinfo.replication_status, ReplicationStatusType::Completed, "{rinfo:?}");
+        let seen = server.join().expect("test HTTP server should finish");
+        assert_eq!(seen.len(), 1, "no version-addressed HEAD on a MintsOwn target: {seen:?}");
+        assert!(seen[0].starts_with("DELETE "), "{seen:?}");
+        assert!(!seen[0].contains("versionId="), "marker creation must remain versionless: {seen:?}");
+    }
+
     /// A target answers `HEAD ?versionId=<delete marker>` with a bodiless
     /// 405, so the SDK reports no error code. That is the marker having
     /// propagated, not a failure (rustfs/backlog#2479).
@@ -6765,6 +6896,76 @@ mod tests {
         server.join().expect("test HTTP server should finish");
     }
 
+    /// AWS answers `HEAD ?versionId=<source marker id>` with a bodiless 400.
+    /// The marker is verified through the latest version instead: a
+    /// versionless 404 with `x-amz-delete-marker: true` proves it propagated.
+    #[tokio::test]
+    async fn resync_verification_accepts_latest_delete_marker_when_target_rejects_source_id() {
+        let (endpoint, server) = spawn_scripted_target_server(2, |line| {
+            if line.contains("versionId=") {
+                empty_response("400 Bad Request")
+            } else {
+                "HTTP/1.1 404 Not Found\r\nX-Amz-Delete-Marker: true\r\nX-Amz-Version-Id: target-marker\r\nContent-Length: 0\r\nConnection: close\r\n\r\n"
+                    .to_string()
+            }
+        });
+        let target = test_target_client(endpoint);
+        let roi = delete_marker_roi();
+        let mut st = TargetReplicationResyncStatus::default();
+
+        let head_result =
+            head_object_for_worker(target.as_ref(), &target.bucket, &roi.name, roi.version_id.map(|v| v.to_string())).await;
+        let (size, err) = verify_resync_head_result(head_result, &roi, &mut st, &target).await;
+
+        assert!(err.is_none(), "{err:?}");
+        assert_eq!((size, st.replicated_count, st.failed_count), (0, 1, 0));
+        let seen = server.join().expect("test HTTP server should finish");
+        assert!(!seen[1].contains("versionId="), "the re-check must address the latest version: {seen:?}");
+    }
+
+    /// A live latest version on a target that rejects the source id means the
+    /// marker has not propagated: the verification stays failed.
+    #[tokio::test]
+    async fn resync_verification_fails_delete_marker_when_latest_target_version_is_live() {
+        let (endpoint, server) = spawn_scripted_target_server(2, |line| {
+            if line.contains("versionId=") {
+                empty_response("400 Bad Request")
+            } else {
+                empty_response("200 OK")
+            }
+        });
+        let target = test_target_client(endpoint);
+        let roi = delete_marker_roi();
+        let mut st = TargetReplicationResyncStatus::default();
+
+        let head_result =
+            head_object_for_worker(target.as_ref(), &target.bucket, &roi.name, roi.version_id.map(|v| v.to_string())).await;
+        let (size, err) = verify_resync_head_result(head_result, &roi, &mut st, &target).await;
+
+        assert!(err.is_some(), "a live latest version must not verify the marker");
+        assert_eq!((size, st.replicated_count, st.failed_count), (0, 0, 1));
+        server.join().expect("test HTTP server should finish");
+    }
+
+    /// A bodiless 403 may be a genuine permission denial: it is not evidence
+    /// that the target mints its own ids, so no latest-version re-check runs.
+    #[tokio::test]
+    async fn resync_verification_keeps_delete_marker_failed_on_bodiless_403() {
+        let (endpoint, server) = spawn_scripted_target_server(1, |_| empty_response("403 Forbidden"));
+        let target = test_target_client(endpoint);
+        let roi = delete_marker_roi();
+        let mut st = TargetReplicationResyncStatus::default();
+
+        let head_result =
+            head_object_for_worker(target.as_ref(), &target.bucket, &roi.name, roi.version_id.map(|v| v.to_string())).await;
+        let (size, err) = verify_resync_head_result(head_result, &roi, &mut st, &target).await;
+
+        assert!(err.is_some(), "a 403 must not count as a propagated delete marker");
+        assert_eq!((size, st.replicated_count, st.failed_count), (0, 0, 1));
+        let seen = server.join().expect("test HTTP server should finish");
+        assert_eq!(seen.len(), 1, "{seen:?}");
+    }
+
     /// Serves exactly `requests` connections, answering each from
     /// `respond(request_line)`, and returns the request lines it saw. Reads
     /// the whole request (headers plus `Content-Length` body, honoring

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

@@ -24,10 +24,11 @@ use crate::cluster::rpc::{gen_tonic_replay_scope_headers, gen_tonic_signature_he
 #[cfg(test)]
 use crate::cluster::rpc::{tonic_boot_epoch_challenge, tonic_boot_epoch_response_headers};
 use crate::disk::error::{DiskError, Error as DiskErrorType, RpcStatusError};
+#[cfg(test)]
 use crate::runtime::sources as runtime_sources;
 use http::{Request as HttpRequest, Response as HttpResponse, Uri};
 use rustfs_protos::{
-    ChannelClass, create_new_channel, get_channel_for_class,
+    ChannelClass, get_channel_for_class,
     proto_gen::node_service::{
         heal_control_service_client::HealControlServiceClient, node_service_client::NodeServiceClient,
         scanner_control_service_client::ScannerControlServiceClient,
@@ -53,7 +54,7 @@ use super::context_propagation::{inject_request_id_into_metadata, inject_trace_c
 /// 3. Subsequent calls will attempt fresh connections
 /// 4. If node is still down, connection will fail fast (3s timeout)
 pub async fn node_service_time_out_client(
-    addr: &String,
+    addr: &str,
     interceptor: TonicInterceptor,
 ) -> Result<NodeServiceClient<InterceptedService<AuthenticatedChannel, TonicInterceptor>>, Box<dyn Error>> {
     // Default to the latency-sensitive control channel; bulk `bytes` RPCs opt in via the
@@ -66,12 +67,9 @@ pub(crate) async fn scanner_control_time_out_client(
     interceptor: TonicInterceptor,
 ) -> crate::error::Result<ScannerControlServiceClient<InterceptedService<AuthenticatedChannel, TonicInterceptor>>> {
     let interceptor = interceptor.with_rpc_audience(addr)?;
-    let channel = match runtime_sources::cached_node_channel(addr).await {
-        Some(channel) => channel,
-        None => create_new_channel(addr)
-            .await
-            .map_err(|err| crate::error::Error::other(err.to_string()))?,
-    };
+    let channel = get_channel_for_class(addr, ChannelClass::Control)
+        .await
+        .map_err(|err| crate::error::Error::other(err.to_string()))?;
     let channel = ReplayScopeChannel::new(channel, interceptor.replay_scope_audience());
     let limit = rustfs_protos::scoped_dirty_usage::SCOPED_DIRTY_USAGE_MAX_REQUEST_BYTES as usize;
     Ok(ScannerControlServiceClient::with_interceptor(channel, interceptor)
@@ -84,10 +82,7 @@ pub async fn heal_control_time_out_client(
     interceptor: TonicInterceptor,
 ) -> Result<HealControlServiceClient<InterceptedService<AuthenticatedChannel, TonicInterceptor>>, Box<dyn Error>> {
     let interceptor = interceptor.with_rpc_audience(addr)?;
-    let channel = match runtime_sources::cached_node_channel(addr).await {
-        Some(channel) => channel,
-        None => create_new_channel(addr).await?,
-    };
+    let channel = get_channel_for_class(addr, ChannelClass::Control).await?;
     let max_message_size = rustfs_protos::HEAL_CONTROL_RPC_MAX_MESSAGE_SIZE;
     let channel = ReplayScopeChannel::new(channel, interceptor.replay_scope_audience());
     Ok(HealControlServiceClient::with_interceptor(channel, interceptor)
@@ -100,10 +95,7 @@ pub async fn tier_mutation_control_time_out_client(
     interceptor: TonicInterceptor,
 ) -> Result<TierMutationControlServiceClient<InterceptedService<AuthenticatedChannel, TonicInterceptor>>, Box<dyn Error>> {
     let interceptor = interceptor.with_rpc_audience(addr)?;
-    let channel = match runtime_sources::cached_node_channel(addr).await {
-        Some(channel) => channel,
-        None => create_new_channel(addr).await?,
-    };
+    let channel = get_channel_for_class(addr, ChannelClass::Control).await?;
     let max_message_size = rustfs_protos::TIER_MUTATION_RPC_MAX_MESSAGE_SIZE;
     let channel = ReplayScopeChannel::new(channel, interceptor.replay_scope_audience());
     Ok(TierMutationControlServiceClient::with_interceptor(channel, interceptor)
@@ -118,22 +110,12 @@ pub async fn tier_mutation_control_time_out_client(
 /// from lock/health RPCs; everything else uses [`ChannelClass::Control`]. When isolation is
 /// disabled the two classes resolve to the same cached channel, i.e. legacy behavior.
 pub async fn node_service_time_out_client_for_class(
-    addr: &String,
+    addr: &str,
     interceptor: TonicInterceptor,
     class: ChannelClass,
 ) -> Result<NodeServiceClient<InterceptedService<AuthenticatedChannel, TonicInterceptor>>, Box<dyn Error>> {
     let interceptor = interceptor.with_rpc_audience(addr)?;
-    let channel = match class {
-        ChannelClass::Control => match runtime_sources::cached_node_channel(addr).await {
-            Some(channel) => {
-                debug!("Using cached gRPC channel for: {}", addr);
-                channel
-            }
-            // No cached connection, create new one.
-            None => create_new_channel(addr).await?,
-        },
-        ChannelClass::Bulk => get_channel_for_class(addr, ChannelClass::Bulk).await?,
-    };
+    let channel = get_channel_for_class(addr, class).await?;
 
     let max_message_size = rustfs_protos::internode_rpc_max_message_size();
     let channel = ReplayScopeChannel::new(channel, interceptor.replay_scope_audience());
@@ -142,8 +124,24 @@ pub async fn node_service_time_out_client_for_class(
         .max_encoding_message_size(max_message_size))
 }
 
+pub(crate) async fn node_service_delete_recovery_client(
+    addr: &str,
+    interceptor: TonicInterceptor,
+    timeout: std::time::Duration,
+) -> Result<NodeServiceClient<InterceptedService<AuthenticatedChannel, TonicInterceptor>>, DiskError> {
+    let interceptor = interceptor.with_rpc_audience(addr).map_err(DiskError::from)?;
+    let channel = rustfs_protos::create_channel_with_timeout(addr, timeout)
+        .await
+        .map_err(|error| DiskError::RemoteClientUnavailable(error.to_string()))?;
+    let channel = ReplayScopeChannel::new(channel, interceptor.replay_scope_audience());
+    let max_message_size = rustfs_protos::internode_rpc_max_message_size();
+    Ok(NodeServiceClient::with_interceptor(channel, interceptor)
+        .max_decoding_message_size(max_message_size)
+        .max_encoding_message_size(max_message_size))
+}
+
 pub async fn node_service_time_out_client_no_auth(
-    addr: &String,
+    addr: &str,
 ) -> Result<NodeServiceClient<InterceptedService<AuthenticatedChannel, TonicInterceptor>>, Box<dyn Error>> {
     node_service_time_out_client(addr, TonicInterceptor::NoOp(NoOpInterceptor)).await
 }
@@ -355,6 +353,13 @@ fn apply_peer_replay_response(
     }
 }
 
+fn rename_data_grpc_stage(path: &str) -> bool {
+    matches!(
+        path,
+        "/node_service.NodeService/RenameData" | "/node_service.NodeService/RenameDataAtIncarnation"
+    )
+}
+
 impl<S, ReqBody, ResBody> Service<HttpRequest<ReqBody>> for ReplayScopeChannel<S>
 where
     S: Service<HttpRequest<ReqBody>, Response = HttpResponse<ResBody>>,
@@ -372,6 +377,12 @@ where
     }
 
     fn call(&mut self, mut request: HttpRequest<ReqBody>) -> Self::Future {
+        let observe_rename_data = rustfs_io_metrics::put_stage_metrics_enabled() && rename_data_grpc_stage(request.uri().path());
+        let request_scope_started = if observe_rename_data {
+            rustfs_io_metrics::put_stage_timer()
+        } else {
+            None
+        };
         let authenticated = self.audience.as_ref().is_some_and(|_| {
             request
                 .headers()
@@ -406,11 +417,30 @@ where
                 }
             }
         }
+        rustfs_io_metrics::record_put_object_stage_duration_from(
+            rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_REQUEST_SCOPE,
+            request_scope_started,
+        );
 
         let audience = self.audience.clone();
         let future = self.inner.call(request);
         Box::pin(async move {
-            let response = future.await?;
+            let response = if observe_rename_data {
+                rustfs_io_metrics::observe_put_stage_future(
+                    future,
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_TRANSPORT_CALL,
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_TRANSPORT_CALL_POLL_PENDING_COUNT,
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_TRANSPORT_CALL_FIRST_PENDING_TO_READY,
+                )
+                .await?
+            } else {
+                future.await?
+            };
+            let replay_response_started = if observe_rename_data {
+                rustfs_io_metrics::put_stage_timer()
+            } else {
+                None
+            };
             if let (Some(audience), Some(challenge)) = (audience, challenge) {
                 let response_state = verify_tonic_peer_replay_capabilities_response(&audience, challenge, response.headers());
                 if let Err(error) = &response_state
@@ -428,6 +458,10 @@ where
                 }
                 apply_peer_replay_response(audience, sent_state, response_state);
             }
+            rustfs_io_metrics::record_put_object_stage_duration_from(
+                rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_REPLAY_RESPONSE,
+                replay_response_started,
+            );
             Ok(response)
         })
     }
@@ -546,6 +580,126 @@ mod tests {
     use tracing_opentelemetry::OpenTelemetrySpanExt;
     use tracing_subscriber::{Registry, layer::SubscriberExt};
 
+    #[tokio::test]
+    async fn delete_recovery_channel_uses_its_scoped_deadline() {
+        use bytes::Bytes;
+        use http_body_util::StreamBody;
+        use hyper::{body::Frame, service::service_fn};
+        use hyper_util::rt::{TokioExecutor, TokioIo};
+        use rustfs_protos::proto_gen::node_service::{RecoverDeleteTransactionRequest, RecoverDeleteTransactionResponse};
+        use tonic_prost::prost::Message;
+        let listener = tokio::net::TcpListener::bind("127.0.0.1:0")
+            .await
+            .expect("bind recovery peer");
+        let addr = format!("http://{}", listener.local_addr().expect("peer address"));
+        let server = tokio::spawn(async move {
+            loop {
+                let (stream, _) = listener.accept().await.expect("accept recovery connection");
+                tokio::spawn(async move {
+                    let service = service_fn(|_request: HttpRequest<hyper::body::Incoming>| async {
+                        tokio::time::sleep(std::time::Duration::from_millis(500)).await;
+                        let payload = RecoverDeleteTransactionResponse {
+                            success: true,
+                            protocol_version: 1,
+                            result: Bytes::new(),
+                            error: None,
+                        }
+                        .encode_to_vec();
+                        let mut data = vec![0];
+                        data.extend_from_slice(&u32::try_from(payload.len()).expect("small response").to_be_bytes());
+                        data.extend_from_slice(&payload);
+                        let mut trailers = http::HeaderMap::new();
+                        trailers.insert("grpc-status", "0".parse().expect("status"));
+                        let frames: Vec<Result<Frame<Bytes>, std::convert::Infallible>> =
+                            vec![Ok(Frame::data(data.into())), Ok(Frame::trailers(trailers))];
+                        let mut response = HttpResponse::new(StreamBody::new(futures::stream::iter(frames)));
+                        response
+                            .headers_mut()
+                            .insert("content-type", "application/grpc".parse().expect("content type"));
+                        Ok::<_, std::convert::Infallible>(response)
+                    });
+                    let _ = hyper::server::conn::http2::Builder::new(TokioExecutor::new())
+                        .serve_connection(TokioIo::new(stream), service)
+                        .await;
+                });
+            }
+        });
+        let request = || RecoverDeleteTransactionRequest {
+            disk: "disk".into(),
+            volume: "bucket".into(),
+            protocol_version: 1,
+            operation: Bytes::new(),
+        };
+        // Seed an existing short-lived control channel to reproduce the inherited deadline.
+        let control = tonic::transport::Endpoint::from_shared(addr.clone())
+            .expect("test endpoint")
+            .timeout(std::time::Duration::from_millis(50))
+            .connect()
+            .await
+            .expect("control connect");
+        rustfs_common::cache_connection(addr.clone(), control).await;
+        let mut foreground = node_service_time_out_client(&addr, TonicInterceptor::NoOp(NoOpInterceptor))
+            .await
+            .expect("control client");
+        let error = foreground
+            .recover_delete_transaction(request())
+            .await
+            .expect_err("short control deadline");
+        assert!(matches!(error.code(), tonic::Code::Cancelled | tonic::Code::DeadlineExceeded));
+        // The 2-second test override bypasses the production range parser, which rejects 2.
+        let mut recovery = node_service_delete_recovery_client(
+            &addr,
+            TonicInterceptor::NoOp(NoOpInterceptor),
+            std::time::Duration::from_secs(2),
+        )
+        .await
+        .expect("scoped recovery client");
+        assert!(
+            recovery
+                .recover_delete_transaction(request())
+                .await
+                .expect("scoped deadline must survive control deadline")
+                .into_inner()
+                .success
+        );
+        server.abort();
+        rustfs_protos::evict_failed_connection(&addr).await;
+    }
+
+    #[tokio::test]
+    async fn rpc_client_factories_do_not_reuse_https_channels_without_tls_generation() {
+        let listener = std::net::TcpListener::bind("127.0.0.1:0").expect("reserve local address");
+        let addr = format!("https://{}", listener.local_addr().expect("local address"));
+        drop(listener);
+        for factory in 0..4 {
+            let placeholder = tonic::transport::Endpoint::from_static("http://127.0.0.1:1").connect_lazy();
+            runtime_sources::cache_test_node_channel(addr.clone(), placeholder).await;
+            let interceptor = TonicInterceptor::NoOp(NoOpInterceptor);
+            let _result = match factory {
+                0 => scanner_control_time_out_client(&addr, interceptor)
+                    .await
+                    .map(|_| ())
+                    .map_err(|e| e.to_string()),
+                1 => heal_control_time_out_client(&addr, interceptor)
+                    .await
+                    .map(|_| ())
+                    .map_err(|e| e.to_string()),
+                2 => tier_mutation_control_time_out_client(&addr, interceptor)
+                    .await
+                    .map(|_| ())
+                    .map_err(|e| e.to_string()),
+                _ => node_service_time_out_client(&addr, interceptor)
+                    .await
+                    .map(|_| ())
+                    .map_err(|e| e.to_string()),
+            };
+            assert!(
+                !runtime_sources::test_node_channel_is_cached(&addr).await,
+                "factory {factory} must evict an unbound HTTPS cache entry before attempting a fresh dial"
+            );
+        }
+    }
+
     #[derive(Clone)]
     struct EpochProofService {
         audience: String,

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

@@ -289,13 +289,9 @@ async fn client(
     >,
     NetworkPeerProbeError,
 > {
-    node_service_time_out_client_for_class(
-        &address.to_owned(),
-        TonicInterceptor::Signature(gen_tonic_signature_interceptor()),
-        class,
-    )
-    .await
-    .map_err(|_| NetworkPeerProbeError::Unreachable)
+    node_service_time_out_client_for_class(address, TonicInterceptor::Signature(gen_tonic_signature_interceptor()), class)
+        .await
+        .map_err(|_| NetworkPeerProbeError::Unreachable)
 }
 
 fn ping_request(payload: &[u8]) -> PingRequest {
@@ -329,6 +325,24 @@ fn map_status(status: tonic::Status) -> NetworkPeerProbeError {
     match status.code() {
         Code::Unavailable | Code::Unknown => NetworkPeerProbeError::Unreachable,
         Code::DeadlineExceeded => NetworkPeerProbeError::TimedOut,
+        Code::Cancelled | Code::Internal => {
+            let mut source = std::error::Error::source(&status);
+            while let Some(error) = source {
+                if error
+                    .downcast_ref::<h2::Error>()
+                    .is_some_and(|error| matches!(error.reason(), Some(h2::Reason::CANCEL | h2::Reason::NO_ERROR)))
+                {
+                    return NetworkPeerProbeError::Unreachable;
+                }
+                source = if let Some(error) = error.downcast_ref::<std::io::Error>() {
+                    // io::Error::source skips its immediate custom error.
+                    error.get_ref().map(|inner| inner as &(dyn std::error::Error + 'static))
+                } else {
+                    error.source()
+                };
+            }
+            NetworkPeerProbeError::ProtocolFailure
+        }
         _ => NetworkPeerProbeError::ProtocolFailure,
     }
 }
@@ -339,6 +353,35 @@ mod tests {
     use crate::layout::endpoint::Endpoint;
     use crate::layout::endpoints::{Endpoints, PoolEndpoints};
 
+    #[test]
+    fn peer_transport_resets_are_unreachable_without_reclassifying_application_errors() {
+        for reason in [h2::Reason::CANCEL, h2::Reason::NO_ERROR] {
+            let status = tonic::Status::from_error(Box::new(h2::Error::from(reason)));
+            assert!(matches!(status.code(), Code::Cancelled | Code::Internal));
+            let code = status.code();
+            assert_eq!(map_status(status), NetworkPeerProbeError::Unreachable);
+            let mut wrapped = tonic::Status::new(code, "wrapped transport reset");
+            wrapped.set_source(Arc::new(std::io::Error::other(h2::Error::from(reason))));
+            assert_eq!(map_status(wrapped), NetworkPeerProbeError::Unreachable);
+        }
+        for reason in [h2::Reason::PROTOCOL_ERROR, h2::Reason::COMPRESSION_ERROR] {
+            let status = tonic::Status::from_error(Box::new(h2::Error::from(reason)));
+            assert_eq!(map_status(status), NetworkPeerProbeError::ProtocolFailure);
+        }
+        for status in [
+            tonic::Status::cancelled("peer cancelled the application request"),
+            tonic::Status::internal("peer rejected the probe"),
+            tonic::Status::unauthenticated("authentication rejected"),
+            tonic::Status::permission_denied("authorization rejected"),
+        ] {
+            assert_eq!(map_status(status), NetworkPeerProbeError::ProtocolFailure);
+        }
+        assert_eq!(
+            map_status(tonic::Status::deadline_exceeded("probe deadline elapsed")),
+            NetworkPeerProbeError::TimedOut
+        );
+    }
+
     fn pacing(duration: Duration) -> Arc<DiagnosticPacing> {
         NetworkPeerProbeClient::from_endpoint_pools(&topology())
             .with_diagnostic_pacing(Instant::now(), duration)

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

@@ -14,7 +14,8 @@
 
 use crate::cluster::rpc::client::{
     AuthenticatedChannel, TonicInterceptor, gen_tonic_signature_interceptor, is_network_like_disk_error,
-    node_service_time_out_client, node_service_time_out_client_for_class, node_service_time_out_client_no_auth,
+    node_service_delete_recovery_client, node_service_time_out_client, node_service_time_out_client_for_class,
+    node_service_time_out_client_no_auth,
 };
 use crate::cluster::rpc::internode_data_transport::{
     InternodeDataTransport, NsScannerCapabilityRequest, NsScannerStreamRequest, ReadStreamRequest, WalkDirStreamRequest,
@@ -22,10 +23,11 @@ use crate::cluster::rpc::internode_data_transport::{
 };
 use crate::disk::error::{Error, Result};
 use crate::disk::{
-    BatchReadVersionReq, BatchReadVersionResp, CheckPartsResp, ConditionalFileUpdate, DeleteOptions, DiskAPI, DiskInfo,
-    DiskInfoOptions, DiskLocation, DiskOption, FileInfoVersions, FileReader, FileWriter, PartTransactionAction, ReadMultipleReq,
-    ReadMultipleResp, ReadOptions, RenameDataResp, SnapshotLeaseToken, UpdateMetadataOpts, VolumeInfo, WalkDirOptions,
-    batch_read_version_one_by_one,
+    BatchReadVersionReq, BatchReadVersionResp, CheckPartsResp, ConditionalFileUpdate, DELETE_RECOVERY_PROTOCOL_VERSION,
+    DeleteOptions, DeleteRecoveryAction, DeleteRecoveryCursor, DeleteRecoveryDigest, DeleteRecoveryInspection,
+    DeleteRecoveryPage, DeleteRecoveryRpcOperation, DeleteRecoveryRpcReply, DiskAPI, DiskInfo, DiskInfoOptions, DiskLocation,
+    DiskOption, FileInfoVersions, FileReader, FileWriter, PartTransactionAction, ReadMultipleReq, ReadMultipleResp, ReadOptions,
+    RenameDataResp, SnapshotLeaseToken, UpdateMetadataOpts, VolumeInfo, WalkDirOptions, batch_read_version_one_by_one,
     disk_store::{
         DEFAULT_RUSTFS_DRIVE_ACTIVE_MONITORING, ENV_RUSTFS_DRIVE_ACTIVE_MONITORING, SKIP_IF_SUCCESS_BEFORE,
         get_drive_active_check_interval, get_drive_active_check_timeout, get_drive_disk_info_timeout, get_drive_list_dir_timeout,
@@ -59,10 +61,10 @@ use rustfs_protos::proto_gen::node_service::{
     CompareAndUpdateFileRequest, DeletePathsRequest, DeleteRequest, DeleteVersionRequest, DeleteVersionsRequest,
     DeleteVersionsResponse, DeleteVolumeRequest, DiskInfoRequest, ListDirRequest, ListVolumesRequest, MakeVolumeRequest,
     MakeVolumesRequest, PreparePartTransactionRequest, ReadAllRequest, ReadMetadataRequest, ReadMultipleRequest,
-    ReadMultipleResponse, ReadPartsRequest, ReadVersionRequest, ReadXlRequest, RenameDataRequest, RenameFileRequest,
-    SettlePartTransactionRequest, SnapshotLeaseReleaseRequest, SnapshotLeaseRenewRequest, SnapshotLeaseRequest,
-    SnapshotLeaseResponse, StatVolumeRequest, UpdateMetadataRequest, VerifyFileRequest, WriteAllRequest, WriteMetadataRequest,
-    node_service_client::NodeServiceClient,
+    ReadMultipleResponse, ReadPartsRequest, ReadVersionRequest, ReadXlRequest, RecoverDeleteTransactionRequest,
+    RenameDataRequest, RenameFileRequest, SettlePartTransactionRequest, SnapshotLeaseReleaseRequest, SnapshotLeaseRenewRequest,
+    SnapshotLeaseRequest, SnapshotLeaseResponse, StatVolumeRequest, UpdateMetadataRequest, VerifyFileRequest, WriteAllRequest,
+    WriteMetadataRequest, node_service_client::NodeServiceClient,
 };
 use serde::{Serialize, de::DeserializeOwned};
 use std::{
@@ -142,6 +144,45 @@ fn decode_delete_versions_errors(response: DeleteVersionsResponse, expected_len:
         .collect()
 }
 
+fn decode_delete_recovery_batch_errors(codes: Vec<Option<u32>>, expected_len: usize) -> Result<Vec<Option<DiskError>>> {
+    if codes.len() != expected_len {
+        return Err(Error::other("malformed recoverable batch delete response length"));
+    }
+    codes
+        .into_iter()
+        .map(|code| match code {
+            None => Ok(None),
+            Some(code) => DiskError::from_u32(code)
+                .map(Some)
+                .ok_or_else(|| Error::other("invalid recoverable batch delete error code")),
+        })
+        .collect()
+}
+
+fn validate_delete_recovery_inspection(
+    volume: &str,
+    object: &str,
+    transaction: Uuid,
+    inspection: &DeleteRecoveryInspection,
+) -> Result<()> {
+    if inspection.protocol_version != DELETE_RECOVERY_PROTOCOL_VERSION
+        || inspection.id.object != object
+        || inspection.id.transaction != transaction
+        || transaction.is_nil()
+    {
+        return Err(Error::other("delete recovery inspection does not match request"));
+    }
+    if let Some(journal) = &inspection.journal
+        && (journal.protocol_version != DELETE_RECOVERY_PROTOCOL_VERSION
+            || journal.volume != volume
+            || journal.object != object
+            || journal.transaction != transaction)
+    {
+        return Err(Error::other("delete recovery journal does not match request"));
+    }
+    Ok(())
+}
+
 fn snapshot_lease_token_from_response(response: SnapshotLeaseResponse) -> Result<SnapshotLeaseToken> {
     if !response.success {
         return Err(response.error.unwrap_or_default().into());
@@ -858,6 +899,75 @@ fn spawn_control_channel_prewarm(addr: String) {
 }
 
 impl RemoteDisk {
+    async fn recover_delete_transaction(
+        &self,
+        volume: &str,
+        operation: DeleteRecoveryRpcOperation,
+    ) -> Result<DeleteRecoveryRpcReply> {
+        let payload = rmp_serde::to_vec_named(&operation)?;
+        if payload.len() > operation.max_wire_payload_bytes() {
+            return Err(Error::other("delete recovery operation exceeds payload limit"));
+        }
+        let settle_timeout = if matches!(&operation, DeleteRecoveryRpcOperation::Settle { .. }) {
+            Some(crate::disk::get_delete_recovery_timeout()?)
+        } else {
+            None
+        };
+        self.execute_with_timeout(
+            || async {
+                let mut client = if let Some(timeout) = settle_timeout {
+                    if let Some(error) = self.offline_bypass_error() {
+                        return Err(error);
+                    }
+                    node_service_delete_recovery_client(
+                        &self.addr,
+                        TonicInterceptor::Signature(gen_tonic_signature_interceptor()),
+                        timeout,
+                    )
+                    .await?
+                } else {
+                    self.get_client().await?
+                };
+                let mut request = Request::new(RecoverDeleteTransactionRequest {
+                    disk: self.endpoint.to_string(),
+                    volume: volume.to_owned(),
+                    protocol_version: DELETE_RECOVERY_PROTOCOL_VERSION,
+                    operation: payload.into(),
+                });
+                let canonical_body = rustfs_protos::canonical_recover_delete_transaction_request_body(request.get_ref())
+                    .map_err(|_| Error::other("recover_delete_transaction request length cannot be represented"))?;
+                crate::cluster::rpc::set_tonic_canonical_body_digest(&mut request, &canonical_body).map_err(Error::other)?;
+                // UNIMPLEMENTED is deliberately propagated: older peers cannot settle transactions.
+                let response = client
+                    .recover_delete_transaction(request)
+                    .await
+                    .map_err(|status| {
+                        if status.code() == Code::Unimplemented {
+                            DiskError::UnsupportedDisk
+                        } else {
+                            status.into()
+                        }
+                    })?
+                    .into_inner();
+                if response.protocol_version != DELETE_RECOVERY_PROTOCOL_VERSION {
+                    return Err(DiskError::UnsupportedDisk);
+                }
+                if !response.success {
+                    return Err(response
+                        .error
+                        .ok_or_else(|| Error::other("missing delete recovery error"))?
+                        .into());
+                }
+                if response.error.is_some() {
+                    return Err(Error::other("malformed delete recovery success"));
+                }
+                rustfs_protos::decode_delete_recovery_payload(&response.result).map_err(Into::into)
+            },
+            settle_timeout.unwrap_or_else(get_max_timeout_duration),
+        )
+        .await
+    }
+
     async fn rename_file_with_durability(
         &self,
         src_volume: &str,
@@ -1920,6 +2030,12 @@ fn encode_msgpack<T: Serialize>(value: &T) -> Result<Vec<u8>> {
     encode_msgpack_with_capacity(value, MSGPACK_ENCODE_CAPACITY_HINT)
 }
 
+// Optional fields are omitted by serde; named maps keep rollback flags and
+// transaction UUIDs independent of field order across peer versions.
+fn encode_delete_options_msgpack(value: &DeleteOptions) -> Result<Vec<u8>> {
+    Ok(rmp_serde::to_vec_named(value)?)
+}
+
 fn encode_file_info_msgpack(value: &FileInfo) -> Result<Vec<u8>> {
     encode_msgpack_with_capacity(value, FILE_INFO_MSGPACK_ENCODE_CAPACITY_HINT)
 }
@@ -2163,6 +2279,7 @@ impl RemoteDisk {
         self.execute_with_timeout_for_op(
             "rename_data",
             || async {
+                let prepare_started = rustfs_io_metrics::put_stage_timer();
                 let file_info = compat_json(fi)?;
                 let file_info_bin = encode_file_info_msgpack(fi)?;
                 let mut client = self.get_client().await?;
@@ -2190,24 +2307,67 @@ impl RemoteDisk {
                 } else {
                     attach_mutation_body_digest(&mut request, canonical_body, "rename_data")?;
                 }
+                rustfs_io_metrics::record_put_object_stage_duration_from(
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_PREPARE,
+                    prepare_started,
+                );
 
-                let response = if incarnation_bound {
-                    // Older peers return Unimplemented before mutation; never downgrade.
-                    client.rename_data_at_incarnation(request).await?
-                } else {
-                    client.rename_data(request).await?
-                }
-                .into_inner();
+                let rpc_started = rustfs_io_metrics::put_stage_timer();
+                let response = rustfs_io_metrics::observe_put_stage_future(
+                    async {
+                        if incarnation_bound {
+                            // Older peers return Unimplemented before mutation; never downgrade.
+                            client.rename_data_at_incarnation(request).await
+                        } else {
+                            client.rename_data(request).await
+                        }
+                    },
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_RPC_AWAIT,
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_RPC_AWAIT_POLL_PENDING_COUNT,
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_RPC_AWAIT_FIRST_PENDING_TO_READY,
+                )
+                .await?;
+                let response_process_started = rustfs_io_metrics::put_stage_timer();
+                let response_headers_started = rustfs_io_metrics::put_stage_timer();
+                let _ = response.metadata().len();
+                rustfs_io_metrics::record_put_object_stage_duration_from(
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_RESPONSE_HEADERS,
+                    response_headers_started,
+                );
+                let into_inner_started = rustfs_io_metrics::put_stage_timer();
+                let response = response.into_inner();
+                rustfs_io_metrics::record_put_object_stage_duration_from(
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_INTO_INNER,
+                    into_inner_started,
+                );
+                rustfs_io_metrics::record_put_object_stage_duration_from(
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_RPC,
+                    rpc_started,
+                );
 
+                let success_check_started = rustfs_io_metrics::put_stage_timer();
                 if !response.success {
                     return Err(response.error.unwrap_or_default().into());
                 }
+                rustfs_io_metrics::record_put_object_stage_duration_from(
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_SUCCESS_CHECK,
+                    success_check_started,
+                );
 
+                let decode_started = rustfs_io_metrics::put_stage_timer();
                 let rename_data_resp = decode_msgpack_or_json::<RenameDataResp>(
                     &response.rename_data_resp_bin,
                     &response.rename_data_resp,
                     "RenameDataResp",
                 )?;
+                rustfs_io_metrics::record_put_object_stage_duration_from(
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_DECODE,
+                    decode_started,
+                );
+                rustfs_io_metrics::record_put_object_stage_duration_from(
+                    rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_REMOTE_CLIENT_RESPONSE_PROCESS,
+                    response_process_started,
+                );
 
                 Ok(rename_data_resp)
             },
@@ -2566,7 +2726,7 @@ impl DiskAPI for RemoteDisk {
                 // `_bin` support for DeleteVersion is new (grpc-optimization P2); always dual-write
                 // JSON + msgpack until its fallback counter has read zero across a release window.
                 let file_info_bin = encode_file_info_msgpack(&fi)?;
-                let opts_bin = encode_msgpack(&opts)?;
+                let opts_bin = encode_delete_options_msgpack(&opts)?;
                 let conditional_marker = opts.expected_delete_marker.is_some();
                 let file_info = serde_json::to_string(&fi)?;
                 let opts = serde_json::to_string(&opts)?;
@@ -2638,7 +2798,7 @@ impl DiskAPI for RemoteDisk {
 
         // `_bin` support for DeleteVersions is new (grpc-optimization P2); always dual-write JSON +
         // msgpack until its fallback counter has read zero across a release window.
-        let opts_bin = match encode_msgpack(&opts) {
+        let opts_bin = match encode_delete_options_msgpack(&opts) {
             Ok(opts_bin) => opts_bin,
             Err(err) => {
                 let mut errors = Vec::with_capacity(versions.len());
@@ -3617,6 +3777,138 @@ impl DiskAPI for RemoteDisk {
     }
 
     #[tracing::instrument(level = "trace", skip_all)]
+    async fn delete_version_recoverable(
+        &self,
+        volume: &str,
+        object: &str,
+        file_info: FileInfo,
+        force_del_marker: bool,
+        opts: DeleteOptions,
+    ) -> Result<()> {
+        match self
+            .recover_delete_transaction(
+                volume,
+                DeleteRecoveryRpcOperation::ApplyVersion {
+                    object: object.to_owned(),
+                    file_info: Box::new(file_info),
+                    force_del_marker,
+                    opts,
+                },
+            )
+            .await?
+        {
+            DeleteRecoveryRpcReply::AppliedVersion => Ok(()),
+            _ => Err(Error::other("unexpected recoverable delete reply")),
+        }
+    }
+
+    async fn delete_versions_recoverable(
+        &self,
+        volume: &str,
+        versions: Vec<FileInfoVersions>,
+        opts: DeleteOptions,
+    ) -> Vec<Option<DiskError>> {
+        let expected_len = versions.len();
+        let result = match self
+            .recover_delete_transaction(volume, DeleteRecoveryRpcOperation::ApplyVersions { versions, opts })
+            .await
+        {
+            Ok(DeleteRecoveryRpcReply::AppliedVersions(codes)) => decode_delete_recovery_batch_errors(codes, expected_len),
+            Ok(_) => Err(Error::other("unexpected recoverable batch delete reply")),
+            Err(error) => Err(error),
+        };
+        match result {
+            Ok(errors) => errors,
+            Err(error) => vec![Some(error); expected_len],
+        }
+    }
+
+    async fn discover_delete_transactions(
+        &self,
+        volume: &str,
+        object: Option<&str>,
+        cursor: Option<DeleteRecoveryCursor>,
+        limit: u32,
+    ) -> Result<DeleteRecoveryPage> {
+        match self
+            .recover_delete_transaction(
+                volume,
+                DeleteRecoveryRpcOperation::Discover {
+                    object: object.map(str::to_owned),
+                    cursor,
+                    limit,
+                },
+            )
+            .await?
+        {
+            DeleteRecoveryRpcReply::Discovered(page) => {
+                if object.is_some_and(|object| page.transactions.iter().any(|id| id.object != object)) {
+                    return Err(Error::other("delete recovery discovery escaped requested object"));
+                }
+                if page.transactions.len() > usize::try_from(limit).map_err(|_| Error::other("invalid discovery limit"))?
+                    || page
+                        .transactions
+                        .iter()
+                        .any(|id| id.transaction.is_nil() || id.object.is_empty())
+                {
+                    return Err(Error::other("malformed delete discovery page"));
+                }
+                Ok(page)
+            }
+            _ => Err(Error::other("unexpected delete discovery reply")),
+        }
+    }
+    async fn inspect_delete_transaction(
+        &self,
+        volume: &str,
+        object: &str,
+        transaction: Uuid,
+    ) -> Result<DeleteRecoveryInspection> {
+        match self
+            .recover_delete_transaction(
+                volume,
+                DeleteRecoveryRpcOperation::Inspect {
+                    object: object.to_owned(),
+                    transaction,
+                },
+            )
+            .await?
+        {
+            DeleteRecoveryRpcReply::Inspected(inspection) => {
+                validate_delete_recovery_inspection(volume, object, transaction, &inspection)?;
+                Ok(inspection)
+            }
+            _ => Err(Error::other("unexpected delete inspection reply")),
+        }
+    }
+    async fn settle_delete_transaction(
+        &self,
+        volume: &str,
+        object: &str,
+        transaction: Uuid,
+        expected_fingerprint: DeleteRecoveryDigest,
+        action: DeleteRecoveryAction,
+    ) -> Result<DeleteRecoveryInspection> {
+        match self
+            .recover_delete_transaction(
+                volume,
+                DeleteRecoveryRpcOperation::Settle {
+                    object: object.to_owned(),
+                    transaction,
+                    expected_fingerprint,
+                    action,
+                },
+            )
+            .await?
+        {
+            DeleteRecoveryRpcReply::Settled(inspection) => {
+                validate_delete_recovery_inspection(volume, object, transaction, &inspection)?;
+                Ok(inspection)
+            }
+            _ => Err(Error::other("unexpected delete settlement reply")),
+        }
+    }
+
     async fn settle_part_transaction(&self, volume: &str, path: &str, action: PartTransactionAction) -> Result<()> {
         self.execute_with_timeout(
             || async {
@@ -4030,6 +4322,45 @@ mod tests {
 
     static INIT: Once = Once::new();
 
+    #[test]
+    fn recoverable_batch_delete_rejects_invalid_error_codes_and_length() {
+        let errors = decode_delete_recovery_batch_errors(vec![None, Some(DiskError::FileNotFound.to_u32())], 2)
+            .expect("valid per-item errors");
+        assert!(errors[0].is_none());
+        assert_eq!(errors[1], Some(DiskError::FileNotFound));
+        assert!(decode_delete_recovery_batch_errors(vec![None], 2).is_err());
+        assert!(decode_delete_recovery_batch_errors(vec![Some(0)], 1).is_err());
+        assert!(decode_delete_recovery_batch_errors(vec![Some(u32::MAX)], 1).is_err());
+    }
+
+    #[test]
+    fn delete_recovery_inspection_rejects_wrong_transaction_object_and_protocol() {
+        let transaction = Uuid::new_v4();
+        let inspection = DeleteRecoveryInspection {
+            protocol_version: DELETE_RECOVERY_PROTOCOL_VERSION,
+            id: crate::disk::DeleteTransactionId {
+                object: "object".into(),
+                transaction,
+            },
+            state: crate::disk::DeleteRecoveryState::Absent,
+            journal: None,
+            current_digest: None,
+            current_versions: Vec::new(),
+            backup_digest: None,
+            rollback_data_complete: false,
+            commit_data_safe: false,
+            explicit_delete_safe: false,
+            fingerprint: [0; 32],
+        };
+        validate_delete_recovery_inspection("bucket", "object", transaction, &inspection).expect("matching inspection");
+        assert!(validate_delete_recovery_inspection("bucket", "other", transaction, &inspection).is_err());
+        assert!(validate_delete_recovery_inspection("bucket", "object", Uuid::new_v4(), &inspection).is_err());
+        assert!(validate_delete_recovery_inspection("bucket", "object", Uuid::nil(), &inspection).is_err());
+        let mut incompatible = inspection;
+        incompatible.protocol_version += 1;
+        assert!(validate_delete_recovery_inspection("bucket", "object", transaction, &incompatible).is_err());
+    }
+
     #[test]
     fn compare_and_update_wire_outcome_fails_closed() {
         assert_eq!(
@@ -4063,32 +4394,19 @@ mod tests {
     }
 
     #[test]
-    fn retired_marker_options_preserve_legacy_positional_wire_shape() {
+    fn delete_options_named_wire_preserves_rollback_and_legacy_marker_fields() {
         #[derive(Debug, serde::Serialize, serde::Deserialize, PartialEq)]
         struct LegacyDeleteOptions {
             recursive: bool,
             immediate: bool,
             undo_write: bool,
+            #[serde(default)]
             undo_delete: bool,
             old_data_dir: Option<Uuid>,
+            #[serde(default, skip_serializing_if = "Option::is_none")]
+            expected_delete_marker: Option<rustfs_filemeta::MetaDeleteMarker>,
         }
-        let legacy = LegacyDeleteOptions {
-            recursive: false,
-            immediate: false,
-            undo_write: false,
-            undo_delete: false,
-            old_data_dir: None,
-        };
-        let original = encode_msgpack(&legacy).unwrap();
-        let current = encode_msgpack(&DeleteOptions::default()).unwrap();
-        assert_eq!(current, original, "ordinary deletes must retain the older peer's positional payload");
-        assert_eq!(rmp_serde::from_slice::<LegacyDeleteOptions>(&current).unwrap(), legacy);
-        assert!(
-            rmp_serde::from_slice::<DeleteOptions>(&original)
-                .unwrap()
-                .expected_delete_marker
-                .is_none()
-        );
+        let transaction = Uuid::new_v4();
         let mut marker = FileInfo {
             deleted: true,
             version_id: Some(Uuid::new_v4()),
@@ -4097,12 +4415,69 @@ mod tests {
         };
         marker.set_delete_marker_incarnation(Uuid::new_v4());
         let marker = rustfs_filemeta::MetaDeleteMarker::from(marker);
+        for expected_delete_marker in [None, Some(marker)] {
+            for (undo_write, undo_delete) in [(true, false), (false, true)] {
+                let legacy = LegacyDeleteOptions {
+                    recursive: false,
+                    immediate: false,
+                    undo_write,
+                    undo_delete,
+                    old_data_dir: Some(transaction),
+                    expected_delete_marker: expected_delete_marker.clone(),
+                };
+                // Older senders use positional arrays. Their unchanged field prefix and
+                // missing optional fields must also decode on the named-map sender's peer.
+                for (encoding, payload) in [
+                    ("positional", rmp_serde::to_vec(&legacy).expect("encode legacy positional options")),
+                    ("named", rmp_serde::to_vec_named(&legacy).expect("encode legacy named options")),
+                ] {
+                    let decoded: DeleteOptions = rmp_serde::from_slice(&payload).expect("decode legacy options on current peer");
+                    assert_eq!(decoded.undo_write, undo_write, "{encoding} undo-write flag");
+                    assert_eq!(decoded.undo_delete, undo_delete, "{encoding} undo-delete flag");
+                    assert_eq!(decoded.old_data_dir, Some(transaction), "{encoding} transaction UUID");
+                    assert_eq!(decoded.expected_delete_marker, expected_delete_marker, "{encoding} marker precondition");
+                    assert!(decoded.delete_recovery_bucket_incarnation.is_none(), "{encoding} omitted incarnation");
+                    assert!(!decoded.recoverable_delete, "{encoding} omitted recovery flag");
+                }
+                for incarnation in [None, Some(Uuid::new_v4())] {
+                    let current = DeleteOptions {
+                        undo_write,
+                        undo_delete,
+                        old_data_dir: Some(transaction),
+                        expected_delete_marker: expected_delete_marker.clone(),
+                        delete_recovery_bucket_incarnation: incarnation,
+                        recoverable_delete: true,
+                        ..Default::default()
+                    };
+                    let payload = encode_delete_options_msgpack(&current).expect("encode current named options");
+                    let decoded: DeleteOptions = rmp_serde::from_slice(&payload).expect("decode current options on current peer");
+                    assert_eq!(decoded.undo_write, undo_write);
+                    assert_eq!(decoded.undo_delete, undo_delete);
+                    assert_eq!(decoded.old_data_dir, Some(transaction));
+                    assert_eq!(decoded.expected_delete_marker, expected_delete_marker);
+                    assert_eq!(decoded.delete_recovery_bucket_incarnation, incarnation);
+                    assert!(decoded.recoverable_delete);
+                    let decoded: LegacyDeleteOptions =
+                        rmp_serde::from_slice(&payload).expect("decode current options on legacy peer");
+                    assert_eq!(decoded, legacy, "unknown recovery fields cannot shift rollback or marker values");
+                }
+            }
+        }
+    }
+
+    #[test]
+    fn malformed_nonempty_delete_options_do_not_fall_back_to_json() {
         let options = DeleteOptions {
-            expected_delete_marker: Some(marker.clone()),
+            undo_write: true,
+            old_data_dir: Some(Uuid::new_v4()),
             ..Default::default()
         };
-        let decoded: DeleteOptions = rmp_serde::from_slice(&encode_msgpack(&options).unwrap()).unwrap();
-        assert_eq!(decoded.expected_delete_marker, Some(marker));
+        let json = serde_json::to_string(&options).expect("valid compatibility JSON");
+        assert!(decode_msgpack_or_json::<DeleteOptions>(&[0xc1], &json, "DeleteOptions").is_err());
+        let decoded: DeleteOptions =
+            decode_msgpack_or_json(&[], &json, "DeleteOptions").expect("absent MessagePack permits JSON compatibility");
+        assert!(decoded.undo_write);
+        assert_eq!(decoded.old_data_dir, options.old_data_dir);
     }
 
     #[test]
@@ -4785,6 +5160,8 @@ mod tests {
             cleanup_data_dir: Some(Uuid::new_v4()),
             sign: Some(vec![0x14, 0x35]),
             old_current_size: Some(crate::disk::OldCurrentSize::Present(64 * 1024)),
+            old_current_source_checked: false,
+            old_current_source: None,
         };
         let json = serde_json::to_string(&response).expect("legacy rename_data JSON response should encode");
         let decode_before = rustfs_io_metrics::internode_metrics::global_internode_metrics().msgpack_json_decode_total_for_test();
@@ -5187,6 +5564,8 @@ mod tests {
             cleanup_data_dir: Some(Uuid::new_v4()),
             sign: Some(vec![1_u8; 32]),
             old_current_size: Some(crate::disk::OldCurrentSize::Present(4096)),
+            old_current_source_checked: false,
+            old_current_source: None,
         };
         let json = serde_json::to_vec(&response).expect("rename data response json should encode");
         let named_msgpack = rmp_serde::encode::to_vec_named(&response).expect("rename data response named msgpack should encode");

crates/ecstore/src/config/com.rs

@@ -495,22 +495,10 @@ pub(crate) async fn read_config_preserve_empty<S>(api: Arc<S>, file: &str) -> Re
 where
     S: EcstoreObjectIO,
 {
-    let (data, _obj) = read_config_preserve_empty_with_metadata_opts(api, file, &ObjectOptions::default()).await?;
+    let (data, _obj) = read_config_with_metadata_inner(api, file, &ObjectOptions::default(), true, None).await?;
     Ok(data)
 }
 
-/// Preserve an empty config payload, its metadata, and caller-supplied read options.
-pub(crate) async fn read_config_preserve_empty_with_metadata_opts<S>(
-    api: Arc<S>,
-    file: &str,
-    opts: &ObjectOptions,
-) -> Result<(Vec<u8>, ObjectInfo)>
-where
-    S: EcstoreObjectIO,
-{
-    read_config_with_metadata_inner(api, file, opts, true, None).await
-}
-
 pub async fn read_config_no_lock<S>(api: Arc<S>, file: &str) -> Result<Vec<u8>>
 where
     S: ObjectIO<
@@ -539,13 +527,15 @@ pub(crate) async fn read_config_no_lock_preserve_empty_with_metadata<S>(api: Arc
 where
     S: EcstoreObjectIO,
 {
-    read_config_preserve_empty_with_metadata_opts(
+    read_config_with_metadata_inner(
         api,
         file,
         &ObjectOptions {
             no_lock: true,
             ..Default::default()
         },
+        true,
+        None,
     )
     .await
 }
@@ -2825,10 +2815,9 @@ mod tests {
         config_task_join_error, configs_semantically_equal, decode_server_config_blob, encode_server_config_blob,
         heal_config_descriptor, is_standard_object_server_config, lookup_configs, map_system_metadata_write_error,
         new_and_save_server_config, read_config, read_config_no_lock_preserve_empty_with_metadata, read_config_preserve_empty,
-        read_config_preserve_empty_with_metadata_opts, read_config_with_metadata, read_config_without_migrate,
-        read_server_config_snapshot, save_config_with_opts_inner, save_server_config, save_server_config_snapshot,
-        save_server_config_snapshot_with_generation, server_config_transaction_lock_path, should_warn_ignored_scalar_section,
-        storage_class_kvs_mut,
+        read_config_with_metadata, read_config_without_migrate, read_server_config_snapshot, save_config_with_opts_inner,
+        save_server_config, save_server_config_snapshot, save_server_config_snapshot_with_generation,
+        server_config_transaction_lock_path, should_warn_ignored_scalar_section, storage_class_kvs_mut,
     };
     use crate::config::{audit, heal, notify, oidc, scanner};
     use crate::disk::{RUSTFS_META_BUCKET, endpoint::Endpoint};
@@ -5076,7 +5065,6 @@ mod tests {
         heal_replacement: Option<Vec<u8>>,
         heal_calls: AtomicUsize,
         write_calls: AtomicUsize,
-        last_get_no_lock: AtomicBool,
         last_put_no_lock: AtomicBool,
         last_put_preconditions: Mutex<Option<HTTPPreconditions>>,
         revision: AtomicUsize,
@@ -5092,7 +5080,6 @@ mod tests {
                 heal_replacement,
                 heal_calls: AtomicUsize::new(0),
                 write_calls: AtomicUsize::new(0),
-                last_get_no_lock: AtomicBool::new(false),
                 last_put_no_lock: AtomicBool::new(false),
                 last_put_preconditions: Mutex::new(None),
                 revision: AtomicUsize::new(1),
@@ -5148,9 +5135,8 @@ mod tests {
             _object: &str,
             _range: Option<HTTPRangeSpec>,
             _h: HeaderMap,
-            opts: &ObjectOptions,
+            _opts: &ObjectOptions,
         ) -> Result<GetObjectReader> {
-            self.last_get_no_lock.store(opts.no_lock, Ordering::SeqCst);
             let data = match &*self.state.lock().expect("state lock poisoned") {
                 RecoveryReadState::Missing => return Err(Error::ConfigNotFound),
                 RecoveryReadState::Blob(data) => data.clone(),
@@ -5445,60 +5431,6 @@ mod tests {
         assert!(data.is_empty());
     }
 
-    #[tokio::test]
-    async fn read_config_preserve_empty_metadata_keeps_empty_and_options() {
-        let store = Arc::new(RecoveryMockStore::new(RecoveryReadState::Blob(Vec::new()), None));
-        for no_lock in [false, true] {
-            let opts = ObjectOptions {
-                no_lock,
-                ..Default::default()
-            };
-            let (data, info) = read_config_preserve_empty_with_metadata_opts(store.clone(), "config/empty.json", &opts)
-                .await
-                .expect("an existing empty config must retain its payload and metadata");
-            assert!(data.is_empty());
-            assert_eq!(info.size, 0);
-            assert_eq!(info.actual_size, 0);
-            assert_eq!(info.etag.as_deref(), Some("config-1"));
-            assert_eq!(info.data_dir, Some(uuid::Uuid::from_u128(1)));
-            assert_eq!(store.last_get_no_lock.load(Ordering::SeqCst), no_lock);
-        }
-        let err = read_config(store, "config/empty.json")
-            .await
-            .expect_err("the existing reader must still treat empty configs as missing");
-        assert!(matches!(err, Error::ConfigNotFound));
-    }
-
-    #[tokio::test]
-    async fn read_config_preserve_empty_metadata_keeps_payload_and_metadata() {
-        let payload = b"original config bytes".to_vec();
-        let store = Arc::new(RecoveryMockStore::new(RecoveryReadState::Blob(payload.clone()), None));
-        let (data, info) =
-            read_config_preserve_empty_with_metadata_opts(store, "config/nonempty.json", &ObjectOptions::default())
-                .await
-                .expect("a nonempty config must retain its payload and metadata");
-        assert_eq!(data, payload);
-        let size = i64::try_from(payload.len()).expect("test payload size fits i64");
-        assert_eq!(info.size, size);
-        assert_eq!(info.actual_size, size);
-        assert_eq!(info.etag.as_deref(), Some("config-1"));
-        assert_eq!(info.data_dir, Some(uuid::Uuid::from_u128(1)));
-    }
-
-    #[tokio::test]
-    async fn read_config_preserve_empty_metadata_preserves_read_errors() {
-        let missing = Arc::new(RecoveryMockStore::new(RecoveryReadState::Missing, None));
-        let err = read_config_preserve_empty_with_metadata_opts(missing, "config/missing.json", &ObjectOptions::default())
-            .await
-            .expect_err("physical absence must remain ConfigNotFound");
-        assert!(matches!(err, Error::ConfigNotFound));
-        let quorum = Arc::new(RecoveryMockStore::new(RecoveryReadState::QuorumError, None));
-        let err = read_config_preserve_empty_with_metadata_opts(quorum, "config/quorum.json", &ObjectOptions::default())
-            .await
-            .expect_err("a quorum failure must retain its typed read error");
-        assert!(matches!(err, Error::ErasureReadQuorum));
-    }
-
     #[async_trait::async_trait]
     impl StorageAdminApi for RecoveryMockStore {
         type BackendInfo = rustfs_madmin::BackendInfo;

crates/ecstore/src/diagnostics/admin_server_info.rs

@@ -15,7 +15,9 @@
 use crate::cluster::rpc::{
     ScannerBucketListing, TonicInterceptor, gen_tonic_signature_interceptor, node_service_time_out_client,
 };
-use crate::data_usage::{DATA_USAGE_CACHE_NAME, DATA_USAGE_ROOT, load_data_usage_from_backend_cached};
+use crate::data_usage::{
+    DATA_USAGE_CACHE_NAME, DATA_USAGE_ROOT, load_admin_data_usage_from_backend_cached as load_server_info_usage_snapshot,
+};
 use crate::error::{Error, Result};
 use crate::{
     disk::endpoint::{Endpoint, EndpointType},
@@ -42,7 +44,11 @@ use std::{
 use time::OffsetDateTime;
 use tokio::time::timeout;
 use tonic::Request;
-use tracing::warn;
+use tracing::{error, trace, warn};
+
+const EVENT_SERVER_PING: &str = "server_ping";
+const LOG_COMPONENT_ECSTORE: &str = "ecstore";
+const LOG_SUBSYSTEM_DIAGNOSTICS: &str = "diagnostics";
 
 use shadow_rs::shadow;
 
@@ -157,11 +163,22 @@ async fn is_server_resolvable(endpoint: &Endpoint) -> Result<()> {
 
         let response: PingResponse = client.ping(request).await?.into_inner();
 
-        let ping_response_body = flatbuffers::root::<PingBody>(&response.body);
-        if let Err(e) = ping_response_body {
-            eprintln!("{e}");
+        if flatbuffers::root::<PingBody>(&response.body).is_err() {
+            error!(
+                event = EVENT_SERVER_PING,
+                component = LOG_COMPONENT_ECSTORE,
+                subsystem = LOG_SUBSYSTEM_DIAGNOSTICS,
+                result = "invalid_response",
+                "Server ping response decoding failed"
+            );
         } else {
-            println!("ping_resp:body(flatbuffer): {ping_response_body:?}");
+            trace!(
+                event = EVENT_SERVER_PING,
+                component = LOG_COMPONENT_ECSTORE,
+                subsystem = LOG_SUBSYSTEM_DIAGNOSTICS,
+                result = "success",
+                "Server ping completed"
+            );
         }
 
         Ok(())
@@ -294,7 +311,7 @@ pub async fn get_server_info(get_pools: bool) -> InfoMessage {
     if let Some(store) = runtime_sources::object_store_handle() {
         mode = ITEM_ONLINE;
         apply_data_usage_result(
-            load_data_usage_from_backend_cached(store.clone()).await,
+            load_server_info_usage_snapshot(store.clone()).await,
             &mut buckets,
             &mut objects,
             &mut versions,
@@ -967,6 +984,93 @@ mod tests {
         assert_eq!(usage.size, 6);
     }
 
+    #[tokio::test]
+    #[serial]
+    async fn server_info_reads_complete_observation_without_promoting_authority() {
+        use crate::bucket::metadata_sys::test_support::isolated_store_over_temp_disks;
+        use crate::config::com::{read_config, save_config};
+        use crate::data_usage::{
+            DATA_USAGE_OBJ_NAME_PATH, DATA_USAGE_OBSERVED_OBJ_NAME_PATH, invalidate_data_usage_snapshot_cache,
+            load_data_usage_from_backend_cached,
+        };
+        use rustfs_data_usage::{BucketUsageInfo, DataUsageInfo};
+        use std::time::{Duration, SystemTime};
+
+        let (_dirs, store) = isolated_store_over_temp_disks().await;
+        let primary = DataUsageInfo {
+            last_update: Some(SystemTime::UNIX_EPOCH + Duration::from_secs(1)),
+            scanner_epoch: Some(1),
+            usage_snapshot_bootstrap_pending: true,
+            usage_snapshot_converged: Some(false),
+            ..Default::default()
+        };
+        let observed = DataUsageInfo {
+            last_update: Some(SystemTime::UNIX_EPOCH + Duration::from_secs(2)),
+            scanner_epoch: Some(1),
+            scanner_cycle: Some(1),
+            buckets_count: 1,
+            objects_total_count: 3,
+            versions_total_count: 4,
+            delete_markers_total_count: 1,
+            objects_total_size: 6,
+            buckets_usage: [(
+                "bucket".to_string(),
+                BucketUsageInfo {
+                    objects_count: 3,
+                    versions_count: 4,
+                    delete_markers_count: 1,
+                    size: 6,
+                    ..Default::default()
+                },
+            )]
+            .into_iter()
+            .collect(),
+            usage_snapshot_complete: true,
+            usage_snapshot_converged: Some(false),
+            usage_snapshot_authoritative_baseline: Some(primary.snapshot_identity()),
+            ..Default::default()
+        };
+        let primary_bytes = serde_json::to_vec(&primary).expect("bootstrap should encode");
+        save_config(store.clone(), DATA_USAGE_OBJ_NAME_PATH.as_str(), primary_bytes.clone())
+            .await
+            .expect("bootstrap should persist");
+        save_config(
+            store.clone(),
+            DATA_USAGE_OBSERVED_OBJ_NAME_PATH.as_str(),
+            serde_json::to_vec(&observed).expect("observation should encode"),
+        )
+        .await
+        .expect("observation should persist");
+        invalidate_data_usage_snapshot_cache().await;
+
+        let displayed = super::load_server_info_usage_snapshot(store.clone())
+            .await
+            .expect("server info should load the complete observation");
+        assert_eq!(displayed.usage_snapshot_converged, Some(false));
+        let mut buckets = rustfs_madmin::Buckets::default();
+        let mut objects = rustfs_madmin::Objects::default();
+        let mut versions = rustfs_madmin::Versions::default();
+        let mut delete_markers = rustfs_madmin::DeleteMarkers::default();
+        let mut usage = rustfs_madmin::Usage::default();
+        apply_data_usage_result(Ok(displayed), &mut buckets, &mut objects, &mut versions, &mut delete_markers, &mut usage);
+        assert!(objects.error.is_none());
+        assert_eq!(
+            (buckets.count, objects.count, versions.count, delete_markers.count, usage.size),
+            (1, 3, 4, 1, 6)
+        );
+        let quota = load_data_usage_from_backend_cached(store.clone())
+            .await
+            .expect("quota should retain its bootstrap state");
+        assert!(!quota.is_complete_bucket_usage_snapshot());
+        assert_eq!(
+            read_config(store, DATA_USAGE_OBJ_NAME_PATH.as_str())
+                .await
+                .expect("primary should remain readable"),
+            primary_bytes
+        );
+        invalidate_data_usage_snapshot_cache().await;
+    }
+
     #[test]
     fn incomplete_data_usage_is_unavailable_in_server_info() {
         let mut buckets = rustfs_madmin::Buckets::default();

crates/ecstore/src/diagnostics/get.rs

@@ -182,6 +182,7 @@ pub(crate) const GET_METADATA_CACHE_REASON_DELETE_MARKER: &str = "delete_marker"
 pub(crate) const GET_METADATA_CACHE_REASON_DIST_ERASURE: &str = "dist_erasure";
 pub(crate) const GET_METADATA_CACHE_REASON_INCL_FREE_VERSIONS: &str = "incl_free_versions";
 pub(crate) const GET_METADATA_CACHE_REASON_INSUFFICIENT_CACHED_QUORUM: &str = "insufficient_cached_quorum";
+pub(crate) const GET_METADATA_CACHE_REASON_BODY_LAYOUT_REQUIRED: &str = "body_layout_required";
 pub(crate) const GET_METADATA_CACHE_REASON_META_BUCKET: &str = "meta_bucket";
 pub(crate) const GET_METADATA_CACHE_REASON_NO_LOCK: &str = "no_lock";
 pub(crate) const GET_METADATA_CACHE_REASON_NOT_FOUND_OR_EXPIRED: &str = "not_found_or_expired";

crates/ecstore/src/disk/disk_store.rs

@@ -12,6 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use crate::disk::cleanup_runtime;
 use crate::disk::io_admission::{
     AdmissionReader, BoxedChunkReader, DiskIoAdmission, DiskIoClass, DiskIoContext, current_disk_io_context,
     with_disk_io_context, with_disk_io_permit,
@@ -466,12 +467,11 @@ impl LocalDiskWrapper {
         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(),
         )
@@ -485,10 +485,10 @@ impl LocalDiskWrapper {
         opts: DeleteOptions,
         namespace_owner: Option<Arc<dyn Send + Sync>>,
     ) -> Result<()> {
-        self.track_disk_health_mutation(
+        self.track_cleanup_mutation(
             "delete",
             DiskMetricMutation::Delete,
-            || async { Box::pin(self.disk.delete_with_namespace_owner(volume, path, opts, namespace_owner)).await },
+            || async { Box::pin(self.cleanup_delete(volume, path, opts, namespace_owner)).await },
             get_max_timeout_duration(),
         )
         .await
@@ -1476,6 +1476,28 @@ impl LocalDiskWrapper {
         )
     }
 
+    async fn cleanup_delete(
+        &self,
+        volume: &str,
+        path: &str,
+        options: DeleteOptions,
+        owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Result<()> {
+        if !cleanup_runtime::enabled()? {
+            return self.disk.delete_with_namespace_owner(volume, path, options, owner).await;
+        }
+        let disk = self.disk.clone();
+        let volume = volume.to_owned();
+        let path = path.to_owned();
+        let guarded = owner.is_some();
+        let operation = async move { disk.delete_with_namespace_owner(&volume, &path, options, owner).await };
+        if guarded {
+            cleanup_runtime::guarded_disk_operation(operation).await
+        } else {
+            cleanup_runtime::disk_operation(operation).await
+        }
+    }
+
     /// Run a delete under an owned coordinator task when a publication guard
     /// is present. This keeps the guard alive if the RPC waiter is cancelled
     /// while the local namespace mutation is still in progress.
@@ -1494,12 +1516,13 @@ impl LocalDiskWrapper {
         } else {
             get_max_timeout_duration()
         };
+        let cleanup_owner = external_guard.clone();
         run_owned_mutation(external_guard, move || async move {
             operation
-                .track_disk_health_mutation(
+                .track_cleanup_mutation(
                     "delete",
                     DiskMetricMutation::Delete,
-                    || async { operation.disk.delete(&volume, &path, options).await },
+                    || async { operation.cleanup_delete(&volume, &path, options, cleanup_owner).await },
                     timeout_duration,
                 )
                 .await
@@ -1816,6 +1839,8 @@ impl LocalDiskWrapper {
                     undo_delete: false,
                     old_data_dir: None,
                     expected_delete_marker: None,
+                    recoverable_delete: false,
+                    delete_recovery_bucket_incarnation: None,
                 },
             )
             .await?;
@@ -2001,10 +2026,86 @@ impl LocalDiskWrapper {
     }
 
     /// Get the current disk ID
+    pub(crate) async fn delete_versions_with_namespace_owner(
+        &self,
+        volume: &str,
+        versions: Vec<FileInfoVersions>,
+        opts: DeleteOptions,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Vec<Option<Error>> {
+        let version_count = versions.len();
+        self.metrics.record_operation_call("delete_versions");
+        // Check if disk is faulty before proceeding
+        if self.health.is_faulty() {
+            self.metrics.record_availability_error();
+            return vec![Some(DiskError::FaultyDisk); versions.len()];
+        }
+
+        // Check if disk is stale
+        if let Err(e) = self.check_disk_stale().await {
+            self.metrics.record_availability_error();
+            return vec![Some(e); versions.len()];
+        }
+
+        let _io_permit = match self.acquire_disk_io_permit().await {
+            Ok(permit) => permit,
+            Err(error) => return vec![Some(error); version_count],
+        };
+
+        // Record operation start
+        self.health.last_started.store(current_unix_nanos(), Ordering::Relaxed);
+        let waiting_guard = self.health.waiting_guard();
+        let metric_waiting_guard = self.metrics.waiting_guard();
+        let started = Instant::now();
+
+        // Execute the operation
+        let result = self
+            .disk
+            .delete_versions_with_namespace_owner(volume, versions, opts, namespace_owner)
+            .await;
+        self.metrics.record_operation_latency("delete_versions", started.elapsed());
+        self.record_batch_delete_error_metrics(&result);
+
+        drop(waiting_guard);
+        drop(metric_waiting_guard);
+        let has_err = result.iter().any(|e| e.is_some());
+        if !has_err {
+            // Log success and decrement waiting counter
+            self.health.record_operation_success(&self.endpoint(), "operation_success");
+            self.metrics.record_mutation_success(DiskMetricMutation::Delete);
+        }
+
+        result
+    }
+
     pub async fn get_current_disk_id(&self) -> Option<Uuid> {
         *self.disk_id.read().await
     }
 
+    pub(crate) async fn settle_delete_transaction_with_namespace_owner(
+        &self,
+        volume: &str,
+        object: &str,
+        transaction: uuid::Uuid,
+        expected_fingerprint: crate::disk::DeleteRecoveryDigest,
+        action: crate::disk::DeleteRecoveryAction,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Result<crate::disk::DeleteRecoveryInspection> {
+        self.track_disk_health_mutation(
+            "settle_delete_transaction",
+            DiskMetricMutation::Delete,
+            || {
+                Box::pin(async {
+                    self.disk
+                        .delete_recovery_settle(volume, object, transaction, expected_fingerprint, action, namespace_owner)
+                        .await
+                })
+            },
+            crate::disk::get_delete_recovery_timeout()?,
+        )
+        .await
+    }
+
     /// Track disk health for an operation.
     /// This method should wrap disk operations to ensure health checking.
     pub async fn track_disk_health<T, F, Fut>(&self, operation: F, timeout_duration: Duration) -> Result<T>
@@ -2015,6 +2116,31 @@ impl LocalDiskWrapper {
         self.track_disk_health_with_op("unknown", operation, timeout_duration).await
     }
 
+    /// Cleanup latency includes waiting behind a node-wide execution budget,
+    /// so a waiter deadline alone is not evidence that this disk is faulty.
+    /// The active disk health probe retains its original timeout policy.
+    async fn track_cleanup_mutation<T, F, Fut>(
+        &self,
+        op: &'static str,
+        mutation: DiskMetricMutation,
+        operation: F,
+        timeout_duration: Duration,
+    ) -> Result<T>
+    where
+        F: FnOnce() -> Fut,
+        Fut: std::future::Future<Output = Result<T>>,
+    {
+        // Runtime initialization must not prevent control-plane release of
+        // reader tokens. Disk I/O entries still propagate configuration errors.
+        let action = if cleanup_runtime::enabled().unwrap_or(false) {
+            TimeoutHealthAction::IgnoreFailure
+        } else {
+            TimeoutHealthAction::MarkFailure
+        };
+        self.track_disk_health_with_op_timeout_action_and_mutation(op, operation, timeout_duration, action, mutation)
+            .await
+    }
+
     async fn track_disk_health_mutation<T, F, Fut>(
         &self,
         op: &'static str,
@@ -2195,6 +2321,87 @@ impl LocalDiskWrapper {
 
 #[async_trait::async_trait]
 impl DiskAPI for LocalDiskWrapper {
+    async fn delete_version_recoverable(
+        &self,
+        volume: &str,
+        object: &str,
+        file_info: FileInfo,
+        force_del_marker: bool,
+        mut opts: DeleteOptions,
+    ) -> Result<()> {
+        crate::disk::validate_delete_recovery_options(&opts)?;
+        if object.starts_with('/') {
+            return Err(DiskError::InvalidPath);
+        }
+        opts.recoverable_delete = true;
+        self.delete_version(volume, object, file_info, force_del_marker, opts).await
+    }
+    async fn delete_versions_recoverable(
+        &self,
+        volume: &str,
+        versions: Vec<FileInfoVersions>,
+        mut opts: DeleteOptions,
+    ) -> Vec<Option<DiskError>> {
+        if let Err(err) = crate::disk::validate_delete_recovery_options(&opts) {
+            return vec![Some(err); versions.len()];
+        }
+        if versions.iter().any(|version| version.name.starts_with('/')) {
+            return vec![Some(DiskError::InvalidPath); versions.len()];
+        }
+        opts.recoverable_delete = true;
+        self.delete_versions(volume, versions, opts).await
+    }
+    async fn discover_delete_transactions(
+        &self,
+        volume: &str,
+        object: Option<&str>,
+        cursor: Option<crate::disk::DeleteRecoveryCursor>,
+        limit: u32,
+    ) -> Result<crate::disk::DeleteRecoveryPage> {
+        self.track_disk_health_with_op(
+            "discover_delete_transactions",
+            || async {
+                self.disk
+                    .discover_delete_transactions(volume, object, cursor.clone(), limit)
+                    .await
+            },
+            get_max_timeout_duration(),
+        )
+        .await
+    }
+    async fn inspect_delete_transaction(
+        &self,
+        volume: &str,
+        object: &str,
+        transaction: uuid::Uuid,
+    ) -> Result<crate::disk::DeleteRecoveryInspection> {
+        self.track_disk_health_with_op(
+            "inspect_delete_transaction",
+            || async { self.disk.inspect_delete_transaction(volume, object, transaction).await },
+            get_max_timeout_duration(),
+        )
+        .await
+    }
+    async fn settle_delete_transaction(
+        &self,
+        volume: &str,
+        object: &str,
+        transaction: uuid::Uuid,
+        expected_fingerprint: crate::disk::DeleteRecoveryDigest,
+        action: crate::disk::DeleteRecoveryAction,
+    ) -> Result<crate::disk::DeleteRecoveryInspection> {
+        self.track_disk_health_mutation(
+            "settle_delete_transaction",
+            DiskMetricMutation::Delete,
+            || async {
+                self.disk
+                    .settle_delete_transaction(volume, object, transaction, expected_fingerprint, action)
+                    .await
+            },
+            crate::disk::get_delete_recovery_timeout()?,
+        )
+        .await
+    }
     fn has_replacement_mount_lease(&self) -> bool {
         self.disk.has_replacement_mount_lease()
     }
@@ -2408,46 +2615,7 @@ impl DiskAPI for LocalDiskWrapper {
     }
 
     async fn delete_versions(&self, volume: &str, versions: Vec<FileInfoVersions>, opts: DeleteOptions) -> Vec<Option<Error>> {
-        let version_count = versions.len();
-        self.metrics.record_operation_call("delete_versions");
-        // Check if disk is faulty before proceeding
-        if self.health.is_faulty() {
-            self.metrics.record_availability_error();
-            return vec![Some(DiskError::FaultyDisk); versions.len()];
-        }
-
-        // Check if disk is stale
-        if let Err(e) = self.check_disk_stale().await {
-            self.metrics.record_availability_error();
-            return vec![Some(e); versions.len()];
-        }
-
-        let _io_permit = match self.acquire_disk_io_permit().await {
-            Ok(permit) => permit,
-            Err(error) => return vec![Some(error); version_count],
-        };
-
-        // Record operation start
-        self.health.last_started.store(current_unix_nanos(), Ordering::Relaxed);
-        let waiting_guard = self.health.waiting_guard();
-        let metric_waiting_guard = self.metrics.waiting_guard();
-        let started = Instant::now();
-
-        // Execute the operation
-        let result = self.disk.delete_versions(volume, versions, opts).await;
-        self.metrics.record_operation_latency("delete_versions", started.elapsed());
-        self.record_batch_delete_error_metrics(&result);
-
-        drop(waiting_guard);
-        drop(metric_waiting_guard);
-        let has_err = result.iter().any(|e| e.is_some());
-        if !has_err {
-            // Log success and decrement waiting counter
-            self.health.record_operation_success(&self.endpoint(), "operation_success");
-            self.metrics.record_mutation_success(DiskMetricMutation::Delete);
-        }
-
-        result
+        self.delete_versions_with_namespace_owner(volume, versions, opts, None).await
     }
 
     async fn delete_paths(&self, volume: &str, paths: &[String]) -> Result<()> {
@@ -2470,9 +2638,37 @@ impl DiskAPI for LocalDiskWrapper {
     }
 
     async fn release_snapshot_lease(&self, volume: &str, path: &str, token: SnapshotLeaseToken) -> Result<()> {
-        self.track_disk_health_with_op(
+        // Quota fence release is control-plane work and may wait for in-flight
+        // mutations. Keep its original waiter lifecycle instead of creating
+        // detached cleanup tasks for repeated fence-release RPCs.
+        if volume == super::RUSTFS_META_BUCKET && super::is_quota_mutation_fence_path(path) {
+            return self
+                .track_disk_health_with_op(
+                    "release_snapshot_lease",
+                    || async { self.disk.release_snapshot_lease(volume, path, token).await },
+                    get_max_timeout_duration(),
+                )
+                .await;
+        }
+        self.track_cleanup_mutation(
             "release_snapshot_lease",
-            || async { self.disk.release_snapshot_lease(volume, path, token).await },
+            DiskMetricMutation::None,
+            || async {
+                if self.disk.release_snapshot_lease_without_cleanup(volume, path, token).await {
+                    return Ok(());
+                }
+                if !cleanup_runtime::enabled().unwrap_or(false) {
+                    // The raw release removes the token before trying the
+                    // reclaim budget. Initialization failure therefore keeps
+                    // the cleanup intent and data, without leaking a reader
+                    // or falling back to unbudgeted filesystem deletion.
+                    return self.disk.release_snapshot_lease(volume, path, token).await;
+                }
+                let disk = self.disk.clone();
+                let volume = volume.to_owned();
+                let path = path.to_owned();
+                cleanup_runtime::release_lease(async move { disk.release_snapshot_lease(&volume, &path, token).await }).await
+            },
             get_max_timeout_duration(),
         )
         .await
@@ -2497,10 +2693,22 @@ impl DiskAPI for LocalDiskWrapper {
         } else {
             get_max_timeout_duration()
         };
-        self.track_disk_health_mutation(
+        self.track_cleanup_mutation(
             "delete_data_dir",
             DiskMetricMutation::Delete,
-            || async { self.disk.delete_data_dir(volume, path, opts).await },
+            || async {
+                if !cleanup_runtime::enabled()? {
+                    return self.disk.delete_data_dir(volume, path, opts).await;
+                }
+                let disk = self.disk.clone();
+                let volume = volume.to_owned();
+                let path = path.to_owned();
+                cleanup_runtime::disk_operation(async move {
+                    let _scope = scope;
+                    disk.delete_data_dir(&volume, &path, opts).await
+                })
+                .await
+            },
             timeout,
         )
         .await
@@ -2747,10 +2955,10 @@ impl DiskAPI for LocalDiskWrapper {
     }
 
     async fn delete(&self, volume: &str, path: &str, opt: DeleteOptions) -> Result<()> {
-        self.track_disk_health_mutation(
+        self.track_cleanup_mutation(
             "delete",
             DiskMetricMutation::Delete,
-            || async { self.disk.delete(volume, path, opt).await },
+            || async { self.cleanup_delete(volume, path, opt, None).await },
             get_max_timeout_duration(),
         )
         .await
@@ -2783,11 +2991,25 @@ impl DiskAPI for LocalDiskWrapper {
     }
 
     async fn write_all(&self, volume: &str, path: &str, data: Bytes) -> Result<()> {
-        self.track_disk_health_mutation(
+        let cleanup = path.rsplit('/').next() == Some(cleanup_runtime::OLD_DATA_CLEANUP_RECEIPT_FILE);
+        self.track_disk_health_with_op_timeout_action_and_mutation(
             "write_all",
-            DiskMetricMutation::Write,
-            || async { self.disk.write_all(volume, path, data).await },
+            || async {
+                if !cleanup || !cleanup_runtime::enabled()? {
+                    return self.disk.write_all(volume, path, data).await;
+                }
+                let disk = self.disk.clone();
+                let volume = volume.to_owned();
+                let path = path.to_owned();
+                cleanup_runtime::disk_operation(async move { disk.write_all(&volume, &path, data).await }).await
+            },
             get_max_timeout_duration(),
+            if cleanup && cleanup_runtime::enabled()? {
+                TimeoutHealthAction::IgnoreFailure
+            } else {
+                TimeoutHealthAction::MarkFailure
+            },
+            DiskMetricMutation::Write,
         )
         .await
     }
@@ -3180,6 +3402,256 @@ mod tests {
         let _ = walk.await;
     }
 
+    #[tokio::test]
+    async fn invalid_cleanup_config_does_not_leak_snapshot_readers() {
+        const CHILD: &str = "RUSTFS_CLEANUP_INVALID_LEASE_TEST_CHILD";
+        if std::env::var_os(CHILD).is_none() {
+            let output = tokio::task::spawn_blocking(|| {
+                std::process::Command::new(std::env::current_exe().expect("test executable"))
+                    .args([
+                        "--exact",
+                        "disk::disk_store::tests::invalid_cleanup_config_does_not_leak_snapshot_readers",
+                        "--nocapture",
+                    ])
+                    .env(CHILD, "1")
+                    .env("RUSTFS_CLEANUP_ISOLATE_ENABLE", "true")
+                    .env("RUSTFS_CLEANUP_ASYNC_THREADS", "0")
+                    .output()
+                    .expect("run with invalid cleanup configuration")
+            })
+            .await
+            .expect("child process task");
+            assert!(
+                output.status.success(),
+                "{}\n{}",
+                String::from_utf8_lossy(&output.stdout),
+                String::from_utf8_lossy(&output.stderr)
+            );
+            assert!(String::from_utf8_lossy(&output.stdout).contains("1 passed"));
+            return;
+        }
+        let dir = tempfile::tempdir().expect("temporary disk");
+        let endpoint = Endpoint::try_from(dir.path().to_str().expect("UTF-8 path")).expect("endpoint");
+        let disk = Arc::new(LocalDisk::new(&endpoint, false).await.expect("local disk"));
+        let wrapper = LocalDiskWrapper::new(disk.clone(), false);
+        let volume = "invalid-cleanup-config";
+        let path = "object/data";
+        wrapper.make_volume(volume).await.expect("bucket");
+        wrapper
+            .write_all(volume, "object/data/part.1", Bytes::from_static(b"live shard"))
+            .await
+            .expect("write shard");
+        let token = wrapper.acquire_snapshot_lease(volume, path).await.expect("GET lease");
+        wrapper
+            .release_snapshot_lease(volume, path, token)
+            .await
+            .expect("control-only release must not depend on cleanup initialization");
+        assert!(matches!(
+            wrapper.renew_snapshot_lease(volume, path, token).await,
+            Err(DiskError::FileNotFound)
+        ));
+
+        let token = wrapper
+            .acquire_snapshot_lease(volume, path)
+            .await
+            .expect("reader of old data");
+        // Register reclaim intent directly; this only updates the lease
+        // registry while a reader is present and performs no filesystem I/O.
+        assert_eq!(
+            disk.delete_data_dir(
+                volume,
+                path,
+                DeleteOptions {
+                    recursive: true,
+                    ..Default::default()
+                }
+            )
+            .await
+            .expect("defer reclaim"),
+            DataDirDeleteStatus::Deferred
+        );
+        assert!(
+            wrapper.release_snapshot_lease(volume, path, token).await.is_err(),
+            "invalid configuration must still reject reclamation"
+        );
+        assert!(
+            matches!(wrapper.renew_snapshot_lease(volume, path, token).await, Err(DiskError::FileNotFound)),
+            "failed reclamation must not strand a reader token"
+        );
+        assert_eq!(
+            wrapper.read_all(volume, "object/data/part.1").await.expect("retained data"),
+            Bytes::from_static(b"live shard"),
+            "configuration failure must not fall back to unbudgeted deletion"
+        );
+    }
+
+    #[tokio::test]
+    async fn rpc_cleanup_entries_share_disk_budget() {
+        const CHILD: &str = "RUSTFS_CLEANUP_RPC_TEST_CHILD";
+        if std::env::var_os(CHILD).is_none() {
+            let output = tokio::task::spawn_blocking(|| {
+                std::process::Command::new(std::env::current_exe().unwrap())
+                    .args([
+                        "--exact",
+                        "disk::disk_store::tests::rpc_cleanup_entries_share_disk_budget",
+                        "--nocapture",
+                    ])
+                    .env(CHILD, "1")
+                    .env("RUSTFS_CLEANUP_ISOLATE_ENABLE", "true")
+                    .env("RUSTFS_CLEANUP_ASYNC_THREADS", "1")
+                    .env("RUSTFS_CLEANUP_DISK_MAX_PENDING", "1")
+                    .env("RUSTFS_CLEANUP_DISK_WORKERS", "1")
+                    .env("RUSTFS_CLEANUP_GC_WORKERS", "1")
+                    .env("RUSTFS_CLEANUP_CPUS", "none")
+                    .env_remove("RUSTFS_PUT_RENAME_TAIL_CLEANUP_COUNTERFACTUAL_SKIP")
+                    .env_remove("RUSTFS_PUT_RENAME_TAIL_CLEANUP_DEFER_HOLD_WORKER")
+                    .env_remove("RUSTFS_PUT_RENAME_TAIL_CLEANUP_ZERO_TARGET_TMP_DELETE_SKIP")
+                    .output()
+                    .unwrap()
+            })
+            .await
+            .unwrap();
+            assert!(
+                output.status.success(),
+                "{}\n{}",
+                String::from_utf8_lossy(&output.stdout),
+                String::from_utf8_lossy(&output.stderr)
+            );
+            assert!(String::from_utf8_lossy(&output.stdout).contains("1 passed"));
+            return;
+        }
+        let dir = tempfile::tempdir().unwrap();
+        let endpoint = Endpoint::try_from(dir.path().to_str().unwrap()).unwrap();
+        let disk = Arc::new(LocalDisk::new(&endpoint, false).await.unwrap());
+        let wrapper = LocalDiskWrapper::new(disk, false);
+        let volume = "cleanup-receiver";
+        let path = "object/old-data";
+        let body_path = format!("{path}/part.1");
+        let receipt_path = format!("{path}/{}", cleanup_runtime::OLD_DATA_CLEANUP_RECEIPT_FILE);
+        wrapper.make_volume(volume).await.unwrap();
+        wrapper
+            .write_all(volume, &body_path, Bytes::from_static(b"old body"))
+            .await
+            .unwrap();
+        let read_lease = wrapper
+            .acquire_snapshot_lease(volume, path)
+            .await
+            .expect("ordinary GET lease");
+        let (entered, entry) = tokio::sync::oneshot::channel();
+        let (release, released) = std::sync::mpsc::channel();
+        let occupied = tokio::spawn(cleanup_runtime::disk_operation(async move {
+            entered.send(()).unwrap();
+            // Deliberately stall the only cleanup scheduler to prove that
+            // ordinary GET completion does not dispatch to it.
+            released
+                .recv_timeout(Duration::from_secs(30))
+                .expect("release stalled scheduler");
+            Ok(())
+        }));
+        tokio::time::timeout(Duration::from_secs(20), entry).await.unwrap().unwrap();
+        tokio::time::timeout(Duration::from_secs(5), wrapper.release_snapshot_lease(volume, path, read_lease))
+            .await
+            .expect("ordinary GET release must progress while the cleanup scheduler is stalled")
+            .expect("release without reclamation");
+        let fence_path = super::super::quota_mutation_fence_path(volume, "object");
+        let fence = wrapper
+            .acquire_snapshot_lease(super::super::RUSTFS_META_BUCKET, &fence_path)
+            .await
+            .unwrap();
+        wrapper
+            .release_snapshot_lease(super::super::RUSTFS_META_BUCKET, &fence_path, fence)
+            .await
+            .unwrap();
+        let timeout = wrapper
+            .track_cleanup_mutation(
+                "delete",
+                DiskMetricMutation::Delete,
+                std::future::pending::<Result<()>>,
+                Duration::from_millis(10),
+            )
+            .await;
+        assert!(matches!(timeout, Err(DiskError::Timeout)));
+        assert!(!wrapper.health.is_faulty(), "cleanup queue delay must not mark a healthy disk faulty");
+        let options = DeleteOptions {
+            recursive: true,
+            ..Default::default()
+        };
+        let drops = Arc::new(std::sync::atomic::AtomicUsize::new(0));
+        // This is the same entry point used after RPC authentication and
+        // scanner publication-lease validation by handle_delete.
+        let result = wrapper
+            .delete_with_publication_guard(volume, path, options.clone(), Some(Arc::new(DropProbe(drops.clone()))))
+            .await;
+        assert!(matches!(result, Err(DiskError::Io(err)) if err.kind() == io::ErrorKind::WouldBlock));
+        assert_eq!(drops.load(Ordering::SeqCst), 1);
+        assert!(matches!(wrapper.delete_data_dir(volume, path, options.clone()).await,
+            Err(DiskError::Io(err)) if err.kind() == io::ErrorKind::WouldBlock));
+        assert!(matches!(wrapper.write_all(volume, &receipt_path, Bytes::from_static(b"receipt")).await,
+            Err(DiskError::Io(err)) if err.kind() == io::ErrorKind::WouldBlock));
+        assert_eq!(wrapper.read_all(volume, &body_path).await.unwrap(), Bytes::from_static(b"old body"));
+        release.send(()).unwrap();
+        occupied.await.unwrap().unwrap();
+        wrapper
+            .write_all(volume, &receipt_path, Bytes::from_static(b"receipt"))
+            .await
+            .unwrap();
+        wrapper
+            .delete_with_publication_guard(volume, path, options, None)
+            .await
+            .unwrap();
+        assert!(matches!(wrapper.read_all(volume, &body_path).await, Err(DiskError::FileNotFound)));
+        // Snapshot readers defer old-data cleanup. Saturation must release
+        // their token while retaining reclaim intent, rather than leaking a
+        // live token or performing an unbudgeted delete on the RPC runtime.
+        wrapper
+            .write_all(volume, &body_path, Bytes::from_static(b"leased body"))
+            .await
+            .unwrap();
+        let lease = wrapper.acquire_snapshot_lease(volume, path).await.unwrap();
+        assert_eq!(
+            wrapper
+                .delete_data_dir(
+                    volume,
+                    path,
+                    DeleteOptions {
+                        recursive: true,
+                        ..Default::default()
+                    }
+                )
+                .await
+                .unwrap(),
+            DataDirDeleteStatus::Deferred
+        );
+        let (entered, entry) = tokio::sync::oneshot::channel();
+        let (release, released) = tokio::sync::oneshot::channel();
+        let occupied = tokio::spawn(cleanup_runtime::disk_operation(async move {
+            entered.send(()).unwrap();
+            released.await.unwrap();
+            Ok(())
+        }));
+        tokio::time::timeout(Duration::from_secs(20), entry).await.unwrap().unwrap();
+        assert!(matches!(wrapper.release_snapshot_lease(volume, path, lease).await,
+            Err(DiskError::Io(err)) if err.kind() == io::ErrorKind::WouldBlock));
+        assert_eq!(wrapper.read_all(volume, &body_path).await.unwrap(), Bytes::from_static(b"leased body"));
+        release.send(()).unwrap();
+        occupied.await.unwrap().unwrap();
+        assert_eq!(
+            wrapper
+                .delete_data_dir(
+                    volume,
+                    path,
+                    DeleteOptions {
+                        recursive: true,
+                        ..Default::default()
+                    }
+                )
+                .await
+                .unwrap(),
+            DataDirDeleteStatus::Deleted
+        );
+        assert!(matches!(wrapper.read_all(volume, &body_path).await, Err(DiskError::FileNotFound)));
+    }
+
     #[tokio::test]
     async fn owned_mutation_keeps_publication_guard_after_waiter_cancellation() {
         let drops = Arc::new(std::sync::atomic::AtomicUsize::new(0));

crates/ecstore/src/disk/error_reduce.rs

@@ -131,6 +131,15 @@ pub fn build_write_quorum_failure_summary(
 
 fn dominant_error_label(errors: &[Option<Error>], ignored_errs: &[Error], dominant_error: Option<&Error>) -> &'static str {
     let Some(dominant_error) = dominant_error else {
+        // The reducer also returns None when every failure is ignored for
+        // quorum voting. That case contains no successful nil entry.
+        if !errors.is_empty() && errors.iter().all(Option::is_some) {
+            return if errors.iter().all(|error| error.as_ref() == Some(&Error::DiskNotFound)) {
+                "disk_not_found"
+            } else {
+                "all_ignored"
+            };
+        }
         return "nil_dominated";
     };
 
@@ -387,9 +396,9 @@ mod tests {
     }
 
     #[test]
-    fn test_build_write_quorum_failure_summary_all_ignored_is_nil_dominated() {
-        // All disks offline (all DiskNotFound). The dominant error must be None and
-        // the label must be "nil_dominated" rather than falling back to "other_error".
+    fn test_build_write_quorum_failure_summary_all_missing_is_disk_not_found() {
+        // Ignoring unavailable disks for quorum reduction must not imply a
+        // successful nil entry in the diagnostic label.
         let errors = vec![
             Some(Error::DiskNotFound),
             Some(Error::DiskNotFound),
@@ -397,7 +406,21 @@ mod tests {
         ];
         let summary = build_write_quorum_failure_summary(&errors, OBJECT_OP_IGNORED_ERRS, 2);
         assert_eq!(summary.dominant_error, None);
-        assert_eq!(summary.dominant_error_label, "nil_dominated");
+        assert_eq!(summary.achieved, 0);
+        assert_eq!(summary.dominant_error_label, "disk_not_found");
+    }
+
+    #[test]
+    fn test_build_write_quorum_failure_summary_all_ignored_has_no_success() {
+        let errors = [Some(Error::FaultyDisk), Some(Error::DiskNotFound)];
+        let summary = build_write_quorum_failure_summary(&errors, OBJECT_OP_IGNORED_ERRS, 2);
+        assert_eq!(summary.dominant_error, None);
+        assert_eq!(summary.achieved, 0);
+        assert_eq!(summary.dominant_error_label, "all_ignored");
+        assert_eq!(
+            reduce_write_quorum_errs(&errors, OBJECT_OP_IGNORED_ERRS, 2),
+            Some(Error::ErasureWriteQuorum)
+        );
     }
 
     #[test]

crates/ecstore/src/disk/fs.rs

@@ -223,12 +223,6 @@ pub fn rename_std(from: impl AsRef<Path>, to: impl AsRef<Path>) -> io::Result<()
     std::fs::rename(from, to)
 }
 
-#[cfg(any(not(windows), test))]
-#[tracing::instrument(level = "debug", skip_all)]
-pub async fn read_file(path: impl AsRef<Path>) -> io::Result<Vec<u8>> {
-    fs::read(path.as_ref()).await
-}
-
 // Bucket existence cache - reduces statx syscalls for repeated bucket checks
 
 /// Cache for bucket directory existence checks.
@@ -614,28 +608,6 @@ mod tests {
         assert_eq!(content, "test content");
     }
 
-    #[tokio::test]
-    async fn test_read_file() {
-        let temp_dir = TempDir::new().unwrap();
-        let file_path = temp_dir.path().join("test_read.txt");
-
-        let test_content = b"This is test content for reading";
-        tokio::fs::write(&file_path, test_content).await.unwrap();
-
-        // Read file
-        let read_content = read_file(&file_path).await.unwrap();
-        assert_eq!(read_content, test_content);
-    }
-
-    #[tokio::test]
-    async fn test_read_file_nonexistent() {
-        let temp_dir = TempDir::new().unwrap();
-        let file_path = temp_dir.path().join("nonexistent.txt");
-
-        // Should fail for non-existent file
-        assert!(read_file(&file_path).await.is_err());
-    }
-
     #[tokio::test]
     async fn test_same_file() {
         let temp_dir = TempDir::new().unwrap();

crates/ecstore/src/disk/local/commit.rs

@@ -47,6 +47,33 @@ use tokio::fs;
 use tracing::{info, warn};
 use uuid::Uuid;
 
+fn transitioned_old_current_source(
+    dst_meta_existed: bool,
+    dst_meta_unparsable: bool,
+    xlmeta: &FileMeta,
+    volume: &str,
+    path: &str,
+    version_id: Uuid,
+) -> (bool, Option<FileInfo>) {
+    if dst_meta_unparsable {
+        return (false, None);
+    }
+    if !dst_meta_existed {
+        return (true, None);
+    }
+    let Ok(old_current) = xlmeta.into_fileinfo(volume, path, "", false, false, true) else {
+        return (false, None);
+    };
+    if old_current.version_id.unwrap_or_default() != version_id {
+        return (true, None);
+    }
+    if old_current.transition_status == rustfs_filemeta::TRANSITION_COMPLETE {
+        (true, Some(old_current))
+    } else {
+        (true, None)
+    }
+}
+
 /// Hold later repair publications after admitting one baseline object. The
 /// fixture arms this on one replacement disk before rejoining the cluster.
 #[cfg(feature = "e2e-test-hooks")]
@@ -214,7 +241,7 @@ pub(super) async fn lock_rename_commit_directories(
     publication_root: &os::PublicationRoot,
     mutation_lease: Arc<os::NamespaceMutationLease>,
 ) -> Result<os::RenameCommitGuard> {
-    #[cfg(windows)]
+    #[cfg(any(windows, target_os = "linux"))]
     let result = {
         let source_parent = source_parent.to_path_buf();
         let destination_parent = destination_parent.to_path_buf();
@@ -222,7 +249,7 @@ pub(super) async fn lock_rename_commit_directories(
         let publication_root = publication_root.clone();
         os::run_blocking_namespace_operation(mutation_lease, move || {
             let result = os::prepare_rename_commit_guard(&source_parent, &destination_parent, &base_dir, &publication_root);
-            #[cfg(test)]
+            #[cfg(all(test, windows))]
             if result.is_ok() {
                 run_destination_commit_directory_preparation(&destination_parent);
             }
@@ -230,7 +257,7 @@ pub(super) async fn lock_rename_commit_directories(
         })
         .await
     };
-    #[cfg(not(windows))]
+    #[cfg(not(any(windows, target_os = "linux")))]
     let result = {
         let _ = mutation_lease;
         os::prepare_rename_commit_guard(source_parent, destination_parent, base_dir, publication_root)
@@ -244,12 +271,12 @@ pub(super) async fn lock_rename_commit_directories(
     result.map_err(to_file_error).map_err(DiskError::from)
 }
 
-async fn read_rename_destination_metadata(
+async fn prepare_rename_destination_metadata(
     file_path: &Path,
     rename_commit_guard: &os::RenameCommitGuard,
     mutation_lease: Arc<os::NamespaceMutationLease>,
 ) -> Result<Option<Bytes>> {
-    #[cfg(windows)]
+    #[cfg(any(windows, target_os = "linux"))]
     let result = {
         let file_path = file_path.to_path_buf();
         let rename_commit_guard = rename_commit_guard.clone();
@@ -258,10 +285,10 @@ async fn read_rename_destination_metadata(
         })
         .await
     };
-    #[cfg(not(windows))]
+    #[cfg(not(any(windows, target_os = "linux")))]
     let _ = (rename_commit_guard, mutation_lease);
-    #[cfg(not(windows))]
-    let result = match super::super::fs::read_file(file_path).await {
+    #[cfg(not(any(windows, target_os = "linux")))]
+    let result = match fs::read(file_path).await {
         Ok(data) => Ok(Some(data)),
         Err(err) if err.kind() == ErrorKind::NotFound => Ok(None),
         Err(err) => Err(err),
@@ -398,7 +425,7 @@ impl LocalDisk {
 
         // xl.meta path
         let src_file_path = self.io_get_object_path(src_volume, format!("{}/{}", src_path, STORAGE_FORMAT_FILE).as_str())?;
-        let dst_file_path = self.io_get_object_path(dst_volume, format!("{}/{}", dst_path, STORAGE_FORMAT_FILE).as_str())?;
+        let dst_file_path = self.io_get_object_open_path(dst_volume, format!("{}/{}", dst_path, STORAGE_FORMAT_FILE).as_str())?;
 
         // data_dir path
         let has_data_dir_path = {
@@ -453,8 +480,7 @@ impl LocalDisk {
             fs::create_dir_all(src_file_parent).await.map_err(to_file_error)?;
         }
         // Acquire the common trees before reading destination metadata. On
-        // Windows this pins the object directory identity across metadata
-        // preparation, data publication, rollback backup, and final commit.
+        // Windows this pins the object directory identity across publication.
         let rename_commit_guard = lock_rename_commit_directories(
             src_file_parent,
             dst_file_parent,
@@ -472,9 +498,12 @@ impl LocalDisk {
             source_volume_dir: &src_volume_dir,
             data_paths: has_data_dir_path.as_ref(),
         };
-        let has_dst_buf =
-            read_rename_destination_metadata(commit.destination_metadata, &commit.directory_guard, commit.mutation_lease.clone())
-                .await?;
+        let has_dst_buf = prepare_rename_destination_metadata(
+            commit.destination_metadata,
+            &commit.directory_guard,
+            commit.mutation_lease.clone(),
+        )
+        .await?;
 
         if no_inline {
             // Non-inline: read xl.meta, parse, write, rename data dir, rename xl.meta
@@ -508,6 +537,14 @@ impl LocalDisk {
             }
 
             let version_id = fi.version_id.unwrap_or_default();
+            let (old_current_source_checked, old_current_source) = transitioned_old_current_source(
+                has_dst_buf.is_some(),
+                dst_meta_unparsable,
+                &xlmeta,
+                dst_volume,
+                dst_path,
+                version_id,
+            );
             let has_old_data_dir = xlmeta.find_unshared_data_dir_for_version(Some(version_id));
             let old_version_exists = xlmeta.find_version(Some(version_id)).is_ok();
             let rollback_data_dir = has_old_data_dir.or_else(|| {
@@ -798,7 +835,9 @@ impl LocalDisk {
                 && let Some(parent) = dst_file_path.parent()
             {
                 let fsync_started = rustfs_io_metrics::put_stage_timer();
-                if let Err(err) = os::fsync_dst_dir_group_commit(parent, Some(commit.mutation_lease.clone())).await {
+                if let Err(err) =
+                    os::fsync_commit_directory(parent, &commit.directory_guard, commit.mutation_lease.clone(), None).await
+                {
                     rustfs_io_metrics::record_put_object_stage_duration_from(
                         rustfs_io_metrics::PUT_STAGE_SET_DISK_RENAME_DST_DIR_FSYNC,
                         fsync_started,
@@ -897,6 +936,8 @@ impl LocalDisk {
                 cleanup_data_dir: has_old_data_dir,
                 sign: version_signature,
                 old_current_size,
+                old_current_source_checked,
+                old_current_source,
             })
         } else {
             // Inline metadata preparation is blocking. The transaction lease is
@@ -910,6 +951,8 @@ impl LocalDisk {
                 None
             };
             let dst_path_for_failpoint = dst_path.to_string();
+            let dst_volume_for_source = dst_volume.to_string();
+            let dst_path_for_source = dst_path.to_string();
             #[cfg(windows)]
             let source_parent = src_file_parent.to_path_buf();
             let rename_commit_guard_for_preparation = commit.directory_guard.clone();
@@ -954,6 +997,14 @@ impl LocalDisk {
                 };
 
                 let version_id = fi.version_id.unwrap_or_default();
+                let (old_current_source_checked, old_current_source) = transitioned_old_current_source(
+                    has_dst_buf.is_some(),
+                    dst_meta_unparsable,
+                    &xlmeta,
+                    &dst_volume_for_source,
+                    &dst_path_for_source,
+                    version_id,
+                );
                 let old_data_dir = xlmeta.find_unshared_data_dir_for_version(Some(version_id));
                 let old_version_exists = xlmeta.find_version(Some(version_id)).is_ok();
                 let rollback_data_dir = old_data_dir.or_else(|| {
@@ -1003,6 +1054,8 @@ impl LocalDisk {
                     old_data_dir,
                     version_signature,
                     old_current_size,
+                    old_current_source_checked,
+                    old_current_source,
                     staged_rollback_path,
                     has_dst_buf.is_none(),
                     prepared_metadata_source,
@@ -1022,6 +1075,8 @@ impl LocalDisk {
                 cleanup_data_dir,
                 version_signature,
                 old_current_size,
+                old_current_source_checked,
+                old_current_source,
                 mut local_rollback_path,
                 destination_was_absent,
                 prepared_metadata_source,
@@ -1147,10 +1202,11 @@ impl LocalDisk {
                     && let Some(dst_parent) = dst_file_path.parent()
                 {
                     let fsync_started = rustfs_io_metrics::put_stage_timer();
-                    if let Err(err) = os::fsync_dst_dir_group_commit_or_namespace_file_sync_limit(
+                    if let Err(err) = os::fsync_commit_directory(
                         dst_parent,
+                        &commit.directory_guard,
                         commit.mutation_lease.clone(),
-                        admission,
+                        Some(admission),
                     )
                     .await
                     {
@@ -1262,6 +1318,8 @@ impl LocalDisk {
                 cleanup_data_dir,
                 sign: version_signature,
                 old_current_size,
+                old_current_source_checked,
+                old_current_source,
             })
         }
     }

crates/ecstore/src/disk/mod.rs

@@ -14,6 +14,7 @@
 
 // #730: disk abstractions still carry staged health and direct-I/O migration paths.
 
+pub(crate) mod cleanup_runtime;
 pub mod disk_store;
 pub mod endpoint;
 pub mod error;
@@ -230,6 +231,105 @@ pub enum Disk {
 }
 
 impl Disk {
+    pub(crate) async fn delete_versions_with_namespace_owner(
+        &self,
+        volume: &str,
+        versions: Vec<FileInfoVersions>,
+        opts: DeleteOptions,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Vec<Option<Error>> {
+        match self {
+            Self::Local(disk) => {
+                disk.delete_versions_with_namespace_owner(volume, versions, opts, namespace_owner)
+                    .await
+            }
+            Self::Remote(disk) => {
+                let _owner = namespace_owner;
+                disk.delete_versions(volume, versions, opts).await
+            }
+        }
+    }
+
+    pub(crate) async fn delete_versions_recoverable_with_namespace_owner(
+        &self,
+        volume: &str,
+        versions: Vec<FileInfoVersions>,
+        mut opts: DeleteOptions,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Vec<Option<Error>> {
+        if let Err(error) = validate_delete_recovery_options(&opts) {
+            return vec![Some(error); versions.len()];
+        }
+        if versions.iter().any(|version| version.name.starts_with('/')) {
+            return vec![Some(DiskError::InvalidPath); versions.len()];
+        }
+        opts.recoverable_delete = true;
+        match self {
+            Self::Local(disk) => {
+                disk.delete_versions_with_namespace_owner(volume, versions, opts, namespace_owner)
+                    .await
+            }
+            Self::Remote(disk) => {
+                let _owner = namespace_owner;
+                disk.delete_versions_recoverable(volume, versions, opts).await
+            }
+        }
+    }
+
+    pub(crate) async fn settle_delete_transaction_with_namespace_owner(
+        &self,
+        volume: &str,
+        object: &str,
+        transaction: Uuid,
+        expected_fingerprint: DeleteRecoveryDigest,
+        action: DeleteRecoveryAction,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Result<DeleteRecoveryInspection> {
+        match self {
+            Self::Local(disk) => {
+                disk.settle_delete_transaction_with_namespace_owner(
+                    volume,
+                    object,
+                    transaction,
+                    expected_fingerprint,
+                    action,
+                    namespace_owner,
+                )
+                .await
+            }
+            Self::Remote(disk) => {
+                let _owner = namespace_owner;
+                disk.settle_delete_transaction(volume, object, transaction, expected_fingerprint, action)
+                    .await
+            }
+        }
+    }
+
+    pub(crate) async fn delete_version_recoverable_with_namespace_owner(
+        &self,
+        volume: &str,
+        path: &str,
+        fi: FileInfo,
+        force_del_marker: bool,
+        mut opts: DeleteOptions,
+        namespace_owner: Option<Arc<dyn Send + Sync>>,
+    ) -> Result<()> {
+        validate_delete_recovery_options(&opts)?;
+        if path.starts_with('/') {
+            return Err(DiskError::InvalidPath);
+        }
+        opts.recoverable_delete = true;
+        match self {
+            Self::Local(disk) => {
+                disk.delete_version_with_namespace_owner(volume, path, fi, force_del_marker, opts, namespace_owner)
+                    .await
+            }
+            Self::Remote(disk) => {
+                disk.delete_version_recoverable(volume, path, fi, force_del_marker, opts)
+                    .await
+            }
+        }
+    }
     pub(crate) async fn set_disk_id_state(&self, id: Option<Uuid>) -> Result<()> {
         match self {
             Disk::Local(local_disk) => {
@@ -257,6 +357,70 @@ impl Disk {
 
 #[async_trait::async_trait]
 impl DiskAPI for Disk {
+    async fn delete_version_recoverable(
+        &self,
+        volume: &str,
+        object: &str,
+        file_info: FileInfo,
+        force_del_marker: bool,
+        opts: DeleteOptions,
+    ) -> Result<()> {
+        self.delete_version_recoverable_with_namespace_owner(volume, object, file_info, force_del_marker, opts, None)
+            .await
+    }
+    async fn delete_versions_recoverable(
+        &self,
+        volume: &str,
+        versions: Vec<FileInfoVersions>,
+        opts: DeleteOptions,
+    ) -> Vec<Option<DiskError>> {
+        match self {
+            Self::Local(disk) => disk.delete_versions_recoverable(volume, versions, opts).await,
+            Self::Remote(disk) => disk.delete_versions_recoverable(volume, versions, opts).await,
+        }
+    }
+    async fn discover_delete_transactions(
+        &self,
+        volume: &str,
+        object: Option<&str>,
+        cursor: Option<DeleteRecoveryCursor>,
+        limit: u32,
+    ) -> Result<DeleteRecoveryPage> {
+        match self {
+            Self::Local(disk) => disk.discover_delete_transactions(volume, object, cursor, limit).await,
+            Self::Remote(disk) => disk.discover_delete_transactions(volume, object, cursor, limit).await,
+        }
+    }
+    async fn inspect_delete_transaction(
+        &self,
+        volume: &str,
+        object: &str,
+        transaction: Uuid,
+    ) -> Result<DeleteRecoveryInspection> {
+        match self {
+            Self::Local(disk) => disk.inspect_delete_transaction(volume, object, transaction).await,
+            Self::Remote(disk) => disk.inspect_delete_transaction(volume, object, transaction).await,
+        }
+    }
+    async fn settle_delete_transaction(
+        &self,
+        volume: &str,
+        object: &str,
+        transaction: Uuid,
+        expected_fingerprint: DeleteRecoveryDigest,
+        action: DeleteRecoveryAction,
+    ) -> Result<DeleteRecoveryInspection> {
+        match self {
+            Self::Local(disk) => {
+                disk.settle_delete_transaction(volume, object, transaction, expected_fingerprint, action)
+                    .await
+            }
+            Self::Remote(disk) => {
+                disk.settle_delete_transaction(volume, object, transaction, expected_fingerprint, action)
+                    .await
+            }
+        }
+    }
     fn to_string(&self) -> String {
         match self {
             Disk::Local(local_disk) => local_disk.to_string(),
@@ -1246,6 +1410,51 @@ pub trait DiskAPI: Debug + Send + Sync + 'static {
         Err(DiskError::MethodNotAllowed)
     }
     async fn delete(&self, volume: &str, path: &str, opt: DeleteOptions) -> Result<()>;
+    async fn discover_delete_transactions(
+        &self,
+        _volume: &str,
+        _object: Option<&str>,
+        _cursor: Option<DeleteRecoveryCursor>,
+        _limit: u32,
+    ) -> Result<DeleteRecoveryPage> {
+        Err(DiskError::MethodNotAllowed)
+    }
+    async fn delete_version_recoverable(
+        &self,
+        _volume: &str,
+        _object: &str,
+        _file_info: FileInfo,
+        _force_del_marker: bool,
+        _opts: DeleteOptions,
+    ) -> Result<()> {
+        Err(DiskError::MethodNotAllowed)
+    }
+    async fn delete_versions_recoverable(
+        &self,
+        _volume: &str,
+        versions: Vec<FileInfoVersions>,
+        _opts: DeleteOptions,
+    ) -> Vec<Option<DiskError>> {
+        vec![Some(DiskError::MethodNotAllowed); versions.len()]
+    }
+    async fn inspect_delete_transaction(
+        &self,
+        _volume: &str,
+        _object: &str,
+        _transaction: Uuid,
+    ) -> Result<DeleteRecoveryInspection> {
+        Err(DiskError::MethodNotAllowed)
+    }
+    async fn settle_delete_transaction(
+        &self,
+        _volume: &str,
+        _object: &str,
+        _transaction: Uuid,
+        _expected_fingerprint: DeleteRecoveryDigest,
+        _action: DeleteRecoveryAction,
+    ) -> Result<DeleteRecoveryInspection> {
+        Err(DiskError::MethodNotAllowed)
+    }
     // VerifyFile
     async fn verify_file(&self, volume: &str, path: &str, fi: &FileInfo) -> Result<CheckPartsResp>;
     // CheckParts
@@ -1568,6 +1777,14 @@ pub struct RenameDataResp {
     /// "cannot vote", never as `Absent`.
     #[serde(default)]
     pub old_current_size: Option<OldCurrentSize>,
+    /// Whether this disk decoded enough of the overwritten current version to
+    /// decide if a transitioned source has to be returned.
+    #[serde(default)]
+    pub old_current_source_checked: bool,
+    /// Transitioned old-current source used by PUT free-version cleanup receipt.
+    /// Empty when the old current is not transitioned or no source was known.
+    #[serde(default)]
+    pub old_current_source: Option<FileInfo>,
 }
 
 #[derive(Debug, Clone, Default, Serialize, Deserialize)]
@@ -1582,6 +1799,269 @@ pub struct DeleteOptions {
     /// Remote calls carrying it must use DeleteRetiredMarker, never DeleteVersion.
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub expected_delete_marker: Option<rustfs_filemeta::MetaDeleteMarker>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub delete_recovery_bucket_incarnation: Option<Uuid>,
+    #[serde(default)]
+    pub recoverable_delete: bool,
+}
+
+pub const DELETE_RECOVERY_PROTOCOL_VERSION: u32 = 1;
+pub(crate) fn get_delete_recovery_timeout() -> Result<std::time::Duration> {
+    let value = std::env::var("RUSTFS_DELETE_RECOVERY_TIMEOUT_SECS");
+    match value {
+        Ok(value) => parse_delete_recovery_timeout(Some(&value)),
+        Err(std::env::VarError::NotPresent) => parse_delete_recovery_timeout(None),
+        Err(std::env::VarError::NotUnicode(_)) => Err(DiskError::other("invalid delete recovery timeout configuration")),
+    }
+}
+
+fn parse_delete_recovery_timeout(value: Option<&str>) -> Result<std::time::Duration> {
+    let seconds = match value {
+        None => 900,
+        Some(value) => value
+            .parse::<u64>()
+            .ok()
+            .filter(|seconds| (30..=86400).contains(seconds))
+            .ok_or_else(|| DiskError::other("delete recovery timeout must be an integer between 30 and 86400 seconds"))?,
+    };
+    Ok(std::time::Duration::from_secs(seconds))
+}
+
+pub(crate) fn validate_delete_recovery_options(opts: &DeleteOptions) -> Result<()> {
+    if opts.old_data_dir.is_none_or(|id| id.is_nil())
+        || opts.undo_write
+        || opts.undo_delete
+        || opts.recursive
+        || opts.immediate
+        || opts.expected_delete_marker.is_some()
+        || opts.delete_recovery_bucket_incarnation.is_some_and(|id| id.is_nil())
+    {
+        return Err(DiskError::FileCorrupt);
+    }
+    Ok(())
+}
+pub type DeleteRecoveryDigest = [u8; 32];
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub struct DeleteTransactionId {
+    pub object: String,
+    pub transaction: Uuid,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub struct DeleteRecoveryCursor {
+    pub token: Uuid,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub struct DeleteRecoveryPage {
+    pub transactions: Vec<DeleteTransactionId>,
+    pub next: Option<DeleteRecoveryCursor>,
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub struct DeleteRecoveryJournal {
+    pub protocol_version: u32,
+    pub volume: String,
+    pub object: String,
+    pub transaction: Uuid,
+    pub bucket_incarnation: Option<Uuid>,
+    pub original_digest: Option<DeleteRecoveryDigest>,
+    pub target_digest: Option<DeleteRecoveryDigest>,
+    pub data_dirs: Vec<Uuid>,
+    pub deleted_versions: Vec<Uuid>,
+    pub requests: Vec<FileInfo>,
+    pub original_versions: Vec<DeleteRecoveryVersionIdentity>,
+    #[serde(default)]
+    pub predecessors: Vec<Uuid>,
+}
+
+impl DeleteRecoveryJournal {
+    /// Checks source generations only; the coordinator must supply a certified
+    /// decision and the receiver must retain its namespace/incarnation fences.
+    pub(crate) fn matches_adoption_source(&self, current: &[DeleteRecoveryVersionIdentity]) -> bool {
+        if self.requests.is_empty() {
+            return false;
+        }
+        let mut originals = std::collections::HashMap::with_capacity(self.original_versions.len());
+        for version in &self.original_versions {
+            originals.entry(version.version_id.unwrap_or_default()).or_insert(version);
+        }
+        let mut locals = std::collections::HashMap::with_capacity(current.len());
+        for version in current {
+            locals.entry(version.version_id.unwrap_or_default()).or_insert(version);
+        }
+        self.requests.iter().all(|request| {
+            if request.deleted && request.mark_deleted {
+                return current.first() == self.original_versions.first();
+            }
+            let id = request.version_id.unwrap_or_default();
+            let Some(original) = originals.get(&id) else {
+                return false;
+            };
+            match locals.get(&id) {
+                Some(local) => local == original,
+                None => !id.is_nil(),
+            }
+        })
+    }
+
+    /// Matches immutable local preparation intent against an independently
+    /// inspected same-transaction commit receipt. This match is not authority
+    /// by itself; the coordinator must supply the certified receipt.
+    pub(crate) fn matches_prepared_authority(&self, authority: &Self) -> bool {
+        if self.protocol_version != DELETE_RECOVERY_PROTOCOL_VERSION
+            || authority.protocol_version != DELETE_RECOVERY_PROTOCOL_VERSION
+            || self.transaction.is_nil()
+            || self.transaction != authority.transaction
+            || self.volume != authority.volume
+            || self.object != authority.object
+            || self.bucket_incarnation != authority.bucket_incarnation
+            || self.bucket_incarnation.is_some_and(|id| id.is_nil())
+            || self.requests.is_empty()
+            || self.requests != authority.requests
+            || self.deleted_versions.iter().collect::<std::collections::HashSet<_>>()
+                != authority.deleted_versions.iter().collect::<std::collections::HashSet<_>>()
+        {
+            return false;
+        }
+        let sources: std::collections::HashMap<_, _> = self
+            .original_versions
+            .iter()
+            .map(|version| (version.version_id.unwrap_or_default(), version))
+            .collect();
+        let certified: std::collections::HashMap<_, _> = authority
+            .original_versions
+            .iter()
+            .map(|version| (version.version_id.unwrap_or_default(), version))
+            .collect();
+        if sources.len() != self.original_versions.len() || certified.len() != authority.original_versions.len() {
+            return false;
+        }
+        self.requests.iter().all(|request| {
+            let (local, remote) = if request.deleted && request.mark_deleted {
+                (self.original_versions.first(), authority.original_versions.first())
+            } else {
+                let id = request.version_id.unwrap_or_default();
+                (sources.get(&id).copied(), certified.get(&id).copied())
+            };
+            match (local, remote) {
+                (Some(local), Some(remote)) => local == remote,
+                (None, None) => self.original_digest.is_none() && authority.original_digest.is_none(),
+                _ => false,
+            }
+        })
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub struct DeleteRecoveryVersionIdentity {
+    pub version_id: Option<Uuid>,
+    pub mod_time: Option<OffsetDateTime>,
+    pub data_dir: Option<Uuid>,
+    pub deleted: bool,
+    pub metadata_digest: DeleteRecoveryDigest,
+}
+
+impl DeleteRecoveryVersionIdentity {
+    pub(crate) fn from_file_info(info: FileInfo) -> error::Result<Self> {
+        local::delete_recovery::generation(info)
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+pub enum DeleteRecoveryState {
+    Absent,
+    NewPrepared,
+    QuorumCommitted,
+    CommitFinished,
+    LegacyAmbiguous,
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub struct DeleteRecoveryInspection {
+    pub protocol_version: u32,
+    pub id: DeleteTransactionId,
+    pub state: DeleteRecoveryState,
+    pub journal: Option<DeleteRecoveryJournal>,
+    pub current_digest: Option<DeleteRecoveryDigest>,
+    pub current_versions: Vec<DeleteRecoveryVersionIdentity>,
+    pub backup_digest: Option<DeleteRecoveryDigest>,
+    pub rollback_data_complete: bool,
+    pub commit_data_safe: bool,
+    pub explicit_delete_safe: bool,
+    pub fingerprint: DeleteRecoveryDigest,
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub enum DeleteRecoveryAction {
+    PublishCommit,
+    FinishCommit,
+    FinalizeCommit,
+    Rollback,
+    DiscardLegacy,
+    AdoptCommit { authority: Box<DeleteRecoveryJournal> },
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub enum DeleteRecoveryRpcOperation {
+    Discover {
+        object: Option<String>,
+        cursor: Option<DeleteRecoveryCursor>,
+        limit: u32,
+    },
+    Inspect {
+        object: String,
+        transaction: Uuid,
+    },
+    Settle {
+        object: String,
+        transaction: Uuid,
+        expected_fingerprint: DeleteRecoveryDigest,
+        action: DeleteRecoveryAction,
+    },
+    ApplyVersion {
+        object: String,
+        file_info: Box<FileInfo>,
+        force_del_marker: bool,
+        opts: DeleteOptions,
+    },
+    ApplyVersions {
+        versions: Vec<FileInfoVersions>,
+        opts: DeleteOptions,
+    },
+}
+
+impl DeleteRecoveryRpcOperation {
+    /// Full metadata application and adoption share the existing mutation envelope.
+    pub fn max_wire_payload_bytes(&self) -> usize {
+        match self {
+            Self::ApplyVersion { .. }
+            | Self::ApplyVersions { .. }
+            | Self::Settle {
+                action: DeleteRecoveryAction::AdoptCommit { .. },
+                ..
+            } => rustfs_protos::DELETE_RECOVERY_MAX_PAYLOAD_BYTES,
+            _ => rustfs_protos::DELETE_RECOVERY_MAX_CONTROL_PAYLOAD_BYTES,
+        }
+    }
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+pub enum DeleteRecoveryRpcReply {
+    Discovered(DeleteRecoveryPage),
+    Inspected(DeleteRecoveryInspection),
+    Settled(DeleteRecoveryInspection),
+    AppliedVersion,
+    AppliedVersions(Vec<Option<u32>>),
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
@@ -1872,6 +2352,8 @@ mod tests {
             undo_delete: false,
             old_data_dir: Some(Uuid::new_v4()),
             expected_delete_marker: None,
+            delete_recovery_bucket_incarnation: None,
+            recoverable_delete: false,
         };
 
         assert!(opts.recursive);
@@ -2044,6 +2526,8 @@ mod tests {
             cleanup_data_dir: Some(uuid),
             sign: Some(signature.clone()),
             old_current_size: Some(OldCurrentSize::Present(42)),
+            old_current_source_checked: false,
+            old_current_source: None,
         };
 
         assert_eq!(resp.old_data_dir, Some(uuid));
@@ -2064,6 +2548,8 @@ mod tests {
                 cleanup_data_dir: Some(Uuid::new_v4()),
                 sign: Some(vec![0x01, 0x02, 0x03]),
                 old_current_size,
+                old_current_source_checked: false,
+                old_current_source: None,
             };
 
             let encoded = rmp_serde::encode::to_vec_named(&resp).expect("named msgpack should encode");
@@ -2101,6 +2587,8 @@ mod tests {
         assert_eq!(decoded.cleanup_data_dir, None);
         assert_eq!(decoded.sign, legacy.sign);
         assert_eq!(decoded.old_current_size, None);
+        assert!(!decoded.old_current_source_checked);
+        assert!(decoded.old_current_source.is_none());
     }
 
     /// Test constants
@@ -2296,3 +2784,183 @@ mod tests {
         }
     }
 }
+
+#[cfg(test)]
+mod delete_recovery_timeout_tests {
+    use super::*;
+    #[test]
+    fn recovery_timeout_defaults_and_validates_range() {
+        assert_eq!(
+            parse_delete_recovery_timeout(None).expect("default timeout"),
+            std::time::Duration::from_secs(900)
+        );
+        for seconds in ["30", "900", "86400"] {
+            assert!(parse_delete_recovery_timeout(Some(seconds)).is_ok());
+        }
+        for value in ["", "invalid", "0", "2", "29", "86401", "-1", "18446744073709551616"] {
+            assert!(parse_delete_recovery_timeout(Some(value)).is_err());
+        }
+    }
+}
+
+#[cfg(test)]
+mod delete_recovery_wire_limit_tests {
+    use super::*;
+
+    #[test]
+    fn boxed_apply_version_preserves_existing_wire_payload() {
+        #[derive(Serialize)]
+        enum PriorOperation<'a> {
+            ApplyVersion {
+                object: &'a str,
+                file_info: &'a FileInfo,
+                force_del_marker: bool,
+                opts: &'a DeleteOptions,
+            },
+        }
+        let file_info = FileInfo {
+            name: "object".into(),
+            version_id: Some(Uuid::new_v4()),
+            metadata: std::collections::HashMap::from([("metadata".into(), "value".into())]),
+            ..Default::default()
+        };
+        let opts = DeleteOptions {
+            old_data_dir: Some(Uuid::new_v4()),
+            recoverable_delete: true,
+            delete_recovery_bucket_incarnation: Some(Uuid::new_v4()),
+            ..Default::default()
+        };
+        let prior = PriorOperation::ApplyVersion {
+            object: "object",
+            file_info: &file_info,
+            force_del_marker: true,
+            opts: &opts,
+        };
+        let encoded = rmp_serde::to_vec_named(&prior).expect("encode the previous unboxed request shape");
+        let operation = DeleteRecoveryRpcOperation::ApplyVersion {
+            object: "object".into(),
+            file_info: Box::new(file_info.clone()),
+            force_del_marker: true,
+            opts: opts.clone(),
+        };
+        assert_eq!(rmp_serde::to_vec_named(&operation).expect("encode boxed request"), encoded);
+        assert_eq!(
+            serde_json::to_value(&operation).expect("boxed request JSON"),
+            serde_json::to_value(prior).expect("previous request JSON")
+        );
+        let decoded: DeleteRecoveryRpcOperation =
+            rustfs_protos::decode_delete_recovery_payload(&encoded).expect("decode the previous payload");
+        assert_eq!(rmp_serde::to_vec_named(&decoded).expect("re-encode decoded prior request"), encoded);
+    }
+    #[test]
+    fn adoption_and_inspection_reply_support_large_version_journals() {
+        let transaction = Uuid::new_v4();
+        let request = FileInfo {
+            metadata: std::collections::HashMap::from([("metadata".into(), "x".repeat(2048))]),
+            ..Default::default()
+        };
+        let journal = DeleteRecoveryJournal {
+            protocol_version: DELETE_RECOVERY_PROTOCOL_VERSION,
+            volume: "bucket".into(),
+            object: "object".into(),
+            transaction,
+            bucket_incarnation: None,
+            original_digest: None,
+            target_digest: None,
+            data_dirs: Vec::new(),
+            deleted_versions: Vec::new(),
+            requests: vec![request; 1000],
+            original_versions: Vec::new(),
+            predecessors: Vec::new(),
+        };
+        let operation = DeleteRecoveryRpcOperation::Settle {
+            object: "object".into(),
+            transaction,
+            expected_fingerprint: [0; 32],
+            action: DeleteRecoveryAction::AdoptCommit {
+                authority: Box::new(journal.clone()),
+            },
+        };
+        let encoded = rmp_serde::to_vec_named(&operation).expect("encode large adoption");
+        assert!(encoded.len() > rustfs_protos::DELETE_RECOVERY_MAX_CONTROL_PAYLOAD_BYTES);
+        assert!(encoded.len() < operation.max_wire_payload_bytes());
+        let decoded: DeleteRecoveryRpcOperation =
+            rustfs_protos::decode_delete_recovery_payload(&encoded).expect("large adoption can decode");
+        assert_eq!(decoded.max_wire_payload_bytes(), rustfs_protos::DELETE_RECOVERY_MAX_PAYLOAD_BYTES);
+        let reply = DeleteRecoveryRpcReply::Inspected(DeleteRecoveryInspection {
+            protocol_version: DELETE_RECOVERY_PROTOCOL_VERSION,
+            id: DeleteTransactionId {
+                object: "object".into(),
+                transaction,
+            },
+            state: DeleteRecoveryState::QuorumCommitted,
+            journal: Some(journal),
+            current_digest: None,
+            current_versions: Vec::new(),
+            backup_digest: None,
+            rollback_data_complete: false,
+            commit_data_safe: true,
+            explicit_delete_safe: false,
+            fingerprint: [0; 32],
+        });
+        let encoded = rmp_serde::to_vec_named(&reply).expect("encode large inspection");
+        assert!(encoded.len() > rustfs_protos::DELETE_RECOVERY_MAX_CONTROL_PAYLOAD_BYTES);
+        assert!(rustfs_protos::decode_delete_recovery_payload::<DeleteRecoveryRpcReply>(&encoded).is_ok());
+        let inspect = DeleteRecoveryRpcOperation::Inspect {
+            object: "object".into(),
+            transaction,
+        };
+        assert_eq!(inspect.max_wire_payload_bytes(), rustfs_protos::DELETE_RECOVERY_MAX_CONTROL_PAYLOAD_BYTES);
+    }
+}
+
+#[cfg(test)]
+mod delete_recovery_adoption_tests {
+    use super::*;
+
+    #[test]
+    fn source_matching_preserves_non_null_absence_and_rejects_foreign_or_missing_null() {
+        let id = Uuid::new_v4();
+        let source = DeleteRecoveryVersionIdentity {
+            version_id: Some(id),
+            mod_time: None,
+            data_dir: None,
+            deleted: false,
+            metadata_digest: [1; 32],
+        };
+        let mut journal = DeleteRecoveryJournal {
+            protocol_version: DELETE_RECOVERY_PROTOCOL_VERSION,
+            volume: "bucket".into(),
+            object: "object".into(),
+            transaction: Uuid::new_v4(),
+            bucket_incarnation: None,
+            original_digest: None,
+            target_digest: None,
+            data_dirs: Vec::new(),
+            deleted_versions: vec![id],
+            requests: vec![FileInfo {
+                version_id: Some(id),
+                ..Default::default()
+            }],
+            original_versions: vec![source.clone()],
+            predecessors: Vec::new(),
+        };
+        assert!(journal.matches_adoption_source(&[]), "certified non-null removal can already be absent");
+        assert!(journal.matches_adoption_source(std::slice::from_ref(&source)));
+        let mut foreign = source.clone();
+        foreign.metadata_digest = [2; 32];
+        assert!(!journal.matches_adoption_source(std::slice::from_ref(&foreign)));
+        journal.original_versions.push(foreign.clone());
+        assert!(
+            journal.matches_adoption_source(&[source.clone(), foreign.clone()]),
+            "retain original first-match semantics"
+        );
+        assert!(!journal.matches_adoption_source(&[foreign, source]));
+        journal.requests[0].version_id = None;
+        journal.original_versions[0].version_id = None;
+        assert!(!journal.matches_adoption_source(&[]), "null absence does not establish its generation");
+        assert!(journal.matches_adoption_source(std::slice::from_ref(&journal.original_versions[0])));
+        journal.requests.clear();
+        assert!(!journal.matches_adoption_source(&journal.original_versions));
+    }
+}

crates/ecstore/src/io_support/bitrot.rs

@@ -174,10 +174,14 @@ async fn open_reader_source(
     metrics_path: Option<&'static str>,
 ) -> disk::error::Result<Option<BoxedObjectReader>> {
     if let Some(data) = inline_data {
+        #[cfg(test)]
+        crate::set_disk::disk_call_counters::record_shard_open(path, true);
         let mut reader = Cursor::new(data);
         reader.set_position(u64::try_from(offset).map_err(|_| DiskError::FileCorrupt)?);
         Ok(Some(ShardReader::InMemory(reader)))
     } else if let Some(disk) = disk {
+        #[cfg(test)]
+        crate::set_disk::disk_call_counters::record_shard_open(path, false);
         open_disk_reader(disk, bucket, path, offset, length, use_mmap_read, metrics_path)
             .await
             .map(Some)

crates/ecstore/src/object_api/mod.rs

@@ -102,10 +102,14 @@ mod body_cache_hook;
 mod encryption;
 mod hook_slot;
 mod object_mutation_hook;
+mod read_condition;
 mod readers;
 mod types;
 pub(crate) mod write_commit_context;
 
+pub use read_condition::{GetObjectReadCondition, GetObjectReadMetadata, with_get_object_read_condition};
+pub(crate) use read_condition::{get_object_read_condition_is_active, get_object_read_condition_is_terminal};
+
 #[cfg(test)]
 mod persisted_metadata_keys_tests;
 

crates/ecstore/src/object_api/read_condition.rs

@@ -0,0 +1,113 @@
+// 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 rustfs_filemeta::{FileInfo, ObjectPartInfo};
+use std::future::Future;
+use std::sync::Arc;
+use time::OffsetDateTime;
+
+/// Borrowed fields needed to decide whether an object read needs a body.
+pub struct GetObjectReadMetadata<'a> {
+    pub bucket: &'a str,
+    pub object: &'a str,
+    pub etag: Option<&'a str>,
+    pub mod_time: Option<OffsetDateTime>,
+    pub parts: &'a [ObjectPartInfo],
+    pub delete_marker: bool,
+}
+
+impl<'a> GetObjectReadMetadata<'a> {
+    pub(crate) fn from_file_info(bucket: &'a str, object: &'a str, info: &'a FileInfo) -> Self {
+        Self {
+            bucket,
+            object,
+            etag: info.metadata.get("etag").map(String::as_str),
+            mod_time: info.mod_time,
+            parts: &info.parts,
+            delete_marker: info.deleted,
+        }
+    }
+}
+
+/// A pure decision on quorum-validated metadata. Authorization belongs to the caller.
+pub trait GetObjectReadCondition: Send + Sync {
+    fn is_terminal(&self, metadata: GetObjectReadMetadata<'_>) -> bool;
+}
+
+tokio::task_local! {
+    static READ_CONDITION: Arc<dyn GetObjectReadCondition>;
+}
+
+/// Applies a condition to this reader call, without inheriting it into shared producers.
+pub async fn with_get_object_read_condition<F: Future>(condition: Arc<dyn GetObjectReadCondition>, read: F) -> F::Output {
+    READ_CONDITION.scope(condition, read).await
+}
+
+pub(crate) fn get_object_read_condition_is_terminal(metadata: GetObjectReadMetadata<'_>) -> bool {
+    READ_CONDITION
+        .try_with(|condition| condition.is_terminal(metadata))
+        .unwrap_or(false)
+}
+
+pub(crate) fn get_object_read_condition_is_active() -> bool {
+    READ_CONDITION.try_with(|_| ()).is_ok()
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    struct Stop;
+    impl GetObjectReadCondition for Stop {
+        fn is_terminal(&self, _: GetObjectReadMetadata<'_>) -> bool {
+            true
+        }
+    }
+    fn metadata() -> GetObjectReadMetadata<'static> {
+        GetObjectReadMetadata {
+            bucket: "bucket",
+            object: "object",
+            etag: Some("etag"),
+            mod_time: None,
+            parts: &[],
+            delete_marker: false,
+        }
+    }
+
+    #[tokio::test]
+    async fn read_condition_scope_does_not_escape_or_enter_shared_producers() {
+        assert!(!get_object_read_condition_is_terminal(metadata()));
+        with_get_object_read_condition(Arc::new(Stop), async {
+            assert!(get_object_read_condition_is_terminal(metadata()));
+            assert!(
+                !tokio::spawn(async { get_object_read_condition_is_terminal(metadata()) })
+                    .await
+                    .expect("shared producer")
+            );
+        })
+        .await;
+        assert!(!get_object_read_condition_is_terminal(metadata()));
+    }
+
+    #[tokio::test]
+    async fn dropping_a_pending_read_removes_its_condition_scope() {
+        let mut read = Box::pin(with_get_object_read_condition(Arc::new(Stop), async {
+            assert!(get_object_read_condition_is_terminal(metadata()));
+            std::future::pending::<()>().await;
+        }));
+        assert!(futures::poll!(&mut read).is_pending());
+        drop(read);
+        assert!(!get_object_read_condition_is_terminal(metadata()));
+    }
+}

crates/ecstore/src/object_api/readers.rs

@@ -528,6 +528,14 @@ pub struct GetObjectReader {
 }
 
 impl GetObjectReader {
+    pub(crate) fn condition_stopped(object_info: ObjectInfo) -> Self {
+        Self {
+            stream: Box::new(ConditionStoppedBody),
+            object_info,
+            buffered_body: None,
+            body_source: GetObjectBodySource::Unprobed,
+        }
+    }
     /// Builds a fully materialized reader from a cache-coordinated body.
     pub fn from_cache_body(mut object_info: ObjectInfo, body: Bytes) -> Result<Self> {
         object_info.size = i64::try_from(body.len()).map_err(|_| Error::other("cached GET body length exceeds i64::MAX"))?;
@@ -552,6 +560,17 @@ impl GetObjectReader {
     }
 }
 
+struct ConditionStoppedBody;
+
+impl AsyncRead for ConditionStoppedBody {
+    fn poll_read(self: Pin<&mut Self>, _cx: &mut Context<'_>, buf: &mut ReadBuf<'_>) -> Poll<std::io::Result<()>> {
+        if buf.remaining() == 0 {
+            return Poll::Ready(Ok(()));
+        }
+        Poll::Ready(Err(std::io::Error::other("object read condition must be handled before body access")))
+    }
+}
+
 enum ReadTransform {
     // Written but never read by production code: the enclosing struct already
     // carries the same pair as `storage_offset`/`storage_length`. They survive

crates/ecstore/src/object_api/types.rs

@@ -938,6 +938,10 @@ pub struct ObjectOptions {
     /// A genuine target write guard; loss fences alone do not authorize a borrowed write.
     #[doc(hidden)]
     pub write_lock_context: Option<WriteLockContext>,
+    /// In-process delete ownership for late filesystem completion. Never
+    /// serialized or copied into object metadata.
+    #[doc(hidden)]
+    pub delete_mutation_owner: Option<Arc<dyn Send + Sync>>,
     /// Internal read-only inspection must not enqueue metadata or payload repairs.
     pub suppress_read_repair: bool,
     /// Control-plane writers that immediately read or CAS the same namespace
@@ -1158,6 +1162,7 @@ impl std::fmt::Debug for ObjectOptions {
             .field("http_preconditions", &self.http_preconditions.is_some())
             .field("delete_replication", &self.delete_replication.is_some())
             .field("delete_replication_config_snapshot", &self.delete_replication_config_snapshot)
+            .field("delete_mutation_owner", &self.delete_mutation_owner.is_some())
             .field("namespace_lock_fence", &self.namespace_lock_fence.is_some())
             .field("bucket_lifecycle_lock_fence", &self.bucket_lifecycle_lock_fence.is_some())
             .field("replication_request", &self.replication_request)

crates/ecstore/src/runtime/sources.rs

@@ -50,6 +50,7 @@ use s3s::dto::BucketLifecycleConfiguration;
 use s3s::region::Region;
 use tokio::sync::{OwnedRwLockReadGuard, RwLock};
 use tokio_util::sync::CancellationToken;
+#[cfg(test)]
 use tonic::transport::Channel;
 use uuid::Uuid;
 
@@ -251,10 +252,6 @@ pub(crate) async fn root_disk_threshold_for_erasure_disk() -> Option<u64> {
     }
 }
 
-pub(crate) async fn cached_node_channel(addr: &str) -> Option<Channel> {
-    rustfs_common::cached_connection(addr).await
-}
-
 #[cfg(test)]
 pub(crate) async fn cache_test_node_channel(addr: String, channel: Channel) {
     rustfs_common::cache_connection(addr, channel).await;

crates/ecstore/src/services/rebalance/entry.rs

@@ -50,7 +50,7 @@ fn ensure_rebalance_entry_active(cancel: &CancellationToken) -> Result<()> {
     Ok(())
 }
 
-#[cfg(test)]
+#[cfg(any(test, feature = "test-util"))]
 tokio::task_local! {
     static REBALANCE_ENTRY_RUN_FENCE_BARRIER: (Arc<tokio::sync::Notify>, Arc<tokio::sync::Notify>);
 }
@@ -268,7 +268,7 @@ impl ECStore {
             self.rebalance_run_guard(rebalance_id.as_ref(), "rebalance entry")
         })
         .await?;
-        #[cfg(test)]
+        #[cfg(any(test, feature = "test-util"))]
         if let Ok((arrived, release)) = REBALANCE_ENTRY_RUN_FENCE_BARRIER.try_with(Clone::clone) {
             arrived.notify_one();
             release.notified().await;
@@ -836,7 +836,8 @@ pub mod test_util {
         _temp_dirs: Vec<tempfile::TempDir>,
         store: Arc<ECStore>,
         cancel: CancellationToken,
-        barrier: crate::data_movement::SourceCleanupDeleteBarrier,
+        entry_arrived: Arc<tokio::sync::Notify>,
+        entry_release: Arc<tokio::sync::Notify>,
         stop_probe: super::super::control::RebalanceStopWaitProbe,
         entry_task: Option<tokio::task::JoinHandle<Result<RebalanceEntryOutcome>>>,
     }
@@ -897,31 +898,36 @@ pub mod test_util {
                 cached: Some(file_meta),
                 reusable: false,
             };
-            let barrier = crate::data_movement::SourceCleanupDeleteBarrier::install(bucket, object);
+            let entry_arrived = Arc::new(tokio::sync::Notify::new());
+            let entry_release = Arc::new(tokio::sync::Notify::new());
             let entry_store = Arc::clone(&store);
             let entry_set = Arc::clone(&set);
             let entry_cancel = cancel.clone();
-            let entry_task = tokio::spawn(async move {
-                entry_store
-                    .rebalance_entry(
-                        RebalanceEntryTarget {
-                            bucket: bucket.to_string(),
-                            pool_index: 0,
-                        },
-                        entry,
-                        entry_set,
-                        Arc::new(RebalanceBucketConfigs::default()),
-                        Arc::from(rebalance_id),
-                        entry_cancel,
-                    )
-                    .await
-            });
+            let entry_task = tokio::spawn(REBALANCE_ENTRY_RUN_FENCE_BARRIER.scope(
+                (Arc::clone(&entry_arrived), Arc::clone(&entry_release)),
+                async move {
+                    entry_store
+                        .rebalance_entry(
+                            RebalanceEntryTarget {
+                                bucket: bucket.to_string(),
+                                pool_index: 0,
+                            },
+                            entry,
+                            entry_set,
+                            Arc::new(RebalanceBucketConfigs::default()),
+                            Arc::from(rebalance_id),
+                            entry_cancel,
+                        )
+                        .await
+                },
+            ));
 
             Self {
                 _temp_dirs,
                 store,
                 cancel,
-                barrier,
+                entry_arrived,
+                entry_release,
                 stop_probe: super::super::control::RebalanceStopWaitProbe::install(rebalance_id),
                 entry_task: Some(entry_task),
             }
@@ -932,7 +938,9 @@ pub mod test_util {
         }
 
         pub async fn wait_until_entry_paused(&self) {
-            self.barrier.wait_until_paused().await;
+            tokio::time::timeout(std::time::Duration::from_secs(30), self.entry_arrived.notified())
+                .await
+                .expect("the entry should acquire its run fence before pausing");
         }
 
         pub async fn wait_until_admission_cancelled(&self) {
@@ -946,7 +954,7 @@ pub mod test_util {
         }
 
         pub fn release_entry(&self) {
-            self.barrier.release();
+            self.entry_release.notify_one();
         }
 
         pub async fn assert_entry_cancelled(&mut self) {
@@ -964,7 +972,7 @@ pub mod test_util {
 
     impl Drop for PausedRebalanceEntryTestFixture {
         fn drop(&mut self) {
-            self.barrier.release();
+            self.entry_release.notify_one();
             if let Some(task) = self.entry_task.take() {
                 task.abort();
             }

crates/ecstore/src/set_disk/conditional_read_bench.rs

@@ -0,0 +1,63 @@
+// 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.
+
+//! Test-util-only controls for the paired read benchmark's dedicated bucket.
+
+use std::sync::atomic::{AtomicBool, Ordering};
+
+static ACTIVE: AtomicBool = AtomicBool::new(false);
+
+/// Keeps metadata off the cache in the benchmark's isolated bucket so GET
+/// metadata reads exercise the configured slow disk. Body-cache policy
+/// remains intact.
+#[must_use]
+pub struct ConditionalReadBenchmarkMetadataGuard {
+    _private: (),
+}
+
+impl ConditionalReadBenchmarkMetadataGuard {
+    pub fn acquire() -> Self {
+        assert!(
+            ACTIVE
+                .compare_exchange(false, true, Ordering::AcqRel, Ordering::Acquire)
+                .is_ok()
+        );
+        Self { _private: () }
+    }
+}
+
+impl Drop for ConditionalReadBenchmarkMetadataGuard {
+    fn drop(&mut self) {
+        ACTIVE.store(false, Ordering::Release);
+    }
+}
+
+pub(super) fn cache_bypass_applies(bucket: &str) -> bool {
+    bucket == "conditional-bench" && ACTIVE.load(Ordering::Acquire)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn benchmark_metadata_profile_is_scoped_to_its_bucket_and_owner() {
+        assert!(!cache_bypass_applies("conditional-bench"));
+        let owner = ConditionalReadBenchmarkMetadataGuard::acquire();
+        assert!(cache_bypass_applies("conditional-bench"));
+        assert!(!cache_bypass_applies("other-bucket"));
+        drop(owner);
+        assert!(!cache_bypass_applies("conditional-bench"));
+    }
+}

crates/ecstore/src/set_disk/core/metadata_read.rs

@@ -66,6 +66,7 @@ use tokio::task::JoinSet;
 pub(in crate::set_disk) struct MetadataReadResult {
     pub(in crate::set_disk) slots: Vec<MetadataDiskObservation>,
     pub(in crate::set_disk) diagnostics: MetadataFanoutDiagnostics,
+    pub(in crate::set_disk) condition_stopped: bool,
 }
 
 impl MetadataReadResult {
@@ -575,6 +576,7 @@ impl MetadataFanoutDiagnostics {
 
     pub(in crate::set_disk) fn record_quorum_candidate_latency(&self, path: &'static str, read_quorum: usize) {
         if let Some(latency) = self.quorum_candidate_latency(read_quorum) {
+            rustfs_io_metrics::record_get_object_metadata_quorum_result(path, "reached");
             rustfs_io_metrics::record_get_object_quorum_reached_latency(path, latency.as_secs_f64());
         }
     }
@@ -655,6 +657,8 @@ async fn inline_body_miss_reason_from_observations<'a>(
         return Some(GET_METADATA_EARLY_STOP_REASON_DATA_READ_INLINE_BODY_VERIFY);
     };
 
+    #[cfg(test)]
+    crate::set_disk::disk_call_counters::record(object, crate::set_disk::disk_call_counters::KIND_BODY_DECODE, 0);
     match try_read_inline_data_shards_direct(&mut readers, erasure.data_shards, read_length, object_size).await {
         Some(body) if body.len() == object_size => None,
         _ => Some(GET_METADATA_EARLY_STOP_REASON_DATA_READ_INLINE_BODY_VERIFY),
@@ -1078,7 +1082,11 @@ impl SetDisks {
                 0,
             );
         }
-        Ok(MetadataReadResult { slots, diagnostics })
+        Ok(MetadataReadResult {
+            slots,
+            diagnostics,
+            condition_stopped: false,
+        })
     }
 
     #[allow(clippy::too_many_arguments)]
@@ -1121,6 +1129,7 @@ impl SetDisks {
         let mut next_fanout_index = 0usize;
         let mut scheduled_count = 0usize;
         let mut force_full_wait = false;
+        let read_condition_active = read_data && crate::object_api::get_object_read_condition_is_active();
         let mut final_miss_reason_override = None;
         let mut non_inline_candidate_eligible = None;
         let mut single_pending_hedge_deadline = None;
@@ -1258,12 +1267,28 @@ impl SetDisks {
                 }
             }
 
-            if !force_full_wait
+            if (!force_full_wait || read_condition_active)
                 && let Some(decision) = accumulator
                     .early_stop_decision()
                     .or_else(|| accumulator.version_early_stop_decision())
             {
-                let should_return_early = if read_data {
+                let condition_stopped = read_condition_active
+                    && decision.reason == crate::diagnostics::get::GET_METADATA_EARLY_STOP_REASON_VALID_QUORUM
+                    && accumulator.candidate.as_ref().is_some_and(|candidate| {
+                        !candidate.deleted
+                            && crate::object_api::get_object_read_condition_is_terminal(
+                                crate::object_api::GetObjectReadMetadata::from_file_info(
+                                    bucket.as_ref(),
+                                    object.as_ref(),
+                                    candidate,
+                                ),
+                            )
+                    });
+                let should_return_early = if condition_stopped {
+                    true
+                } else if force_full_wait {
+                    false
+                } else if read_data {
                     match accumulator.candidate.as_ref() {
                         Some(_candidate) if non_inline_candidate_eligible == Some(true) => {
                             accumulator.candidate_has_read_reserve()
@@ -1332,7 +1357,11 @@ impl SetDisks {
                         cancelled_count,
                     );
                     let diagnostics = MetadataFanoutDiagnostics::new(fanout_start.elapsed(), observations);
-                    return Ok(MetadataReadResult { slots, diagnostics });
+                    return Ok(MetadataReadResult {
+                        slots,
+                        diagnostics,
+                        condition_stopped,
+                    });
                 }
             }
 
@@ -1410,7 +1439,11 @@ impl SetDisks {
         rustfs_io_metrics::record_get_object_metadata_early_stop_saved_responses(metrics_path, 0);
         rustfs_io_metrics::record_get_object_metadata_fanout_lifecycle(metrics_path, scheduled_count, scheduled_count, 0);
         let diagnostics = MetadataFanoutDiagnostics::new(fanout_start.elapsed(), observations);
-        Ok(MetadataReadResult { slots, diagnostics })
+        Ok(MetadataReadResult {
+            slots,
+            diagnostics,
+            condition_stopped: false,
+        })
     }
 
     /// Test-only seam that records one per-disk `read_version` metadata RPC for

crates/ecstore/src/set_disk/mod.rs

@@ -166,7 +166,7 @@ use sha2::Sha256;
 use std::hash::{BuildHasher, Hash, Hasher};
 use std::mem::{self};
 use std::pin::Pin;
-use std::sync::atomic::{AtomicU64, Ordering};
+use std::sync::atomic::{AtomicBool, AtomicU64, Ordering};
 use std::sync::{Arc, OnceLock};
 use std::task::{Context, Poll};
 use std::time::{Instant, SystemTime, UNIX_EPOCH};
@@ -410,6 +410,12 @@ impl Drop for ObjectLockDiagGuard {
         }
 
         let hold = self.acquired_at.elapsed();
+        if self.op == "put_object_commit" && self.mode == "write" && rustfs_io_metrics::put_stage_metrics_enabled() {
+            rustfs_io_metrics::record_put_object_stage_duration(
+                rustfs_io_metrics::PUT_STAGE_PUT_OBJECT_COMMIT_NAMESPACE_LOCK_HELD,
+                hold.as_secs_f64() * 1000.0,
+            );
+        }
         record_object_lock_diag_hold_duration(self.op, self.mode, hold);
         let threshold = get_object_lock_diag_slow_hold_threshold();
         if hold >= threshold {
@@ -793,6 +799,11 @@ const DEFAULT_RUSTFS_GET_METADATA_TWO_PHASE_READ_PLAN_ENABLE: bool = false;
 const ENV_RUSTFS_GET_METADATA_EARLY_STOP_BOUNDED_FANOUT: &str = "RUSTFS_GET_METADATA_EARLY_STOP_BOUNDED_FANOUT";
 const DEFAULT_RUSTFS_GET_METADATA_EARLY_STOP_BOUNDED_FANOUT: bool = true;
 
+#[cfg(feature = "test-util")]
+mod conditional_read_bench;
+#[cfg(feature = "test-util")]
+pub use conditional_read_bench::ConditionalReadBenchmarkMetadataGuard;
+
 const ENV_RUSTFS_GET_METADATA_SLOWTAIL_FAULT_DELAY_MS: &str = "RUSTFS_GET_METADATA_SLOWTAIL_FAULT_DELAY_MS";
 const ENV_RUSTFS_GET_METADATA_SLOWTAIL_FAULT_DISKS: &str = "RUSTFS_GET_METADATA_SLOWTAIL_FAULT_DISKS";
 const ENV_RUSTFS_GET_METADATA_SLOWTAIL_FAULT_BUCKET: &str = "RUSTFS_GET_METADATA_SLOWTAIL_FAULT_BUCKET";
@@ -874,6 +885,7 @@ mod ctx;
 mod metadata;
 mod ops;
 pub(crate) use ops::bucket::BucketInfoQuorum;
+pub(crate) use ops::delete_recovery::{DeleteRecoveryContext, DeleteRecoverySummary};
 pub(crate) use ops::heal::AbsenceProofRequest;
 #[cfg(test)]
 pub(crate) use ops::heal::DanglingDeleteFailure;
@@ -918,16 +930,22 @@ pub use ops::heal_walk::HealWalkVersion;
 pub(in crate::set_disk) struct GetObjectFileInfo {
     owned: Option<OwnedGetObjectFileInfo>,
     shared: Option<Arc<GetObjectMetadataCacheEntry>>,
+    condition_stopped: bool,
+    late_metadata_fanout: Option<LateMetadataFanout>,
 }
 
 struct OwnedGetObjectFileInfo {
     fi: FileInfo,
     parts_metadata: Vec<FileInfo>,
     online_disks: Vec<Option<DiskStore>>,
-    late_metadata_fanout_disks: Option<Vec<Option<DiskStore>>>,
 }
 
-type OwnedGetObjectFileInfoParts = (FileInfo, Vec<FileInfo>, Vec<Option<DiskStore>>, Option<Vec<Option<DiskStore>>>);
+enum LateMetadataFanout {
+    CapturedDisks(Vec<Option<DiskStore>>),
+    RefreshCurrentSet,
+}
+
+type OwnedGetObjectFileInfoParts = (FileInfo, Vec<FileInfo>, Vec<Option<DiskStore>>, Option<LateMetadataFanout>);
 
 impl GetObjectFileInfo {
     fn owned(fi: FileInfo, parts_metadata: Vec<FileInfo>, online_disks: Vec<Option<DiskStore>>) -> Self {
@@ -936,9 +954,10 @@ impl GetObjectFileInfo {
                 fi,
                 parts_metadata,
                 online_disks,
-                late_metadata_fanout_disks: None,
             }),
             shared: None,
+            condition_stopped: false,
+            late_metadata_fanout: None,
         }
     }
 
@@ -948,24 +967,27 @@ impl GetObjectFileInfo {
         online_disks: Vec<Option<DiskStore>>,
         late_metadata_fanout_disks: Vec<Option<DiskStore>>,
     ) -> Self {
-        Self {
-            owned: Some(OwnedGetObjectFileInfo {
-                fi,
-                parts_metadata,
-                online_disks,
-                late_metadata_fanout_disks: Some(late_metadata_fanout_disks),
-            }),
-            shared: None,
-        }
+        let mut snapshot = Self::owned(fi, parts_metadata, online_disks);
+        snapshot.late_metadata_fanout = Some(LateMetadataFanout::CapturedDisks(late_metadata_fanout_disks));
+        snapshot
     }
 
     fn shared(entry: Arc<GetObjectMetadataCacheEntry>) -> Self {
+        let condition_stopped = entry.capability == GetObjectMetadataCacheCapability::ConditionOnly;
         Self {
             owned: None,
             shared: Some(entry),
+            condition_stopped,
+            late_metadata_fanout: None,
         }
     }
 
+    fn shared_with_late_metadata_refresh(entry: Arc<GetObjectMetadataCacheEntry>) -> Self {
+        let mut snapshot = Self::shared(entry);
+        snapshot.late_metadata_fanout = Some(LateMetadataFanout::RefreshCurrentSet);
+        snapshot
+    }
+
     fn fi(&self) -> &FileInfo {
         match (&self.owned, &self.shared) {
             (Some(snapshot), None) => &snapshot.fi,
@@ -991,9 +1013,7 @@ impl GetObjectFileInfo {
     }
 
     fn has_late_metadata_fanout(&self) -> bool {
-        self.owned
-            .as_ref()
-            .is_some_and(|snapshot| snapshot.late_metadata_fanout_disks.is_some())
+        self.late_metadata_fanout.is_some()
     }
 
     fn into_owned(self) -> (FileInfo, Vec<FileInfo>, Vec<Option<DiskStore>>) {
@@ -1002,16 +1022,22 @@ impl GetObjectFileInfo {
     }
 
     fn into_owned_with_late_metadata_fanout(self) -> OwnedGetObjectFileInfoParts {
-        match (self.owned, self.shared) {
-            (Some(snapshot), None) => (
-                snapshot.fi,
-                snapshot.parts_metadata,
-                snapshot.online_disks,
-                snapshot.late_metadata_fanout_disks,
-            ),
+        let Self {
+            owned,
+            shared,
+            late_metadata_fanout,
+            ..
+        } = self;
+        match (owned, shared) {
+            (Some(snapshot), None) => (snapshot.fi, snapshot.parts_metadata, snapshot.online_disks, late_metadata_fanout),
             (None, Some(entry)) => match Arc::try_unwrap(entry) {
-                Ok(entry) => (entry.fi, entry.parts_metadata, entry.online_disks, None),
-                Err(entry) => (entry.fi.clone(), entry.parts_metadata.clone(), entry.online_disks.clone(), None),
+                Ok(entry) => (entry.fi, entry.parts_metadata, entry.online_disks, late_metadata_fanout),
+                Err(entry) => (
+                    entry.fi.clone(),
+                    entry.parts_metadata.clone(),
+                    entry.online_disks.clone(),
+                    late_metadata_fanout,
+                ),
             },
             _ => unreachable!("GET metadata snapshot representation must be exclusive"),
         }
@@ -1143,6 +1169,248 @@ mod prepared_get_object_metadata_tests {
         assert!(take_prepared_get_object_metadata().is_none());
     }
 
+    struct StopReadForEtag(String);
+    impl crate::object_api::GetObjectReadCondition for StopReadForEtag {
+        fn is_terminal(&self, metadata: crate::object_api::GetObjectReadMetadata<'_>) -> bool {
+            !metadata.delete_marker && metadata.etag == Some(self.0.as_str())
+        }
+    }
+
+    #[tokio::test]
+    #[serial_test::serial(body_cache_hook)]
+    async fn read_condition_stops_on_quorum_before_slow_disk_and_body_verification() {
+        use crate::object_api::with_get_object_read_condition;
+        for size in [4096, 1_300_000] {
+            let (dirs, set) = make_local_set_disks(4, 2).await;
+            let bucket = "read-condition-quorum";
+            let object = object_with_initial_data_shards(bucket, &format!("read-condition-{size}"));
+            let opts = ObjectOptions {
+                no_lock: true,
+                ..Default::default()
+            };
+            set.make_bucket(bucket, &MakeBucketOptions::default())
+                .await
+                .expect("create bucket");
+            let mut put = PutObjReader::from_vec(vec![b'v'; size]);
+            let info = set.put_object(bucket, &object, &mut put, &opts).await.expect("write object");
+            for dir in &dirs {
+                let object_path = dir.path().join(bucket).join(&object);
+                if size == 4096 {
+                    let path = object_path.join("xl.meta");
+                    let mut bytes = tokio::fs::read(&path).await.expect("read inline metadata");
+                    *bytes.last_mut().expect("inline payload") ^= 1;
+                    tokio::fs::write(path, bytes).await.expect("damage inline data");
+                } else {
+                    tokio::fs::remove_dir_all(object_path.join(info.data_dir.expect("data directory").to_string()))
+                        .await
+                        .expect("remove shards");
+                }
+            }
+            let slow = bounded_initial_parity_disk_index(bucket, &object);
+            let barrier = rename_fanout_barrier::arm(&object, slow, rename_fanout_barrier::PHASE_READ_VERSION);
+            let calls = disk_call_counters::observe(&object);
+            let read_opts = ObjectOptions {
+                metadata_cache_safe: true,
+                ..opts
+            };
+            let followup_opts = read_opts.clone();
+            let condition = Arc::new(StopReadForEtag(info.etag.clone().expect("committed ETag")));
+            let set_for_read = Arc::clone(&set);
+            let object_for_read = object.clone();
+            let mut read = tokio::spawn(async move {
+                with_get_object_read_condition(
+                    condition,
+                    set_for_read.get_object_reader(bucket, &object_for_read, None, HeaderMap::new(), &read_opts),
+                )
+                .await
+            });
+            tokio::time::timeout(READ_VERSION_BARRIER_GUARD, barrier.wait_until_paused())
+                .await
+                .expect("slow metadata read is pending");
+            let mut reader = tokio::time::timeout(READ_VERSION_BARRIER_GUARD, &mut read)
+                .await
+                .expect("condition must not wait for the slow disk or decode damaged data")
+                .expect("reader task")
+                .expect("metadata-only result");
+            assert_eq!(reader.object_info.etag, info.etag);
+            assert_eq!(
+                calls.total(disk_call_counters::KIND_INLINE_SHARD_OPEN),
+                0,
+                "damaged inline data was never opened"
+            );
+            assert_eq!(
+                calls.total(disk_call_counters::KIND_DISK_SHARD_OPEN),
+                0,
+                "missing shards were never opened"
+            );
+            assert_eq!(
+                calls.total(disk_call_counters::KIND_BODY_DECODE),
+                0,
+                "the body verifier and decoder were never entered"
+            );
+            let mut body = Vec::new();
+            assert!(
+                reader.stream.read_to_end(&mut body).await.is_err(),
+                "terminal metadata cannot be served as a successful body"
+            );
+            assert!(body.is_empty());
+            let first_calls = calls.total(disk_call_counters::KIND_READ_VERSION);
+            drop(reader);
+            barrier.release();
+            let snapshot = set
+                .get_object_fileinfo_for_get_object_reader(bucket, &object, &followup_opts, true, true)
+                .await
+                .expect("fresh metadata after terminal scope");
+            assert!(!snapshot.condition_stopped);
+            assert!(
+                calls.total(disk_call_counters::KIND_READ_VERSION) > first_calls,
+                "an incomplete terminal snapshot must not satisfy the next metadata read from cache"
+            );
+        }
+    }
+
+    #[tokio::test]
+    #[serial_test::serial(body_cache_hook)]
+    async fn read_condition_skips_shard_open_and_decode_with_live_positive_controls() {
+        for size in [4096, 1_300_000] {
+            let (_dirs, set) = make_local_set_disks(4, 2).await;
+            let bucket = "read-condition-body-counters";
+            let object = object_with_initial_data_shards(bucket, &format!("body-counters-{size}"));
+            let opts = ObjectOptions {
+                no_lock: true,
+                ..Default::default()
+            };
+            let payload: Vec<_> = (0..size)
+                .map(|index| u8::try_from(index % 251).expect("pattern byte"))
+                .collect();
+            set.make_bucket(bucket, &MakeBucketOptions::default())
+                .await
+                .expect("create bucket");
+            let mut put = PutObjReader::from_vec(payload.clone());
+            let info = set.put_object(bucket, &object, &mut put, &opts).await.expect("write object");
+            let calls = disk_call_counters::observe(&object);
+            let reader = crate::object_api::with_get_object_read_condition(
+                Arc::new(StopReadForEtag(info.etag.expect("committed ETag"))),
+                set.get_object_reader(bucket, &object, None, HeaderMap::new(), &opts),
+            )
+            .await
+            .expect("terminal metadata reader");
+            drop(reader);
+            assert!(calls.total(disk_call_counters::KIND_READ_VERSION) > 0, "metadata really was read");
+            assert_eq!(
+                calls.total(disk_call_counters::KIND_INLINE_SHARD_OPEN),
+                0,
+                "no inline shard source opened"
+            );
+            assert_eq!(calls.total(disk_call_counters::KIND_DISK_SHARD_OPEN), 0, "no disk shard source opened");
+            assert_eq!(
+                calls.total(disk_call_counters::KIND_BODY_DECODE),
+                0,
+                "no body verifier or decoder entered"
+            );
+
+            let mut reader = crate::object_api::with_get_object_read_condition(
+                Arc::new(StopReadForEtag("different-etag".to_owned())),
+                set.get_object_reader(bucket, &object, None, HeaderMap::new(), &opts),
+            )
+            .await
+            .expect("unmatched condition opens the intact object");
+            let mut body = Vec::new();
+            reader.stream.read_to_end(&mut body).await.expect("positive-control body");
+            assert_eq!(body, payload);
+            let source_kind = if size == 4096 {
+                disk_call_counters::KIND_INLINE_SHARD_OPEN
+            } else {
+                disk_call_counters::KIND_DISK_SHARD_OPEN
+            };
+            assert!(
+                calls.total(source_kind) > 0,
+                "the same source counter observes the positive control: size={size}, inline={}, disk={}, decode={}",
+                calls.total(disk_call_counters::KIND_INLINE_SHARD_OPEN),
+                calls.total(disk_call_counters::KIND_DISK_SHARD_OPEN),
+                calls.total(disk_call_counters::KIND_BODY_DECODE),
+            );
+            assert!(
+                calls.total(disk_call_counters::KIND_BODY_DECODE) > 0,
+                "the same decode counter observes the positive control"
+            );
+        }
+    }
+
+    #[tokio::test]
+    #[serial_test::serial(body_cache_hook)]
+    async fn read_condition_rechecks_selected_metadata_before_body_access() {
+        struct Changing(std::sync::atomic::AtomicUsize);
+        impl crate::object_api::GetObjectReadCondition for Changing {
+            fn is_terminal(&self, _: crate::object_api::GetObjectReadMetadata<'_>) -> bool {
+                self.0.fetch_add(1, std::sync::atomic::Ordering::Relaxed) == 0
+            }
+        }
+        let (_dirs, set) = make_local_set_disks(4, 2).await;
+        let bucket = "read-condition-consistency";
+        let object = object_with_initial_data_shards(bucket, "read-condition-consistency");
+        let opts = ObjectOptions {
+            no_lock: true,
+            ..Default::default()
+        };
+        set.make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("create bucket");
+        let mut put = PutObjReader::from_vec(vec![b'x'; 4096]);
+        set.put_object(bucket, &object, &mut put, &opts).await.expect("write object");
+        let result = crate::object_api::with_get_object_read_condition(
+            Arc::new(Changing(std::sync::atomic::AtomicUsize::new(0))),
+            set.get_object_reader(bucket, &object, None, HeaderMap::new(), &opts),
+        )
+        .await;
+        assert!(result.is_err(), "an unverified terminal snapshot must never become a 200 reader");
+    }
+
+    #[tokio::test]
+    #[serial_test::serial(body_cache_hook)]
+    async fn read_condition_miss_preserves_inline_early_stop_and_exact_body() {
+        let (_dirs, set) = make_local_set_disks(4, 2).await;
+        let bucket = "read-condition-miss";
+        let object = object_with_initial_data_shards(bucket, "read-condition-miss");
+        let opts = ObjectOptions {
+            no_lock: true,
+            ..Default::default()
+        };
+        let payload: Vec<u8> = (0..8192)
+            .map(|index| u8::try_from(index % 251).expect("pattern byte"))
+            .collect();
+        set.make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("create bucket");
+        let mut put = PutObjReader::from_vec(payload.clone());
+        set.put_object(bucket, &object, &mut put, &opts)
+            .await
+            .expect("write inline object");
+        let slow = bounded_initial_parity_disk_index(bucket, &object);
+        let barrier = rename_fanout_barrier::arm(&object, slow, rename_fanout_barrier::PHASE_READ_VERSION);
+        let set_for_read = Arc::clone(&set);
+        let object_for_read = object.clone();
+        let mut read = tokio::spawn(async move {
+            crate::object_api::with_get_object_read_condition(
+                Arc::new(StopReadForEtag("other".to_owned())),
+                set_for_read.get_object_reader(bucket, &object_for_read, None, HeaderMap::new(), &opts),
+            )
+            .await
+        });
+        tokio::time::timeout(READ_VERSION_BARRIER_GUARD, barrier.wait_until_paused())
+            .await
+            .expect("slow metadata read is pending");
+        let mut reader = tokio::time::timeout(READ_VERSION_BARRIER_GUARD, &mut read)
+            .await
+            .expect("condition miss preserves the normal early stop")
+            .expect("reader task")
+            .expect("reader");
+        let mut bytes = Vec::new();
+        reader.stream.read_to_end(&mut bytes).await.expect("normal body");
+        assert_eq!(bytes, payload);
+        barrier.release();
+    }
+
     #[test]
     fn cache_hit_consumers_release_snapshot_at_legacy_ownership() {
         let fi = FileInfo {
@@ -1151,6 +1419,7 @@ mod prepared_get_object_metadata_tests {
         };
         let cached = Arc::new(GetObjectMetadataCacheEntry {
             created_at: Instant::now(),
+            capability: GetObjectMetadataCacheCapability::Full,
             parts_metadata: vec![fi.clone()],
             fi,
             online_disks: vec![None],
@@ -1159,11 +1428,11 @@ mod prepared_get_object_metadata_tests {
         let snapshot = GetObjectFileInfo::shared(Arc::clone(&cached));
 
         assert!(snapshot.has_valid_representation());
-        assert!(std::mem::size_of::<GetObjectFileInfo>() >= std::mem::size_of::<OwnedGetObjectFileInfo>());
-        assert!(
-            std::mem::size_of::<GetObjectFileInfo>()
-                <= std::mem::size_of::<OwnedGetObjectFileInfo>() + 2 * std::mem::size_of::<usize>()
-        );
+        // Keep the original payload budget after moving the captured disks out
+        // of OwnedGetObjectFileInfo.
+        let owned_payload_size = std::mem::size_of::<(OwnedGetObjectFileInfo, Option<Vec<Option<DiskStore>>>)>();
+        assert!(std::mem::size_of::<GetObjectFileInfo>() >= owned_payload_size);
+        assert!(std::mem::size_of::<GetObjectFileInfo>() <= owned_payload_size + 2 * std::mem::size_of::<usize>());
         assert_eq!(Arc::strong_count(&cached), 2, "a cache hit must add one snapshot reference");
         assert_eq!(snapshot.fi().name, "object");
         assert_eq!(snapshot.parts_metadata().len(), 1);
@@ -4083,6 +4352,9 @@ pub struct SetDisks {
     /// writes skip the global registry mutex (backlog#1315). `Arc` so clones of
     /// a set share one generation marker.
     capacity_dirty_generation: Arc<AtomicU64>,
+    /// Per-set proof that successful rename_data peers return the old-current
+    /// source capability needed before PUT may skip its pre-rename lookup.
+    tier_free_version_rename_data_source_capable: Arc<AtomicBool>,
     /// Orphan prefixes whose last purge scan met data that can never be
     /// purged by listing (residue without a committed marker, an in-flight
     /// write), keyed by `bucket/prefix` with the time of that scan. Empty
@@ -4338,6 +4610,14 @@ struct GetObjectMetadataCacheKey {
     hash: u64,
 }
 
+#[derive(Clone, Copy, Debug)]
+struct GetObjectMetadataCacheLookupKey<'a> {
+    bucket: &'a str,
+    object: &'a str,
+    generation: u64,
+    hash: u64,
+}
+
 #[derive(Clone, Copy, Debug, Eq, PartialEq)]
 struct GetObjectMetadataCacheGeneration {
     index: usize,
@@ -4426,6 +4706,33 @@ impl GetObjectMetadataCacheKey {
     }
 }
 
+impl<'a> GetObjectMetadataCacheLookupKey<'a> {
+    fn new(bucket: &'a str, object: &'a str, generation: GetObjectMetadataCacheGeneration) -> Self {
+        Self {
+            bucket,
+            object,
+            generation: generation.value,
+            hash: generation.hash,
+        }
+    }
+}
+
+impl Hash for GetObjectMetadataCacheLookupKey<'_> {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.hash.hash(state);
+        self.generation.hash(state);
+    }
+}
+
+impl moka::Equivalent<GetObjectMetadataCacheKey> for GetObjectMetadataCacheLookupKey<'_> {
+    fn equivalent(&self, key: &GetObjectMetadataCacheKey) -> bool {
+        self.bucket == key.bucket.as_ref()
+            && self.object == key.object.as_ref()
+            && self.generation == key.generation
+            && self.hash == key.hash
+    }
+}
+
 impl Hash for GetObjectMetadataCacheKey {
     fn hash<H: Hasher>(&self, state: &mut H) {
         self.hash.hash(state);
@@ -4433,6 +4740,15 @@ impl Hash for GetObjectMetadataCacheKey {
     }
 }
 
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+enum GetObjectMetadataCacheCapability {
+    Full,
+    // A quorum-selected layout can serve body reads, but is not a mutation layout.
+    QuorumSelected,
+    // Quorum-validated metadata used only to evaluate a terminal read condition.
+    ConditionOnly,
+}
+
 #[derive(Debug)]
 struct GetObjectMetadataCacheEntry {
     #[allow(dead_code)] // Kept for debugging; moka handles TTL internally
@@ -4441,6 +4757,7 @@ struct GetObjectMetadataCacheEntry {
     parts_metadata: Vec<FileInfo>,
     online_disks: Vec<Option<DiskStore>>,
     read_quorum: usize,
+    capability: GetObjectMetadataCacheCapability,
 }
 
 #[derive(Clone, Debug)]
@@ -4971,6 +5288,7 @@ impl SetDisks {
             ctx,
             capacity_scope_cache: Arc::new(std::sync::RwLock::new(CapacityScopeCache::default())),
             capacity_dirty_generation: Arc::new(AtomicU64::new(u64::MAX)),
+            tier_free_version_rename_data_source_capable: Arc::new(AtomicBool::new(false)),
             orphan_purge_backoff: Arc::new(std::sync::Mutex::new(HashMap::new())),
             #[cfg(test)]
             storage_class_config_override: Arc::new(std::sync::RwLock::new(None)),
@@ -4985,6 +5303,15 @@ impl SetDisks {
         &self.ctx
     }
 
+    pub(in crate::set_disk) fn tier_free_version_rename_data_source_capable(&self) -> bool {
+        self.tier_free_version_rename_data_source_capable.load(Ordering::Acquire)
+    }
+
+    pub(in crate::set_disk) fn observe_tier_free_version_rename_data_source_capability(&self, capable: bool) {
+        self.tier_free_version_rename_data_source_capable
+            .store(capable, Ordering::Release);
+    }
+
     #[cfg(test)]
     pub(crate) fn set_instance_ctx_for_test(&mut self, ctx: Arc<InstanceContext>) {
         self.local_lock_manager = ctx.lock_manager();

crates/ecstore/src/set_disk/ops/bucket.rs

@@ -31,6 +31,7 @@ use crate::disk::DiskAPI;
 pub(crate) enum BucketInfoQuorum {
     Read,
     Write,
+    HealCandidates,
 }
 
 impl SetDisks {
@@ -104,7 +105,9 @@ impl SetDisks {
             // class. A namespace read must intersect that majority; object
             // readers still enforce the persisted layout's data-shard quorum.
             BucketInfoQuorum::Read => reduce_read_quorum_errs(&errs, BUCKET_OP_IGNORED_ERRS, disk_count.div_ceil(2).max(1)),
-            BucketInfoQuorum::Write => reduce_write_quorum_errs(&errs, BUCKET_OP_IGNORED_ERRS, disk_count / 2 + 1),
+            BucketInfoQuorum::Write | BucketInfoQuorum::HealCandidates => {
+                reduce_write_quorum_errs(&errs, BUCKET_OP_IGNORED_ERRS, disk_count / 2 + 1)
+            }
         };
         if let Some(err) = error {
             return Err(err.into());
@@ -123,8 +126,42 @@ impl SetDisks {
     }
 
     pub(crate) async fn list_bucket_for_scanner(&self, _opts: &BucketOptions) -> Result<(Vec<BucketInfo>, bool)> {
+        self.list_bucket_with_quorum(BucketInfoQuorum::Write).await
+    }
+
+    pub(crate) async fn bucket_needs_reconstruction(&self, bucket: &str) -> Result<bool> {
+        let results = join_all(self.disk_inventory().await.into_iter().map(|disk| async move {
+            match disk {
+                Some(disk) => disk.stat_volume(bucket).await,
+                None => Err(DiskError::DiskNotFound),
+            }
+        }))
+        .await;
+        let mut present = false;
+        let mut missing = false;
+        for result in results {
+            match result {
+                Ok(_) => present = true,
+                Err(DiskError::VolumeNotFound) => missing = true,
+                Err(error) => return Err(error.into()),
+            }
+        }
+        if !present {
+            return Err(Error::VolumeNotFound);
+        }
+        Ok(missing)
+    }
+
+    pub(crate) async fn list_bucket_with_quorum(&self, quorum: BucketInfoQuorum) -> Result<(Vec<BucketInfo>, bool)> {
         let disks = self.disk_inventory().await;
         let write_quorum = (disks.len() / 2) + 1;
+        // Discovery uses the same namespace read boundary as stat_bucket.
+        // It grants no authority to mutate buckets or accept object reads.
+        let required = match quorum {
+            BucketInfoQuorum::Read => disks.len().div_ceil(2).max(1),
+            BucketInfoQuorum::Write => write_quorum,
+            BucketInfoQuorum::HealCandidates => disks.len().max(1),
+        };
 
         let mut futures = Vec::with_capacity(disks.len());
         for disk in disks {
@@ -153,7 +190,13 @@ impl SetDisks {
             }
         }
 
-        if let Some(err) = reduce_write_quorum_errs(&errs, BUCKET_OP_IGNORED_ERRS, write_quorum) {
+        let error = match quorum {
+            BucketInfoQuorum::Read => reduce_read_quorum_errs(&errs, BUCKET_OP_IGNORED_ERRS, required),
+            BucketInfoQuorum::Write | BucketInfoQuorum::HealCandidates => {
+                reduce_write_quorum_errs(&errs, BUCKET_OP_IGNORED_ERRS, required)
+            }
+        };
+        if let Some(err) = error {
             return Err(err.into());
         }
 
@@ -176,10 +219,17 @@ impl SetDisks {
             }
         }
 
-        topology_complete &= counts.values().all(|(count, _)| *count >= write_quorum);
+        let minimum_presence = if matches!(quorum, BucketInfoQuorum::HealCandidates) {
+            // Candidate presence is not mutation authority. Every disk must answer,
+            // and the heal owner validates the committed generation before recreation.
+            1
+        } else {
+            topology_complete &= counts.values().all(|(count, _)| *count >= write_quorum);
+            required
+        };
         let mut buckets = counts
             .into_values()
-            .filter_map(|(count, bucket)| (count >= write_quorum).then_some(bucket))
+            .filter_map(|(count, bucket)| (count >= minimum_presence).then_some(bucket))
             .collect::<Vec<_>>();
         buckets.sort_by(|left, right| left.name.cmp(&right.name));
         Ok((buckets, topology_complete))

crates/ecstore/src/set_disk/ops/mod.rs

@@ -19,6 +19,7 @@
 
 pub(crate) mod bitrot_self_verify;
 pub(crate) mod bucket;
+pub(crate) mod delete_recovery;
 pub(crate) mod heal;
 pub(crate) mod heal_walk;
 pub(crate) mod list;

crates/ecstore/src/store/bucket.rs

@@ -20,6 +20,7 @@ use crate::bucket::{
 use crate::error::is_err_bucket_not_found;
 use crate::runtime::sources as runtime_sources;
 use crate::set_disk::{BucketInfoQuorum, get_lock_acquire_timeout};
+use crate::storage_api_contracts::admin::DiskSetSelector;
 use crate::storage_api_contracts::bucket::{BUCKET_LIFECYCLE_LOCK_OBJECT, SRBucketDeleteOp};
 use crate::storage_api_contracts::namespace::NamespaceLocking as _;
 use futures::stream::{self, StreamExt};
@@ -1151,7 +1152,7 @@ impl ECStore {
         let mut scoped_results = futures::future::join_all(self.bucket_sets().map(|(pool_index, set_index, set)| async move {
             let result = match quorum {
                 BucketInfoQuorum::Read => set.stat_bucket_with_quorum(bucket, quorum).await,
-                BucketInfoQuorum::Write => set.get_bucket_info(bucket, opts).await,
+                BucketInfoQuorum::Write | BucketInfoQuorum::HealCandidates => set.get_bucket_info(bucket, opts).await,
             };
             (pool_index, set_index, result)
         }))
@@ -1231,34 +1232,70 @@ impl ECStore {
     #[instrument(skip(self))]
     pub(super) async fn handle_list_bucket(&self, opts: &BucketOptions) -> Result<Vec<BucketInfo>> {
         // TODO(backlog): support cached bucket listing via opts.cached
-        Ok(self.list_bucket_from_sets(opts, usize::MAX).await?.buckets)
+        Ok(self
+            .list_bucket_from_sets(opts, usize::MAX, BucketInfoQuorum::Write, None)
+            .await?
+            .buckets)
     }
 
     pub async fn list_bucket_for_scanner(&self, opts: &BucketOptions) -> Result<crate::cluster::rpc::ScannerBucketListing> {
-        self.list_bucket_from_sets(opts, SCANNER_BUCKET_LIST_SET_CONCURRENCY).await
+        self.list_bucket_from_sets(opts, SCANNER_BUCKET_LIST_SET_CONCURRENCY, BucketInfoQuorum::Write, None)
+            .await
+    }
+
+    /// Discover candidates for metadata initialization at each set's namespace
+    /// read quorum. Metadata loading still validates persisted configuration;
+    /// this listing must not be used as a complete scanner or mutation view.
+    pub async fn list_buckets_for_metadata_startup(&self) -> Result<Vec<BucketInfo>> {
+        Ok(self
+            .list_bucket_from_sets(
+                &BucketOptions {
+                    no_metadata: true,
+                    ..Default::default()
+                },
+                SCANNER_BUCKET_LIST_SET_CONCURRENCY,
+                BucketInfoQuorum::Read,
+                None,
+            )
+            .await?
+            .buckets)
+    }
+
+    /// Preserve namespace completeness for administrator traversal. A scoped
+    /// request must not depend on unrelated sets being available.
+    pub async fn list_buckets_for_admin_heal(
+        &self,
+        selector: Option<DiskSetSelector>,
+    ) -> Result<crate::cluster::rpc::ScannerBucketListing> {
+        self.list_bucket_from_sets(
+            &BucketOptions {
+                no_metadata: true,
+                ..Default::default()
+            },
+            SCANNER_BUCKET_LIST_SET_CONCURRENCY,
+            BucketInfoQuorum::HealCandidates,
+            selector,
+        )
+        .await
     }
 
     async fn list_bucket_from_sets(
         &self,
         opts: &BucketOptions,
         max_concurrency: usize,
+        quorum: BucketInfoQuorum,
+        selector: Option<DiskSetSelector>,
     ) -> Result<crate::cluster::rpc::ScannerBucketListing> {
-        let set_count = self.pools.iter().map(|pool| pool.disk_set.len()).sum();
+        let sets = self
+            .bucket_sets()
+            .filter(|(pool, set, _)| selector.is_none_or(|scope| scope.pool_idx == *pool && scope.set_idx == *set))
+            .collect::<Vec<_>>();
+        let set_count = sets.len();
         let concurrency = bucket_list_set_concurrency(set_count, max_concurrency);
-        let deleted = opts.deleted;
-        let cached = opts.cached;
-        let no_metadata = opts.no_metadata;
-        let mut set_listings = stream::iter(self.bucket_sets().map(move |(pool_index, set_index, set)| {
-            let opts = BucketOptions {
-                deleted,
-                cached,
-                no_metadata,
-            };
-            async move {
-                set.list_bucket_for_scanner(&opts)
-                    .await
-                    .map(|(buckets, complete)| (pool_index, set_index, buckets, complete))
-            }
+        let mut set_listings = stream::iter(sets.into_iter().map(move |(pool_index, set_index, set)| async move {
+            set.list_bucket_with_quorum(quorum)
+                .await
+                .map(|(buckets, complete)| (pool_index, set_index, buckets, complete))
         }))
         .buffer_unordered(concurrency);
         let mut topology_complete = set_count != 0;
@@ -2224,6 +2261,64 @@ mod tests {
         }
     }
 
+    #[tokio::test]
+    async fn admin_bucket_discovery_preserves_completeness_and_scope() {
+        use crate::storage_api_contracts::admin::DiskSetSelector;
+        let (temp, store) = setup_bucket_quorum_test_env(&[2, 2], None).await;
+        let empty = store
+            .list_buckets_for_admin_heal(None)
+            .await
+            .expect("empty namespace enumeration");
+        assert!(empty.buckets.is_empty());
+        assert!(empty.topology_complete);
+        for disk in 0..2 {
+            tokio::fs::create_dir(temp.path().join(format!("pool0-disk{disk}/healthy-bucket")))
+                .await
+                .expect("healthy bucket");
+        }
+        tokio::fs::create_dir(temp.path().join("pool1-disk0/partial-bucket"))
+            .await
+            .expect("single surviving bucket directory");
+        let global = store
+            .list_buckets_for_admin_heal(None)
+            .await
+            .expect("online disk enumeration");
+        assert!(global.topology_complete, "every disk answered the candidate inventory");
+        assert_eq!(
+            global.buckets.iter().map(|b| b.name.as_str()).collect::<Vec<_>>(),
+            ["healthy-bucket", "partial-bucket"]
+        );
+        let healthy = store
+            .list_buckets_for_admin_heal(Some(DiskSetSelector::new(0, 0)))
+            .await
+            .expect("healthy scope");
+        assert!(healthy.topology_complete);
+        assert_eq!(healthy.buckets.len(), 1);
+        assert_eq!(healthy.set_buckets.len(), 1);
+        let partial = store
+            .list_buckets_for_admin_heal(Some(DiskSetSelector::new(1, 0)))
+            .await
+            .expect("partial scope");
+        assert!(partial.topology_complete);
+        assert_eq!(partial.buckets[0].name, "partial-bucket");
+        let invalid = store
+            .list_buckets_for_admin_heal(Some(DiskSetSelector::new(0, 1)))
+            .await
+            .expect("unconfigured scope");
+        assert!(!invalid.topology_complete);
+        assert!(invalid.buckets.is_empty());
+        let offline = take_set_disks_offline(&store, &store.pools[1].disk_set[0], &[0]).await;
+        assert!(store.list_buckets_for_admin_heal(None).await.is_err(), "quorum errors must remain errors");
+        assert!(
+            store
+                .list_buckets_for_admin_heal(Some(DiskSetSelector::new(0, 0)))
+                .await
+                .expect("unrelated offline pool")
+                .topology_complete
+        );
+        restore_set_disks(&store, &store.pools[1].disk_set[0], offline).await;
+    }
+
     #[test]
     fn should_not_override_when_metadata_created_is_unix_epoch() {
         assert!(!should_override_created_from_metadata(OffsetDateTime::UNIX_EPOCH));
@@ -3543,6 +3638,79 @@ mod tests {
         assert_eq!(info_err, StorageError::ErasureReadQuorum);
     }
 
+    #[tokio::test]
+    #[serial]
+    async fn metadata_startup_listing_uses_each_sets_read_quorum() {
+        for drives_per_pool in [vec![2], vec![3], vec![4], vec![6], vec![8], vec![4, 6]] {
+            let (_temp_dir, store) = setup_bucket_quorum_test_env(&drives_per_pool, None).await;
+            let bucket = "metadata-startup-read-quorum";
+            for pool in &store.pools {
+                pool.disk_set[0]
+                    .make_bucket(bucket, &MakeBucketOptions::default())
+                    .await
+                    .expect("healthy set should create the startup candidate");
+            }
+            let mut offline = Vec::new();
+            for (pool_index, &width) in drives_per_pool.iter().enumerate() {
+                let set = &store.pools[pool_index].disk_set[0];
+                let indexes = (width.div_ceil(2)..width).collect::<Vec<_>>();
+                offline.push(take_set_disks_offline(&store, set, &indexes).await);
+            }
+            let listing = store
+                .list_buckets_for_metadata_startup()
+                .await
+                .expect("startup must discover buckets at every set's read quorum");
+            assert_eq!(listing.len(), 1);
+            assert_eq!(listing[0].name, bucket);
+            if drives_per_pool.iter().any(|width| width % 2 == 0) {
+                let error = store
+                    .list_bucket_for_scanner(&BucketOptions::default())
+                    .await
+                    .expect_err("startup discovery must not weaken scanner quorum");
+                assert_eq!(error, StorageError::ErasureWriteQuorum);
+                let error = store
+                    .list_bucket(&BucketOptions {
+                        no_metadata: true,
+                        ..Default::default()
+                    })
+                    .await
+                    .expect_err("ordinary listing must retain its existing quorum");
+                assert_eq!(error, StorageError::ErasureWriteQuorum);
+            }
+            // A healthy earlier pool must not hide read-quorum loss in a later pool.
+            let last_pool = store.pools.len() - 1;
+            let set = &store.pools[last_pool].disk_set[0];
+            let extra = take_set_disks_offline(&store, set, &[0]).await;
+            let error = store
+                .list_buckets_for_metadata_startup()
+                .await
+                .expect_err("every set must retain read quorum for startup discovery");
+            assert_eq!(error, StorageError::ErasureReadQuorum);
+            restore_set_disks(&store, set, extra).await;
+            for (pool_index, disks) in offline.into_iter().enumerate() {
+                restore_set_disks(&store, &store.pools[pool_index].disk_set[0], disks).await;
+            }
+        }
+    }
+
+    #[tokio::test]
+    #[serial]
+    async fn metadata_startup_listing_does_not_accept_single_disk_bucket_remnants() {
+        let (temp_dir, store) = setup_bucket_quorum_test_env(&[4], None).await;
+        let bucket = "incomplete-startup-bucket";
+        tokio::fs::create_dir(temp_dir.path().join("pool0-disk0").join(bucket))
+            .await
+            .expect("create one uncommitted directory remnant");
+        let set = &store.pools[0].disk_set[0];
+        let offline = take_set_disks_offline(&store, set, &[2, 3]).await;
+        let listing = store
+            .list_buckets_for_metadata_startup()
+            .await
+            .expect("two responsive disks satisfy discovery quorum");
+        assert!(listing.iter().all(|info| info.name != bucket));
+        restore_set_disks(&store, set, offline).await;
+    }
+
     #[tokio::test]
     #[serial]
     async fn scanner_bucket_listing_unions_every_erasure_set() {

crates/ecstore/src/store/heal.rs

@@ -270,13 +270,151 @@ impl ECStore {
         .map_err(|error| Error::other_with_context("bucket heal owner task failed", error))?
     }
 
+    /// Recover hidden delete journals through the lifecycle-owned heal boundary.
+    pub async fn recover_delete_transactions(self: &Arc<Self>, bucket: &str, opts: &HealOpts) -> Result<HealResultItem> {
+        Ok(self
+            .recover_delete_transactions_at_incarnation(bucket, None, None, opts)
+            .await?
+            .unwrap_or_else(|| HealResultItem {
+                heal_item_type: HealItemType::Bucket.to_string(),
+                bucket: bucket.to_owned(),
+                detail: "delete recovery: discovered=0".to_owned(),
+                ..Default::default()
+            }))
+    }
+
+    pub async fn recover_delete_transactions_at_incarnation(
+        self: &Arc<Self>,
+        bucket: &str,
+        object: Option<&str>,
+        expected: Option<Uuid>,
+        opts: &HealOpts,
+    ) -> Result<Option<HealResultItem>> {
+        self.recover_delete_transactions_in_scope(bucket, object, None, expected, opts)
+            .await
+    }
+
+    pub async fn recover_delete_prefix_at_incarnation(
+        self: &Arc<Self>,
+        bucket: &str,
+        prefix: &str,
+        expected: Option<Uuid>,
+        opts: &HealOpts,
+    ) -> Result<Option<HealResultItem>> {
+        self.recover_delete_transactions_in_scope(bucket, None, Some(prefix), expected, opts)
+            .await
+    }
+
+    async fn recover_delete_transactions_in_scope(
+        self: &Arc<Self>,
+        bucket: &str,
+        object: Option<&str>,
+        prefix: Option<&str>,
+        expected: Option<Uuid>,
+        opts: &HealOpts,
+    ) -> Result<Option<HealResultItem>> {
+        if opts.no_lock && !opts.dry_run {
+            return Err(StorageError::InvalidArgument(
+                "heal".to_string(),
+                "no_lock".to_string(),
+                "delete recovery requires namespace locking".to_string(),
+            ));
+        }
+        let expected = match expected {
+            Some(expected) => expected,
+            None => self.bucket_incarnation_id_from_disk(bucket).await?,
+        };
+        let object = object.map(str::to_owned);
+        let prefix = prefix.map(str::to_owned);
+        self.run_bucket_heal_at_incarnation(bucket, expected, opts, |store, bucket, opts| async move {
+            let scope = bucket_heal_scope(&bucket).ok_or(StorageError::PreconditionFailed)?;
+            let explicit_delete = opts.remove
+                && metadata_sys::has_authoritative_never_versioned_state_in(&store.ctx, &bucket).await?
+                && matches!(
+                    metadata_sys::get_object_lock_config_state_in(&store.ctx, &bucket).await?,
+                    metadata_sys::ObjectLockConfigState::ConfirmedAbsent
+                );
+            let movement_gate = store.ctx.data_movement_operation_gate();
+            let _movement_guard = movement_gate.read().await;
+            let can_commit = || scope.check().is_ok();
+            let recovery_context = crate::set_disk::DeleteRecoveryContext {
+                dry_run: opts.dry_run,
+                explicit_delete,
+                namespace_owner: (!opts.dry_run)
+                    .then(|| Arc::new((scope.clone(), store.ctx.begin_namespace_commit())) as Arc<dyn Send + Sync>),
+                ..crate::set_disk::DeleteRecoveryContext::new(Some(scope.incarnation), &can_commit)
+            };
+            let mut summary = crate::set_disk::DeleteRecoverySummary::default();
+            for pool in store.get_pools_for_heal_object(&opts)? {
+                if store.is_suspended(pool.pool_idx).await || store.is_pool_rebalancing(pool.pool_idx).await {
+                    return Err(StorageError::SlowDown);
+                }
+                for (set_index, set) in pool.disk_set.iter().enumerate() {
+                    if opts.set.is_some_and(|requested| requested != set_index) {
+                        continue;
+                    }
+                    if let Some(object) = &object {
+                        if !Arc::ptr_eq(set, &pool.get_disks_by_key(object)) {
+                            continue;
+                        }
+                        summary.add(set.recover_delete_object(&bucket, object, recovery_context.clone()).await?);
+                    } else if let Some(prefix) = &prefix {
+                        summary.add(set.recover_delete_prefix(&bucket, prefix, recovery_context.clone()).await?);
+                    } else {
+                        summary.add(set.recover_delete_bucket(&bucket, recovery_context.clone()).await?);
+                    }
+                }
+            }
+            if summary.pending > 0 {
+                return Err(StorageError::ErasureReadQuorum);
+            }
+            if object.is_some() && !opts.dry_run && summary.blocked > 0 {
+                return Err(StorageError::SlowDown);
+            }
+            if summary.discovered == 0 {
+                return Ok(None);
+            }
+            Ok(Some(HealResultItem {
+                heal_item_type: if object.is_some() {
+                    HealItemType::Object
+                } else {
+                    HealItemType::Bucket
+                }
+                .to_string(),
+                object: object.or(prefix).unwrap_or_default(),
+                bucket,
+                detail: format!(
+                    "delete recovery{}: discovered={}, committed={}, rolled_back={}, reclaimed={}, blocked={}",
+                    if opts.dry_run { " (dry run)" } else { "" },
+                    summary.discovered,
+                    summary.committed,
+                    summary.rolled_back,
+                    summary.reclaimed,
+                    summary.blocked
+                ),
+                ..Default::default()
+            }))
+        })
+        .await
+    }
+
     pub async fn heal_bucket_at_incarnation(
         self: &Arc<Self>,
         bucket: &str,
         expected: uuid::Uuid,
         opts: &HealOpts,
     ) -> Result<HealResultItem> {
-        self.run_bucket_heal_at_incarnation(bucket, expected, opts, |store, bucket, opts| async move {
+        self.run_bucket_heal_at_incarnation(bucket, expected, opts, move |store, bucket, mut opts| async move {
+            if !opts.recreate && !opts.remove {
+                let mut missing = false;
+                for (pool, set) in store.heal_erasure_set_scopes(&opts).await? {
+                    missing |= store.pools[pool].get_disks(set).bucket_needs_reconstruction(&bucket).await?;
+                }
+                if missing {
+                    crate::bucket::metadata_sys::validate_bucket_reconstruction_in(&store.ctx, &bucket, expected).await?;
+                    opts.recreate = true;
+                }
+            }
             store.heal_bucket(&bucket, &opts).await
         })
         .await
@@ -1323,6 +1461,178 @@ mod tests {
         .expect("minimal pool should build")
     }
 
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn admin_bucket_recovery_recreates_missing_directories_at_current_incarnation() {
+        let (root, store, shutdown) = multi_pool_heal_store().await;
+        let bucket = "admin-bucket-recovery";
+        store
+            .make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("create a committed bucket generation");
+        let incarnation = store
+            .bucket_incarnation_id_from_disk(bucket)
+            .await
+            .expect("persisted generation");
+        for disk in 1..4 {
+            tokio::fs::remove_dir(root.path().join(format!("pool0-disk{disk}/{bucket}")))
+                .await
+                .expect("remove only an empty fixture bucket directory");
+        }
+        store
+            .heal_bucket_at_incarnation(bucket, incarnation, &HealOpts::default())
+            .await
+            .expect("default administrator recovery should use committed generation authority");
+        for disk in 0..4 {
+            assert!(root.path().join(format!("pool0-disk{disk}/{bucket}")).is_dir());
+        }
+        assert_eq!(
+            store
+                .bucket_incarnation_id_from_disk(bucket)
+                .await
+                .expect("recovered generation"),
+            incarnation
+        );
+        shutdown.cancel();
+    }
+
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn admin_bucket_recovery_rejects_missing_corrupt_uncommitted_and_retired_authority() {
+        let (root, store, shutdown) = multi_pool_heal_store().await;
+        for state in ["missing", "corrupt", "uncommitted", "mismatched", "retired"] {
+            let bucket = format!("recovery-authority-{state}");
+            store
+                .make_bucket(&bucket, &MakeBucketOptions::default())
+                .await
+                .expect("create fixture generation");
+            let incarnation = store
+                .bucket_incarnation_id_from_disk(&bucket)
+                .await
+                .expect("fixture identity");
+            let (mut metadata, persisted) = metadata_sys::get_config_from_disk_with_presence_in(&store.ctx, &bucket)
+                .await
+                .expect("committed metadata");
+            assert!(persisted && metadata.bucket_creation_committed);
+            let path = metadata.save_file_path();
+            match state {
+                "missing" => delete_config(store.clone(), &path).await.expect("remove metadata authority"),
+                "corrupt" => save_config(store.clone(), &path, b"corrupt bucket metadata".to_vec())
+                    .await
+                    .expect("corrupt authority"),
+                "uncommitted" | "mismatched" => {
+                    if state == "uncommitted" {
+                        metadata.bucket_creation_committed = false;
+                    } else {
+                        metadata.bucket_incarnation_id = Uuid::new_v4();
+                    }
+                    metadata
+                        .save_with_store_committed(store.clone())
+                        .await
+                        .expect("inject inconsistent metadata authority");
+                }
+                "retired" => {
+                    // A returning disk may retain a directory after retirement was committed.
+                    crate::bucket::retirement::commit_retirement(store.clone(), &bucket, incarnation, &ObjectOptions::default())
+                        .await
+                        .expect("persist retirement evidence");
+                }
+                _ => unreachable!("fixed fixture states"),
+            }
+            for disk in 1..4 {
+                tokio::fs::remove_dir(root.path().join(format!("pool0-disk{disk}/{bucket}")))
+                    .await
+                    .expect("lose fixture directories");
+            }
+            assert!(
+                store
+                    .heal_bucket_at_incarnation(&bucket, incarnation, &HealOpts::default())
+                    .await
+                    .is_err(),
+                "must reject {state}"
+            );
+            for disk in 1..4 {
+                assert!(
+                    !root.path().join(format!("pool0-disk{disk}/{bucket}")).exists(),
+                    "{state} cannot authorize reconstruction"
+                );
+            }
+        }
+        shutdown.cancel();
+    }
+
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn admin_bucket_recovery_rejects_offline_disks_and_fully_missing_sets() {
+        for offline in [false, true] {
+            let (root, store, shutdown) = multi_pool_heal_store().await;
+            let bucket = "recovery-unproven-set";
+            store
+                .make_bucket(bucket, &MakeBucketOptions::default())
+                .await
+                .expect("fixture bucket");
+            let incarnation = store.bucket_incarnation_id_from_disk(bucket).await.expect("fixture identity");
+            let first_missing = usize::from(offline);
+            for disk in first_missing..4 {
+                tokio::fs::remove_dir(root.path().join(format!("pool0-disk{disk}/{bucket}")))
+                    .await
+                    .expect("remove fixture directories");
+            }
+            if offline {
+                let disk = store.pools[0].disk_set[0].disks.read().await[0]
+                    .clone()
+                    .expect("online fixture disk");
+                disk.close().await.expect("stop disk monitoring");
+                disk.force_offline_for_test();
+            }
+            assert!(
+                store
+                    .heal_bucket_at_incarnation(bucket, incarnation, &HealOpts::default())
+                    .await
+                    .is_err()
+            );
+            for disk in first_missing..4 {
+                assert!(!root.path().join(format!("pool0-disk{disk}/{bucket}")).exists());
+            }
+            shutdown.cancel();
+        }
+    }
+
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn admin_bucket_recovery_dry_run_preserves_missing_directories() {
+        let (root, store, shutdown) = multi_pool_heal_store().await;
+        let bucket = "recovery-dry-run";
+        store
+            .make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("fixture bucket");
+        let incarnation = store.bucket_incarnation_id_from_disk(bucket).await.expect("fixture identity");
+        for disk in 1..4 {
+            tokio::fs::remove_dir(root.path().join(format!("pool0-disk{disk}/{bucket}")))
+                .await
+                .expect("remove fixture directories");
+        }
+        store
+            .heal_bucket_at_incarnation(
+                bucket,
+                incarnation,
+                &HealOpts {
+                    dry_run: true,
+                    ..Default::default()
+                },
+            )
+            .await
+            .expect("dry run can inspect eligible recovery");
+        for disk in 1..4 {
+            assert!(
+                !root.path().join(format!("pool0-disk{disk}/{bucket}")).exists(),
+                "dry run must not reconstruct directories"
+            );
+        }
+        shutdown.cancel();
+    }
+
     #[tokio::test]
     #[serial_test::serial]
     async fn bucket_incarnation_heal_preserves_successor_shards_and_allows_fresh_repair() {

crates/ecstore/src/store/init.rs

@@ -108,6 +108,10 @@ fn should_retry_format_load(err: &Error) -> bool {
     )
 }
 
+fn format_load_retry_exhausted(err: &Error, retries: u32) -> bool {
+    retries >= 10 && !matches!(err, Error::ErasureReadQuorum | Error::FirstDiskWait | Error::NotFirstDisk)
+}
+
 fn should_auto_start_rebalance_after_init(decommission_running: bool, rebalance_resume_required: bool) -> bool {
     rebalance_resume_required && !decommission_running
 }
@@ -484,7 +488,7 @@ impl ECStore {
             check_disk_fatal_errs(&errs)?;
 
             let loaded_format = {
-                let mut times = 0;
+                let mut times = 0_u32;
                 let mut interval = 1;
                 loop {
                     match init_format::connect_load_init_formats_with_instance_ctx(
@@ -500,13 +504,13 @@ impl ECStore {
                         Ok(fm) => break Ok(fm),
                         Err(e) if !should_retry_format_load(&e) => break Err(e),
                         // Wrap the final error if we are giving up
-                        Err(e) if times >= 10 => {
+                        Err(e) if format_load_retry_exhausted(&e, times) => {
                             break Err(Error::other(format!("store init failed to load formats after {times} retries: {e}")));
                         }
                         // Retrying so just drop the error
                         Err(_) => {}
                     }
-                    times += 1;
+                    times = times.saturating_add(1);
                     if interval < 16 {
                         interval *= 2;
                     }
@@ -519,6 +523,12 @@ impl ECStore {
                         "Retrying storage format load"
                     );
                     select! {
+                        _ = ctx.cancelled() => {
+                            return Err(Error::other(std::io::Error::new(
+                                std::io::ErrorKind::Interrupted,
+                                "storage format initialization cancelled",
+                            )));
+                        }
                         _ = tokio::signal::ctrl_c() => {
                             info!(
                                 event = EVENT_STORE_FORMAT_RETRY,
@@ -2201,6 +2211,23 @@ mod tests {
         assert_eq!(attempts.load(Ordering::SeqCst), 2);
     }
 
+    #[test]
+    fn format_peer_wait_does_not_exhaust_the_retry_budget() {
+        for retries in [0, 9, 10, u32::MAX] {
+            for error in [
+                StorageError::ErasureReadQuorum,
+                StorageError::FirstDiskWait,
+                StorageError::NotFirstDisk,
+            ] {
+                assert!(
+                    !super::format_load_retry_exhausted(&error, retries),
+                    "peer availability must remain retryable"
+                );
+            }
+            assert_eq!(super::format_load_retry_exhausted(&StorageError::Timeout, retries), retries >= 10);
+        }
+    }
+
     #[test]
     fn test_should_retry_format_load_rejects_permanent_corruption() {
         assert!(!should_retry_format_load(&StorageError::CorruptedFormat));
@@ -9567,8 +9594,18 @@ mod tests {
     #[cfg(feature = "test-util")]
     async fn decommission_migrates_and_verifies_registered_durable_ilm_records_scenario() {
         let temp_dir = tempfile::tempdir().expect("create temp store dir");
-        let (ctx, store, _shutdown) =
-            without_storage_class_env(build_isolated_test_store(temp_dir.path(), "durable-ilm-decommission", &[4, 4])).await;
+        // The explicit recovery below must be the only journal consumer,
+        // including when receipt validation outlasts the background interval.
+        let mut instance_ctx = crate::runtime::instance::InstanceContext::new();
+        instance_ctx.suppress_tier_delete_journal_recovery_for_test();
+        let (ctx, store, _shutdown) = without_storage_class_env(build_isolated_test_store_with_layout(
+            temp_dir.path(),
+            "durable-ilm-decommission",
+            &[(1, 4), (1, 4)],
+            CancellationToken::new(),
+            Some(Arc::new(instance_ctx)),
+        ))
+        .await;
 
         let tier_name = "DECOMMISSION-ILM";
         let backend = register_transition_reconcile_test_tier(&ctx.tier_config_mgr(), tier_name).await;
@@ -10040,6 +10077,16 @@ mod tests {
         .await
         .expect("manual task receipt should restore after the corruption check");
 
+        assert!(matches!(
+            com::read_config(store.pools[0].clone(), &tier_path).await,
+            Err(Error::ConfigNotFound)
+        ));
+        assert_eq!(
+            com::read_config(store.pools[1].clone(), &tier_path)
+                .await
+                .expect("manual recovery must still observe the exact migrated tier journal"),
+            tier_bytes
+        );
         let tier_stats = recover_tier_delete_journal_entries(store.clone(), 100, None)
             .await
             .expect("tier journal recovery should consume the migrated record before completion");
@@ -10796,8 +10843,19 @@ mod tests {
         const JOURNAL_COUNT: usize = 65;
 
         let temp_dir = tempfile::tempdir().expect("create bounded rollback store dir");
-        let (ctx, store, _shutdown) =
-            without_storage_class_env(build_isolated_test_store(temp_dir.path(), "bounded-dispatch-rollback", &[4])).await;
+        // The test's worker must be the only one to roll back the seeded manifest: if
+        // the setup outlasts TIER_DELETE_JOURNAL_RECOVERY_INTERVAL, a background pass
+        // rolls it back before the slow-delete hook is installed.
+        let mut instance_ctx = crate::runtime::instance::InstanceContext::new();
+        instance_ctx.suppress_tier_delete_journal_recovery_for_test();
+        let (ctx, store, _shutdown) = without_storage_class_env(build_isolated_test_store_with_layout(
+            temp_dir.path(),
+            "bounded-dispatch-rollback",
+            &[(1, 4)],
+            CancellationToken::new(),
+            Some(Arc::new(instance_ctx)),
+        ))
+        .await;
         crate::bucket::metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
         let bucket = "bounded-dispatch-rollback-bucket";
         store
@@ -10953,8 +11011,19 @@ mod tests {
     #[serial_test::serial(storage_class_env)]
     async fn dispatch_manifest_page_timeout_detaches_one_deduplicated_worker_until_convergence() {
         let temp_dir = tempfile::tempdir().expect("create detached rollback store dir");
-        let (ctx, store, _shutdown) =
-            without_storage_class_env(build_isolated_test_store(temp_dir.path(), "detached-dispatch-rollback", &[4])).await;
+        // The test's own passes must be the only ones to process the seeded manifest:
+        // if the setup outlasts TIER_DELETE_JOURNAL_RECOVERY_INTERVAL, a background
+        // pass rolls it back before the slow-delete hook is installed.
+        let mut instance_ctx = crate::runtime::instance::InstanceContext::new();
+        instance_ctx.suppress_tier_delete_journal_recovery_for_test();
+        let (ctx, store, _shutdown) = without_storage_class_env(build_isolated_test_store_with_layout(
+            temp_dir.path(),
+            "detached-dispatch-rollback",
+            &[(1, 4)],
+            CancellationToken::new(),
+            Some(Arc::new(instance_ctx)),
+        ))
+        .await;
         crate::bucket::metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
         let bucket = "detached-dispatch-rollback-bucket";
         store
@@ -11022,9 +11091,20 @@ mod tests {
         const WORKER_LIMIT: usize = 4;
 
         let temp_dir = tempfile::tempdir().expect("create bounded detached rollback store dir");
-        let (ctx, store, shutdown) =
-            without_storage_class_env(build_isolated_test_store(temp_dir.path(), "bounded-detached-dispatch-rollback", &[4]))
-                .await;
+        // The test's passes must be the only ones to process the seeded manifests: if
+        // the setup outlasts TIER_DELETE_JOURNAL_RECOVERY_INTERVAL, a background pass
+        // rolls them back before the pause hook is installed. The restarted store
+        // below keeps background recovery, which its half of the test relies on.
+        let mut instance_ctx = crate::runtime::instance::InstanceContext::new();
+        instance_ctx.suppress_tier_delete_journal_recovery_for_test();
+        let (ctx, store, shutdown) = without_storage_class_env(build_isolated_test_store_with_layout(
+            temp_dir.path(),
+            "bounded-detached-dispatch-rollback",
+            &[(1, 4)],
+            CancellationToken::new(),
+            Some(Arc::new(instance_ctx)),
+        ))
+        .await;
         crate::bucket::metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
         let tier_name = "BOUNDED-DETACHED-DISPATCH-ROLLBACK";
         let _backend = register_mock_tier(&ctx.tier_config_mgr(), tier_name).await;
@@ -17346,6 +17426,19 @@ mod tests {
     #[tokio::test]
     #[serial_test::serial(storage_class_env)]
     async fn delete_holds_bucket_incarnation_sentinel_through_object_commit() {
+        // The recreation waits for the delete's sentinel on its lifecycle write lock
+        // from before `barrier.release()`, so that lock wait also has to cover the
+        // delete's commit. Raise the lock acquire timeout from the 5 s default to give
+        // the commit room on a starved runner; a kept sentinel still fails the test,
+        // after 30 s instead of 5 s.
+        temp_env::async_with_vars(
+            [(rustfs_config::ENV_OBJECT_LOCK_ACQUIRE_TIMEOUT, Some("30"))],
+            delete_holds_bucket_incarnation_sentinel_through_object_commit_body(),
+        )
+        .await;
+    }
+
+    async fn delete_holds_bucket_incarnation_sentinel_through_object_commit_body() {
         let temp = tempfile::tempdir().expect("create temp store dir");
         let (_ctx, store, _shutdown) =
             without_storage_class_env(build_isolated_test_store(temp.path(), "delete-bucket-incarnation", &[4])).await;
@@ -17418,13 +17511,19 @@ mod tests {
             "bucket delete and same-name recreation must wait for the in-flight object delete"
         );
         barrier.release();
-        delete
+        // The recreation's lock wait started before the release; give the delete as long.
+        tokio::time::timeout(Duration::from_secs(30), delete)
             .await
+            .expect("delete should finish once the barrier is released")
             .expect("delete task should join")
             .expect("old-generation delete should complete");
-        tokio::time::timeout(Duration::from_secs(10), recreate)
+        // If the delete kept the sentinel, `delete_bucket` would fail on its lifecycle
+        // write lock after the lock acquire timeout (30 s here). The rest of the
+        // recreation is fsync-bound IO on four disks, so this bound, above the lock
+        // timeout, only stops a hang.
+        tokio::time::timeout(Duration::from_secs(120), recreate)
             .await
-            .expect("bucket recreation should finish after delete releases the sentinel")
+            .expect("bucket recreation should not hang after delete releases the sentinel")
             .expect("bucket recreation task should join")
             .expect("bucket recreation should succeed");
 
@@ -17477,6 +17576,19 @@ mod tests {
     #[tokio::test]
     #[serial_test::serial(storage_class_env)]
     async fn delete_objects_holds_bucket_incarnation_sentinel_through_commit() {
+        // The recreation waits for the delete's sentinel on its lifecycle write lock
+        // from before `barrier.release()`, so that lock wait also has to cover the
+        // delete's commit. Raise the lock acquire timeout from the 5 s default to give
+        // the commit room on a starved runner; a kept sentinel still fails the test,
+        // after 30 s instead of 5 s.
+        temp_env::async_with_vars(
+            [(rustfs_config::ENV_OBJECT_LOCK_ACQUIRE_TIMEOUT, Some("30"))],
+            delete_objects_holds_bucket_incarnation_sentinel_through_commit_body(),
+        )
+        .await;
+    }
+
+    async fn delete_objects_holds_bucket_incarnation_sentinel_through_commit_body() {
         let temp = tempfile::tempdir().expect("create temp store dir");
         let (_ctx, store, _shutdown) =
             without_storage_class_env(build_isolated_test_store(temp.path(), "batch-delete-bucket-incarnation", &[4])).await;
@@ -17538,14 +17650,22 @@ mod tests {
             "bucket recreation must wait for the in-flight batch delete"
         );
         barrier.release();
-        let (_deleted, errors) = delete.await.expect("batch delete task should join");
+        // The recreation's lock wait started before the release; give the delete as long.
+        let (_deleted, errors) = tokio::time::timeout(Duration::from_secs(30), delete)
+            .await
+            .expect("batch delete should finish once the barrier is released")
+            .expect("batch delete task should join");
         assert!(
             errors.iter().all(Option::is_none),
             "old-generation batch delete should complete: {errors:?}"
         );
-        tokio::time::timeout(Duration::from_secs(10), recreate)
+        // If the delete kept the sentinel, `delete_bucket` would fail on its lifecycle
+        // write lock after the lock acquire timeout (30 s here). The rest of the
+        // recreation is fsync-bound IO on four disks, so this bound, above the lock
+        // timeout, only stops a hang.
+        tokio::time::timeout(Duration::from_secs(120), recreate)
             .await
-            .expect("bucket recreation should finish after batch delete releases the sentinel")
+            .expect("bucket recreation should not hang after batch delete releases the sentinel")
             .expect("bucket recreation task should join")
             .expect("bucket recreation should succeed");
         store
@@ -22023,7 +22143,6 @@ mod tests {
             .make_bucket(bucket, &MakeBucketOptions::default())
             .await
             .expect("kill-point bucket should be created");
-        let _compaction_proof = install_transition_transaction_compaction_fleet_proof_for_test("object-transaction-fencing-test");
 
         for (index, point, expected_state, expected_revision, expected_committed, expected_recovered) in [
             (
@@ -22067,10 +22186,14 @@ mod tests {
                 .await
                 .expect("kill-point source should be written");
             let remote_before = backend.object_count().await;
+            // Earlier cancellation/recovery cases must not consume this
+            // transaction's 30-second homogeneous-fleet proof window.
+            let _compaction_proof =
+                install_transition_transaction_compaction_fleet_proof_for_test("object-transaction-fencing-test");
             let barrier = TransitionTransactionKillPointBarrier::install(bucket, &object, point);
             let transition_store = store.clone();
             let transition_object = object.clone();
-            let transition = tokio::spawn(async move {
+            let mut transition = tokio::spawn(async move {
                 transition_store
                     .transition_object(
                         bucket,
@@ -22089,7 +22212,12 @@ mod tests {
                     )
                     .await
             });
-            barrier.wait_until_paused().await;
+            tokio::select! {
+                _ = barrier.wait_until_paused() => {}
+                result = &mut transition => {
+                    panic!("transition exited before reaching {point:?}: {result:?}");
+                }
+            }
             transition.abort();
             assert!(
                 transition

crates/ecstore/src/store/init_format.rs

@@ -134,6 +134,18 @@ pub(crate) async fn connect_load_init_formats_with_instance_ctx(
     {
         return Err(Error::CorruptedFormat);
     }
+    if matches!(format_quorum, Some(Err(Error::ErasureReadQuorum)))
+        && let Some((format, members)) = select_existing_format_read_quorum(&formats, &errs, set_drive_count)
+    {
+        retain_format_quorum_members(instance_ctx, disks, &format, &members, set_drive_count).await?;
+        // Loading an existing format grants no authority to initialize pool
+        // metadata or migrate formats. The persisted storage classes and each
+        // object's EC metadata still determine service and operation quorums.
+        return Ok(LoadedFormat {
+            format,
+            pool_meta_bootstrap_authority: PoolMetaBootstrapAuthority::None,
+        });
+    }
     let resumable_partial_migration = formats_present > 0
         && errs.iter().any(|error| matches!(error, Some(DiskError::UnformattedDisk)))
         && format_quorum.as_ref().is_some_and(Result::is_err)
@@ -578,6 +590,43 @@ pub fn get_format_erasure_in_quorum(formats: &[Option<FormatV3>], slot_offset: u
     select_format_erasure_in_quorum(formats, slot_offset).map(|(format, _)| format)
 }
 
+/// Load a consistent existing layout with enough members per set to read the
+/// max-parity system metadata containing the persisted storage classes. This
+/// is not object-read admission: STANDARD, RRS and historical objects can have
+/// different EC layouts, which are validated after this bootstrap read.
+/// Format migration and healing still use strict-majority election.
+fn select_existing_format_read_quorum(
+    formats: &[Option<FormatV3>],
+    errors: &[Option<DiskError>],
+    set_drive_count: usize,
+) -> Option<(FormatV3, Vec<bool>)> {
+    if set_drive_count == 0 || formats.is_empty() || formats.len() != errors.len() {
+        return None;
+    }
+    let metadata_parity = set_drive_count / 2;
+    let metadata_read_quorum = set_drive_count - metadata_parity;
+    let reference = formats.iter().flatten().next()?;
+    if !formats_match_reference_slots(formats, reference, 0)
+        || check_format_erasure_value_for_topology(reference, formats.len(), set_drive_count).is_err()
+        || formats.iter().zip(errors).any(|(format, error)| match (format, error) {
+            (Some(_), None) => false,
+            (None, Some(DiskError::DiskNotFound)) => false,
+            (None, Some(error)) => !is_network_like_disk_error(error),
+            _ => true,
+        })
+        || formats
+            .chunks(set_drive_count)
+            .any(|set| set.iter().flatten().count() < metadata_read_quorum)
+    {
+        return None;
+    }
+
+    let mut format = reference.clone();
+    format.erasure.this = Uuid::nil();
+    format.disk_info = None;
+    Some((format, formats.iter().map(Option::is_some).collect()))
+}
+
 pub(crate) fn select_format_erasure_in_quorum(formats: &[Option<FormatV3>], slot_offset: usize) -> Result<(FormatV3, Vec<bool>)> {
     let mut candidates = HashMap::new();
     let formats_present = formats.iter().flatten().count();
@@ -1120,6 +1169,190 @@ mod tests {
         assert!(!formats_match_reference_slots(&formats, &reference, 0));
     }
 
+    fn half_pool_formats() -> (Vec<Option<FormatV3>>, Vec<Option<DiskError>>) {
+        let reference = FormatV3::new(6, 4);
+        let formats = (0..24)
+            .map(|slot| {
+                (slot % 4 < 2).then(|| {
+                    let mut format = reference.clone();
+                    format.erasure.this = reference.erasure.sets[slot / 4][slot % 4];
+                    format
+                })
+            })
+            .collect::<Vec<_>>();
+        let errors = formats
+            .iter()
+            .map(|format| format.is_none().then_some(DiskError::DiskNotFound))
+            .collect();
+        (formats, errors)
+    }
+
+    #[test]
+    fn existing_format_read_quorum_follows_each_sets_metadata_layout() {
+        // These are system-metadata EC boundaries, independent of user-object
+        // storage classes and of how many drives are hosted by each node.
+        for (set_width, read_quorum) in [(1, 1), (2, 1), (3, 2), (4, 2), (6, 3), (8, 4), (12, 6), (16, 8)] {
+            for set_count in [1, 2, 5] {
+                let reference = FormatV3::new(set_count, set_width);
+                let mut formats = (0..set_count * set_width)
+                    .map(|slot| {
+                        let set = slot / set_width;
+                        let drive = slot % set_width;
+                        // Rotate surviving slots between sets so admission
+                        // cannot depend on the first drive or node being up.
+                        ((drive + set) % set_width < read_quorum).then(|| {
+                            let mut format = reference.clone();
+                            format.erasure.this = reference.erasure.sets[set][drive];
+                            format
+                        })
+                    })
+                    .collect::<Vec<_>>();
+                let mut errors = formats
+                    .iter()
+                    .map(|format| format.is_none().then_some(DiskError::DiskNotFound))
+                    .collect::<Vec<_>>();
+                let (loaded, members) = select_existing_format_read_quorum(&formats, &errors, set_width)
+                    .unwrap_or_else(|| panic!("metadata read quorum must load {set_count} sets of {set_width} drives"));
+                assert_eq!(loaded.shared_identity(), reference.shared_identity());
+                assert!(loaded.erasure.this.is_nil());
+                assert_eq!(members.iter().filter(|member| **member).count(), set_count * read_quorum);
+
+                // More members in other sets cannot compensate for one set
+                // falling below its own metadata reconstruction boundary.
+                for slot in 0..(set_count - 1) * set_width {
+                    let mut format = reference.clone();
+                    format.erasure.this = reference.erasure.sets[slot / set_width][slot % set_width];
+                    formats[slot] = Some(format);
+                    errors[slot] = None;
+                }
+                let lost_slot = formats
+                    .iter()
+                    .rposition(Option::is_some)
+                    .expect("at least one surviving drive");
+                formats[lost_slot] = None;
+                errors[lost_slot] = Some(DiskError::DiskNotFound);
+                assert!(
+                    select_existing_format_read_quorum(&formats, &errors, set_width).is_none(),
+                    "one set below read quorum must block {set_count} sets of {set_width} drives"
+                );
+            }
+        }
+    }
+
+    #[test]
+    fn existing_format_read_quorum_preserves_election_and_set_boundaries() {
+        let (mut formats, mut errors) = half_pool_formats();
+        assert!(select_existing_format_read_quorum(&[], &[], 0).is_none());
+        assert!(select_existing_format_read_quorum(&formats, &errors[..23], 4).is_none());
+        assert!(select_existing_format_read_quorum(&formats, &errors, 0).is_none());
+        assert!(select_existing_format_read_quorum(&formats, &errors, 6).is_none());
+        assert!(select_existing_format_read_quorum(&formats, &errors, 4).is_some());
+        assert!(matches!(get_format_erasure_in_quorum(&formats, 0), Err(Error::ErasureReadQuorum)));
+
+        let mut extra = formats[0].as_ref().expect("seed format").clone();
+        extra.erasure.this = extra.erasure.sets[0][2];
+        formats[2] = Some(extra);
+        errors[2] = None;
+        assert!(get_format_erasure_in_quorum(&formats, 0).is_ok(), "13 votes must establish a majority");
+        formats[5] = None;
+        errors[5] = Some(DiskError::DiskNotFound);
+        assert!(
+            select_existing_format_read_quorum(&formats, &errors, 4).is_none(),
+            "12 votes cannot compensate for a set with only one readable drive"
+        );
+        formats[2] = None;
+        errors[2] = Some(DiskError::DiskNotFound);
+        assert!(select_existing_format_read_quorum(&formats, &errors, 4).is_none(), "11 votes must wait");
+    }
+
+    #[test]
+    fn existing_format_read_quorum_rejects_conflicts_invalid_slots_and_blank_disks() {
+        for failure in [
+            "conflict",
+            "wrong_slot",
+            "nil",
+            "duplicate",
+            "blank",
+            "corrupt",
+            "access_denied",
+            "tie",
+        ] {
+            let (mut formats, mut errors) = half_pool_formats();
+            match failure {
+                "conflict" => formats[1].as_mut().expect("second format").id = Uuid::new_v4(),
+                "wrong_slot" => {
+                    let first_id = formats[0].as_ref().expect("first format").erasure.this;
+                    formats[1].as_mut().expect("second format").erasure.this = first_id;
+                }
+                "nil" => {
+                    for format in formats.iter_mut().flatten() {
+                        format.id = Uuid::nil();
+                    }
+                }
+                "duplicate" => {
+                    for format in formats.iter_mut().flatten() {
+                        format.erasure.sets[5][3] = format.erasure.sets[0][0];
+                    }
+                }
+                "blank" => errors[2] = Some(DiskError::UnformattedDisk),
+                "corrupt" => errors[2] = Some(DiskError::CorruptedFormat),
+                "access_denied" => errors[2] = Some(DiskError::FileAccessDenied),
+                "tie" => {
+                    let other = FormatV3::new(6, 4);
+                    for slot in 0..24 {
+                        if formats[slot].is_none() {
+                            let mut format = other.clone();
+                            format.erasure.this = other.erasure.sets[slot / 4][slot % 4];
+                            formats[slot] = Some(format);
+                            errors[slot] = None;
+                        }
+                    }
+                }
+                _ => unreachable!(),
+            }
+            assert!(
+                select_existing_format_read_quorum(&formats, &errors, 4).is_none(),
+                "{failure} must not authorize degraded startup"
+            );
+        }
+    }
+
+    #[tokio::test]
+    async fn existing_format_load_accepts_each_half_of_four_by_six_without_format_writes() {
+        let (_root, all_disks) = local_disks_with_set_width(24, 4).await;
+        let reference = FormatV3::new(6, 4);
+        for (slot, disk) in all_disks.iter().enumerate() {
+            let mut format = reference.clone();
+            format.erasure.this = reference.erasure.sets[slot / 4][slot % 4];
+            save_format_file(disk, &Some(format)).await.expect("persist existing format");
+        }
+        let before = format_bytes(&all_disks).await;
+        for first in 0..4 {
+            for second in first + 1..4 {
+                let mut disks = all_disks.clone();
+                for (slot, disk) in disks.iter_mut().enumerate() {
+                    if slot % 4 != first && slot % 4 != second {
+                        *disk = None;
+                    }
+                }
+                let loaded = connect_load_init_formats_with_instance_ctx(
+                    &Arc::new(InstanceContext::new()),
+                    first == 0,
+                    &mut disks,
+                    6,
+                    4,
+                    None,
+                )
+                .await
+                .expect("two nodes must load the existing format");
+                assert_eq!(loaded.format.shared_identity(), reference.shared_identity());
+                assert!(matches!(loaded.pool_meta_bootstrap_authority, PoolMetaBootstrapAuthority::None));
+                assert_eq!(disks.iter().flatten().count(), 12);
+                assert_eq!(format_bytes(&all_disks).await, before, "read startup must never rewrite formats");
+            }
+        }
+    }
+
     #[tokio::test]
     async fn existing_format_load_succeeds_with_a_strict_majority() {
         let (_temp_dir, mut disks) = two_local_disks_with_missing_third().await;

crates/ecstore/src/store/list_objects.rs

@@ -9221,7 +9221,7 @@ mod test {
         };
         let snapshot = PersistentListMetadataObject::from_object_info(&object);
         assert!(snapshot.to_object_info("bucket").etag.is_none());
-        let mut old_snapshot = snapshot.clone();
+        let mut old_snapshot = snapshot;
         old_snapshot.etag = object.etag.clone();
         let old_contents = format!(
             "# rustfs-listobjects-key-only-v2\n{}\n",

crates/ecstore/src/store/object.rs

@@ -4849,7 +4849,9 @@ impl ECStore {
         opts: ObjectOptions,
         guards: impl IntoIterator<Item = rustfs_lock::NamespaceLockGuard>,
     ) -> Result<ObjectInfo> {
-        let guards: Vec<_> = guards.into_iter().collect();
+        let guards = Arc::new(guards.into_iter().collect::<Vec<_>>());
+        let mut opts = opts;
+        opts.delete_mutation_owner = Some(Arc::new((opts.delete_mutation_owner.take(), guards.clone())));
         let store = Arc::clone(self);
         let bucket = bucket.to_owned();
         let object = object.to_owned();
@@ -4861,7 +4863,7 @@ impl ECStore {
                 if result.is_ok() {
                     list_objects::observe_list_objects_mutation(&store, &bucket).await;
                 }
-                // Keep: the only use of `guards` here, it moves them into the task so the locks outlive a dropped caller.
+                // Transient options also retain these guards through late disk I/O.
                 drop(guards);
                 result
             })
@@ -4930,7 +4932,7 @@ impl ECStore {
             let mut opts = opts;
             let receipt_sink = install_tier_free_version_receipt_sink(&mut opts);
             let result = self
-                .handle_delete_object_with_journal_inner(bucket, object, opts, tier_journal_api)
+                .handle_delete_object_with_journal_inner_boxed(bucket, object, opts, tier_journal_api)
                 .await;
             enqueue_recorded_tier_free_versions(self, receipt_sink).await;
             result
@@ -4938,6 +4940,16 @@ impl ECStore {
         .await
     }
 
+    fn handle_delete_object_with_journal_inner_boxed<'a>(
+        &'a self,
+        bucket: &'a str,
+        object: &'a str,
+        opts: ObjectOptions,
+        tier_journal_api: Option<Arc<ECStore>>,
+    ) -> impl std::future::Future<Output = Result<ObjectInfo>> + Send + 'a {
+        Box::pin(self.handle_delete_object_with_journal_inner(bucket, object, opts, tier_journal_api))
+    }
+
     async fn handle_delete_object_with_journal_inner(
         &self,
         bucket: &str,
@@ -4955,9 +4967,9 @@ impl ECStore {
             if is_meta_bucketname(bucket) || (opts.delete_prefix && opts.bucket_lifecycle_lock_fence.is_some()) {
                 None
             } else if opts.delete_prefix {
-                Some(self.acquire_bucket_lifecycle_write_lock(bucket).await?)
+                Some(Arc::new(self.acquire_bucket_lifecycle_write_lock(bucket).await?))
             } else {
-                Some(self.acquire_bucket_lifecycle_read_lock(bucket).await?)
+                Some(Arc::new(self.acquire_bucket_lifecycle_read_lock(bucket).await?))
             };
         let object = if opts.delete_prefix && !opts.delete_prefix_object {
             object.to_owned()
@@ -4975,6 +4987,7 @@ impl ECStore {
         opts.tier_delete_journal_api = tier_journal_api.clone();
         if let Some(guard) = _bucket_lifecycle_guard.as_ref() {
             opts.add_bucket_lifecycle_lock_guard(guard);
+            opts.delete_mutation_owner = Some(Arc::new((opts.delete_mutation_owner.take(), guard.clone())));
         }
 
         if !is_meta_bucketname(bucket) {
@@ -5227,6 +5240,40 @@ impl ECStore {
                 return Ok(obj);
             }
             Err(err) => {
+                if (is_err_object_not_found(&err) || is_err_version_not_found(&err))
+                    && (opts.version_id.is_some() || (!opts.versioned && !opts.version_suspended))
+                    && opts.http_preconditions.is_none()
+                    && !opts.replication_request
+                {
+                    let mut reclaimed = None;
+                    for pool in &self.pools {
+                        if self.is_suspended(pool.pool_idx).await || self.is_pool_rebalancing(pool.pool_idx).await {
+                            return Err(StorageError::SlowDown);
+                        }
+                        let pool_idx = pool.pool_idx;
+                        let pool = pool.clone();
+                        match self
+                            .run_external_decommission_capacity_object_delete(
+                                pool_idx,
+                                bucket,
+                                object,
+                                object,
+                                opts.clone(),
+                                |opts| async move { pool.delete_object(bucket, object, opts).await },
+                            )
+                            .await
+                        {
+                            Ok(result) => reclaimed = Some(result),
+                            Err(missing) if is_err_object_not_found(&missing) || is_err_version_not_found(&missing) => {}
+                            Err(error) => return Err(error),
+                        }
+                    }
+                    if let Some(mut result) = reclaimed {
+                        result.bucket = bucket.to_owned();
+                        result.name = decode_dir_object(object);
+                        return Ok(result);
+                    }
+                }
                 // A folder key (`prefix/`) with no object metadata may still exist on
                 // disk as an orphan empty-directory tree (issue #4189): listings show
                 // it as a common prefix, but no regular delete path can remove it.
@@ -5422,12 +5469,13 @@ impl ECStore {
             None
         } else {
             match self.acquire_bucket_lifecycle_read_lock(bucket).await {
-                Ok(guard) => Some(guard),
+                Ok(guard) => Some(Arc::new(guard)),
                 Err(err) => return return_batch_delete_lock_error_with_accounting(objects.as_slice(), err),
             }
         };
         if let Some(guard) = _bucket_lifecycle_guard.as_ref() {
             opts.add_bucket_lifecycle_lock_guard(guard);
+            opts.delete_mutation_owner = Some(Arc::new((opts.delete_mutation_owner.take(), guard.clone())));
         }
         if opts.delete_replication_config_snapshot.is_none() {
             match ReplicationObjectBridge::delete_request_config_in(&self.ctx, bucket).await {
@@ -5493,10 +5541,11 @@ impl ECStore {
         if current_bucket_incarnation_id.is_some() {
             pause_delete_after_object_lock_snapshot(bucket).await;
         }
-        let _object_lock_guards = match self.acquire_delete_objects_write_locks(bucket, &objects, &mut opts).await {
+        let _object_lock_guards = Arc::new(match self.acquire_delete_objects_write_locks(bucket, &objects, &mut opts).await {
             Ok(guards) => guards,
             Err(err) => return return_batch_delete_lock_error_with_accounting(objects.as_slice(), err),
-        };
+        });
+        opts.delete_mutation_owner = Some(Arc::new((opts.delete_mutation_owner.take(), _object_lock_guards.clone())));
         #[cfg(any(test, feature = "test-util"))]
         if !_object_lock_guards.is_empty() {
             notify_delete_namespace_acquired(bucket);
@@ -9241,8 +9290,31 @@ mod tests {
                 .expect("read original marker");
             assert_eq!(info.delete_marker_incarnation(), Some(old_incarnation));
         }
-        // Taking four actual members offline preserves their existing xl.meta.
-        // All deletion/recreation work below goes through the normal store API.
+        // Save each stale member's inline versions before converging the old
+        // bucket's delete receipts, then reintroduce them after recreation.
+        let mut stale_metadata = Vec::new();
+        for disk in before.iter().skip(12).flatten() {
+            for version in &historical {
+                let info = disk
+                    .read_version(
+                        "",
+                        bucket,
+                        "history.bin",
+                        &version.to_string(),
+                        &crate::disk::ReadOptions {
+                            read_data: true,
+                            ..Default::default()
+                        },
+                    )
+                    .await
+                    .expect("read stale inline data version");
+                assert!(info.inline_data() && info.data.is_some(), "fixture payload must be preserved in xl.meta");
+            }
+            stale_metadata.push((
+                disk.read_all(bucket, "history.bin/xl.meta").await.unwrap(),
+                disk.read_all(bucket, &format!("{object}/xl.meta")).await.unwrap(),
+            ));
+        }
         for slot in 12..16 {
             set.disks.write().await[slot] = None;
         }
@@ -9272,6 +9344,29 @@ mod tests {
             )
             .await
             .expect("delete exact old marker with four drives offline");
+        let pending = store
+            .handle_delete_bucket(bucket, &crate::storage_api_contracts::bucket::DeleteBucketOptions::default())
+            .await
+            .expect_err("offline participants require retained delete receipts");
+        assert!(matches!(
+            pending,
+            StorageError::BucketNotEmpty(_) | StorageError::BucketNotEmptyWithDetails { .. }
+        ));
+        assert!(
+            !crate::bucket::retirement::is_retired(store.clone(), bucket, old_incarnation)
+                .await
+                .unwrap()
+        );
+        *set.disks.write().await = before.clone();
+        store
+            .recover_delete_transactions_at_incarnation(
+                bucket,
+                None,
+                Some(old_incarnation),
+                &rustfs_heal_contracts::heal_channel::HealOpts::default(),
+            )
+            .await
+            .expect("converge old bucket deletes on all participants");
         store
             .handle_delete_bucket(bucket, &crate::storage_api_contracts::bucket::DeleteBucketOptions::default())
             .await
@@ -9281,6 +9376,9 @@ mod tests {
                 .await
                 .expect("durable retirement")
         );
+        for slot in 12..16 {
+            set.disks.write().await[slot] = None;
+        }
         store
             .handle_make_bucket(bucket, &create)
             .await
@@ -9306,6 +9404,13 @@ mod tests {
                 .expect("write new generation");
             new_versions.push((key, info.version_id.unwrap(), payload));
         }
+        for (slot, (history, marker)) in (12..16).zip(stale_metadata) {
+            for (key, bytes) in [("history.bin", history), (object, marker)] {
+                let root = dirs[slot].path().join(bucket).join(key);
+                tokio::fs::create_dir_all(&root).await.unwrap();
+                tokio::fs::write(root.join("xl.meta"), bytes).await.unwrap();
+            }
+        }
         *set.disks.write().await = before.clone();
         let heal = rustfs_heal_contracts::heal_channel::HealOpts {
             remove: true,
@@ -11477,3 +11582,160 @@ mod tests {
         .await;
     }
 }
+
+#[cfg(test)]
+mod missing_key_delete_quorum_tests {
+    use super::*;
+    use crate::bucket::metadata_sys::{init_bucket_metadata_sys, test_support::isolated_store_over_temp_disks};
+    use crate::storage_api_contracts::bucket::{BucketOperations, BucketOptions, MakeBucketOptions};
+
+    async fn degraded_store(bucket: &str) -> (Vec<tempfile::TempDir>, Arc<ECStore>) {
+        let (dirs, store) = isolated_store_over_temp_disks().await;
+        init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+        store
+            .make_bucket(bucket, &MakeBucketOptions::default())
+            .await
+            .expect("create authoritative never-versioned bucket");
+        store.pools[0].disk_set[0].disks.write().await[0] = None;
+        let info = store
+            .get_bucket_info(bucket, &BucketOptions::default())
+            .await
+            .expect("existing bucket must remain readable at quorum");
+        assert_eq!(info.name, bucket);
+        let lookup = store
+            .get_object_info_for_delete(bucket, "never-existed", &ObjectOptions::default())
+            .await;
+        assert!(
+            lookup.as_ref().is_err_and(is_err_object_not_found),
+            "remaining disks must establish absence through production lookup: {lookup:?}"
+        );
+        (dirs, store)
+    }
+
+    #[tokio::test]
+    async fn single_delete_missing_key_preserves_quorum_noop() {
+        let (_dirs, store) = degraded_store("missing-key-quorum-single").await;
+        let result = store
+            .delete_object_with_tier_delete_journal("missing-key-quorum-single", "never-existed", ObjectOptions::default())
+            .await;
+        assert!(
+            result.as_ref().is_err_and(is_err_object_not_found),
+            "single missing-key delete must preserve its idempotent not-found outcome at quorum: {result:?}"
+        );
+    }
+
+    #[tokio::test]
+    async fn batch_delete_missing_key_preserves_quorum_noop() {
+        let (_dirs, store) = degraded_store("missing-key-quorum-batch").await;
+        let (_, errors, _) = store
+            .delete_objects_with_tier_delete_journal_and_accounting(
+                "missing-key-quorum-batch",
+                vec![ObjectToDelete {
+                    object_name: "never-existed".to_owned(),
+                    ..Default::default()
+                }],
+                ObjectOptions::default(),
+            )
+            .await;
+        assert_eq!(errors.len(), 1, "batch DELETE must return one result per key");
+        assert!(
+            errors.iter().all(Option::is_none),
+            "batch missing-key delete must preserve its idempotent success at quorum: {errors:?}"
+        );
+    }
+
+    #[cfg(not(windows))]
+    #[tokio::test]
+    async fn batch_delete_timeout_retains_physical_key_guard() {
+        use crate::disk::os::prepared_publication_test_hooks as hooks;
+        use crate::disk::{DiskAPI, ReadOptions};
+        use crate::storage_api_contracts::object::ObjectIO;
+        temp_env::async_with_vars([("RUSTFS_DELETE_RECOVERY_TIMEOUT_SECS", Some("30"))], async {
+            let (_dirs, store) = isolated_store_over_temp_disks().await;
+            init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+            let bucket = "review-batch-late-owner";
+            let key = "object";
+            store
+                .make_bucket(bucket, &MakeBucketOptions::default())
+                .await
+                .expect("fixture bucket");
+            let mut reader = PutObjReader::from_vec(vec![0x51; 1024 * 1024]);
+            store
+                .put_object(
+                    bucket,
+                    key,
+                    &mut reader,
+                    &ObjectOptions {
+                        write_completion: crate::object_api::WriteCompletion::TailDrained,
+                        ..Default::default()
+                    },
+                )
+                .await
+                .expect("seed real external object");
+            let disks = store.pools[0].disk_set[0].disk_inventory().await;
+            let disk = disks[0].as_ref().expect("online drive");
+            let info = disk
+                .read_version("", bucket, key, "", &ReadOptions::default())
+                .await
+                .expect("seed metadata");
+            assert!(!info.inline_data());
+            let crate::disk::Disk::Local(local) = disk.as_ref() else {
+                panic!("local disk");
+            };
+            let part = local
+                .get_disk()
+                .get_object_path_for_io(bucket, key)
+                .expect("leased key")
+                .join(info.data_dir.expect("external data directory").to_string())
+                .join("part.1");
+            let (entered_tx, entered_rx) = tokio::sync::oneshot::channel();
+            let (release, release_rx) = std::sync::mpsc::channel::<()>();
+            let hook = hooks::install_at(hooks::Stage::Remove, &part, move || {
+                let _ = entered_tx.send(());
+                let _ = release_rx.recv();
+            });
+            let deleting = store.clone();
+            let delete = tokio::spawn(async move {
+                deleting
+                    .delete_objects_with_tier_delete_journal_and_accounting(
+                        bucket,
+                        vec![ObjectToDelete {
+                            object_name: key.to_owned(),
+                            ..Default::default()
+                        }],
+                        ObjectOptions::default(),
+                    )
+                    .await
+            });
+            tokio::time::timeout(Duration::from_secs(10), entered_rx)
+                .await
+                .expect("physical cleanup entry")
+                .expect("entry hook");
+            assert!(part.exists(), "payload unlink has not run");
+            let (_, errors, _) = tokio::time::timeout(Duration::from_secs(40), delete)
+                .await
+                .expect("real settlement timeout must return")
+                .expect("delete task");
+            assert!(errors.iter().all(Option::is_none), "certified batch response: {errors:?}");
+            assert!(
+                store.ctx.namespace_commits_pending(),
+                "background physical operation must retain its owner"
+            );
+            let ns = store.handle_new_ns_lock(bucket, key).await.expect("same production key lock");
+            let probe = ns.get_write_lock(Duration::from_millis(300)).await;
+            let acquired_while_pending = probe.is_ok();
+            drop(probe);
+            drop(release);
+            drop(hook);
+            assert!(
+                !acquired_while_pending,
+                "batch released the exact-key guard while physical cleanup was still paused after its real recovery timeout"
+            );
+            let _released = ns
+                .get_write_lock(Duration::from_secs(10))
+                .await
+                .expect("physical cleanup must eventually release the exact-key guard");
+        })
+        .await;
+    }
+}

crates/ecstore/src/store/peer.rs

@@ -18,6 +18,11 @@ use crate::disk::error::{DiskError, Result as DiskResult};
 use crate::disk::{DeleteOptions, Disk, RenameDataGuards, RenameDataResp};
 use crate::runtime::instance::{InstanceContext, NamespaceCommitGuard};
 use crate::runtime::sources as runtime_sources;
+use rustfs_io_metrics::{
+    PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_ADMIT, PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_LOCAL_RENAME,
+    PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_LOOKUP, PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_OVERHEAD,
+    PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_TOTAL,
+};
 use tracing::{debug, error};
 
 const LOG_COMPONENT_ECSTORE: &str = "ecstore";
@@ -331,15 +336,38 @@ async fn rename_local_data_with_ctx(
     destination: (&str, &str),
     mut guards: RenameDataGuards,
 ) -> DiskResult<RenameDataResp> {
+    let handler_started = rustfs_io_metrics::put_stage_timer();
+    let lookup_started = rustfs_io_metrics::put_stage_timer();
     let (disk, disk_id) = local_disk_candidate(ctx, disk_ref).await?;
+    let lookup_ms = lookup_started.map(|started| started.elapsed().as_secs_f64() * 1000.0);
+    if let Some(duration_ms) = lookup_ms {
+        rustfs_io_metrics::record_put_object_stage_duration(PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_LOOKUP, duration_ms);
+    }
     let mutates_namespace = !is_meta_bucketname(source.0) || !is_meta_bucketname(destination.0);
+    let admit_started = rustfs_io_metrics::put_stage_timer();
     let owner = admit_local_disk(ctx, &disk, disk_id, mutates_namespace).await?;
+    let admit_ms = admit_started.map(|started| started.elapsed().as_secs_f64() * 1000.0);
+    if let Some(duration_ms) = admit_ms {
+        rustfs_io_metrics::record_put_object_stage_duration(PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_ADMIT, duration_ms);
+    }
     guards.namespace_owner = owner.as_ref().map(|owner| owner.clone() as Arc<dyn Send + Sync>);
+    let local_rename_started = rustfs_io_metrics::put_stage_timer();
     let result = disk
         .rename_data_borrowed_with_fence_observed(source.0, source.1, fi, destination.0, destination.1, guards)
         .await
         .result;
+    let local_rename_ms = local_rename_started.map(|started| started.elapsed().as_secs_f64() * 1000.0);
+    if let Some(duration_ms) = local_rename_ms {
+        rustfs_io_metrics::record_put_object_stage_duration(PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_LOCAL_RENAME, duration_ms);
+    }
     drop(owner);
+    if let Some(handler_started) = handler_started {
+        let handler_ms = handler_started.elapsed().as_secs_f64() * 1000.0;
+        rustfs_io_metrics::record_put_object_stage_duration(PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_TOTAL, handler_ms);
+        let child_ms = lookup_ms.unwrap_or(0.0) + admit_ms.unwrap_or(0.0) + local_rename_ms.unwrap_or(0.0);
+        let overhead_ms = (handler_ms - child_ms).max(0.0);
+        rustfs_io_metrics::record_put_object_stage_duration(PUT_STAGE_SET_DISK_RENAME_REMOTE_HANDLER_OVERHEAD, overhead_ms);
+    }
     result
 }
 
@@ -1669,7 +1697,9 @@ mod tests {
     #[serial_test::serial]
     async fn target_ordinary_timeout_keeps_its_physical_namespace_owner() {
         use crate::disk::os::prepared_publication_test_hooks as hooks;
-        temp_env::async_with_vars([(rustfs_config::ENV_DRIVE_MAX_TIMEOUT_DURATION, Some("1"))], async {
+        // Allow real fixture I/O to finish under load. Expire the operation
+        // with virtual time only after its physical publication is paused.
+        temp_env::async_with_vars([(rustfs_config::ENV_DRIVE_MAX_TIMEOUT_DURATION, Some("60"))], async {
             let ctx = Arc::new(InstanceContext::new());
             let store = super::super::tests::build_store_with_ctx(ctx.clone());
             let root = tempfile::tempdir().expect("root");
@@ -1703,7 +1733,7 @@ mod tests {
             .await
             .expect("bounded entry");
             tokio::time::pause();
-            tokio::time::advance(std::time::Duration::from_secs(2)).await;
+            tokio::time::advance(std::time::Duration::from_secs(61)).await;
             tokio::time::resume();
             let result = tokio::time::timeout(std::time::Duration::from_secs(5), &mut rename)
                 .await

crates/ecstore/tests/cleanup_isolation_test.rs

@@ -0,0 +1,174 @@
+// Copyright 2026 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.
+
+#![cfg(target_os = "linux")]
+
+use bytes::Bytes;
+mod storage_api;
+
+use std::time::Duration;
+use storage_api::cleanup_isolation::{DeleteOptions, DiskAPI, DiskError, DiskOption, Endpoint, new_disk};
+
+#[tokio::test]
+async fn cleanup_disk_and_periodic_gc_use_affinity_without_changing_foreground() {
+    const CHILD: &str = "RUSTFS_CLEANUP_LINUX_TEST_CHILD";
+    let foreground = rustix::thread::sched_getaffinity(None).unwrap();
+    let cpu = (0..rustix::thread::CpuSet::MAX_CPU)
+        .find(|cpu| foreground.is_set(*cpu))
+        .expect("the test process must have an allowed CPU");
+    if std::env::var_os(CHILD).is_none() {
+        for mode in ["custom", "default", "none", "restricted"] {
+            if mode == "default" && (!foreground.is_set(1) || !foreground.is_set(2)) {
+                continue;
+            }
+            let output = tokio::task::spawn_blocking(move || {
+                let mut command = std::process::Command::new(std::env::current_exe().unwrap());
+                command
+                    .args([
+                        "--exact",
+                        "cleanup_disk_and_periodic_gc_use_affinity_without_changing_foreground",
+                        "--nocapture",
+                    ])
+                    .env(CHILD, mode)
+                    .env("RUSTFS_CLEANUP_ISOLATE_ENABLE", "true")
+                    .env("RUSTFS_PUT_RENAME_TAIL_CLEANUP_MAX_PENDING", "4")
+                    .env("RUSTFS_PUT_RENAME_TAIL_CLEANUP_DEFER_WORKERS", "1")
+                    .env("RUSTFS_CLEANUP_DISK_MAX_PENDING", "4")
+                    .env("RUSTFS_CLEANUP_DISK_WORKERS", "1")
+                    .env("RUSTFS_CLEANUP_GC_WORKERS", "1")
+                    .env("RUSTFS_CLEANUP_BLOCKING_THREADS", "1")
+                    .env("RUSTFS_CLEANUP_ASYNC_THREADS", "1")
+                    .env_remove("RUSTFS_CLEANUP_CPUS")
+                    .env_remove("RUSTFS_PUT_RENAME_TAIL_CLEANUP_COUNTERFACTUAL_SKIP")
+                    .env_remove("RUSTFS_PUT_RENAME_TAIL_CLEANUP_DEFER_HOLD_WORKER")
+                    .env_remove("RUSTFS_PUT_RENAME_TAIL_CLEANUP_ZERO_TARGET_TMP_DELETE_SKIP");
+                match mode {
+                    "custom" => {
+                        command.env("RUSTFS_CLEANUP_CPUS", cpu.to_string());
+                    }
+                    "none" => {
+                        command.env("RUSTFS_CLEANUP_CPUS", "none");
+                    }
+                    _ => {}
+                }
+                command.output().unwrap()
+            })
+            .await
+            .unwrap();
+            assert!(
+                output.status.success(),
+                "{}\n{}",
+                String::from_utf8_lossy(&output.stdout),
+                String::from_utf8_lossy(&output.stderr)
+            );
+            assert!(String::from_utf8_lossy(&output.stdout).contains("1 passed"));
+        }
+        return;
+    }
+
+    let mode = std::env::var(CHILD).unwrap();
+    if mode == "restricted" {
+        let mut restricted = rustix::thread::CpuSet::new();
+        restricted.set(cpu);
+        rustix::thread::sched_setaffinity(None, &restricted).expect("restrict initialization to one allowed CPU");
+    }
+    let dir = tempfile::tempdir().unwrap();
+    let disk = new_disk(&Endpoint::try_from(dir.path().to_str().unwrap()).unwrap(), &DiskOption::default())
+        .await
+        .unwrap();
+    let bucket = "cleanup-affinity";
+    disk.make_volume(bucket).await.unwrap();
+    disk.write_all(bucket, "old/part.1", Bytes::from_static(b"garbage"))
+        .await
+        .unwrap();
+    let receipt = disk
+        .write_all(bucket, "old/.rustfs-old-data-cleanup-receipt.json", Bytes::from_static(b"receipt"))
+        .await;
+    if mode == "restricted" {
+        assert!(
+            receipt
+                .expect_err("unavailable default CPUs must reject cleanup")
+                .to_string()
+                .contains("allowed CPU set")
+        );
+        assert_eq!(disk.read_all(bucket, "old/part.1").await.unwrap(), Bytes::from_static(b"garbage"));
+        return;
+    }
+    receipt.unwrap();
+    let expected_mask = match mode.as_str() {
+        "default" => "1-2".to_owned(),
+        "none" => std::fs::read_to_string("/proc/thread-self/status")
+            .unwrap()
+            .lines()
+            .find_map(|line| line.strip_prefix("Cpus_allowed_list:"))
+            .unwrap()
+            .trim()
+            .to_owned(),
+        _ => cpu.to_string(),
+    };
+    disk.write_all(bucket, "live/part.1", Bytes::from_static(b"keep"))
+        .await
+        .unwrap();
+    disk.delete(
+        bucket,
+        "old",
+        DeleteOptions {
+            recursive: true,
+            ..Default::default()
+        },
+    )
+    .await
+    .unwrap();
+    assert!(matches!(disk.read_all(bucket, "old/part.1").await, Err(DiskError::FileNotFound)));
+
+    // One async worker and one lazily started blocking worker. Inspect the
+    // real process threads, including the threads executing filesystem work.
+    let mut cleanup_threads = 0;
+    for entry in std::fs::read_dir("/proc/self/task").unwrap() {
+        let task = entry.unwrap().path();
+        let Ok(name) = std::fs::read_to_string(task.join("comm")) else {
+            continue;
+        };
+        if name.trim() != "rustfs-cleanup" {
+            continue;
+        }
+        cleanup_threads += 1;
+        let status = std::fs::read_to_string(task.join("status")).unwrap();
+        let mask = status
+            .lines()
+            .find_map(|line| line.strip_prefix("Cpus_allowed_list:"))
+            .unwrap()
+            .trim();
+        assert_eq!(mask, expected_mask);
+    }
+    assert_eq!(cleanup_threads, 2, "async and blocking workers must both use the configured runtime");
+    assert_eq!(rustix::thread::sched_getaffinity(None).unwrap(), foreground);
+
+    let trash = dir.path().join(".rustfs.sys/tmp/.trash");
+    assert!(std::fs::read_dir(&trash).unwrap().next().is_some());
+    // Advance the foreground timer that dispatches the five-minute periodic
+    // scan; the dedicated runtime then performs the actual physical removal.
+    tokio::time::pause();
+    tokio::time::advance(Duration::from_secs(301)).await;
+    tokio::time::resume();
+    tokio::time::timeout(Duration::from_secs(20), async {
+        while std::fs::read_dir(&trash).unwrap().next().is_some() {
+            tokio::time::sleep(Duration::from_millis(5)).await;
+        }
+    })
+    .await
+    .expect("periodic GC must physically remove trash");
+    assert_eq!(disk.read_all(bucket, "live/part.1").await.unwrap(), Bytes::from_static(b"keep"));
+    assert_eq!(rustix::thread::sched_getaffinity(None).unwrap(), foreground);
+}

crates/ecstore/tests/fd_relative_io_test.rs

@@ -0,0 +1,283 @@
+// Copyright 2026 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.
+
+#![cfg(all(target_os = "linux", feature = "test-util"))]
+
+mod storage_api;
+
+use bytes::Bytes;
+use rustfs_filemeta::{FileInfo, ObjectPartInfo};
+use std::{path::Path, time::Duration};
+use storage_api::fd_relative::{
+    DiskAPI, DiskError, DiskOption, DiskStore, Endpoint, LocalPublicationPause, LocalPublicationStage, ReadOptions, new_disk,
+};
+use time::OffsetDateTime;
+use uuid::Uuid;
+
+const RUSTFS_META_TMP_BUCKET: &str = ".rustfs.sys/tmp";
+
+async fn disk_at(path: &Path) -> DiskStore {
+    new_disk(
+        &Endpoint::try_from(path.to_str().expect("UTF-8 test path")).expect("endpoint"),
+        &DiskOption::default(),
+    )
+    .await
+    .expect("local disk")
+}
+
+fn metadata(data: Bytes, inline: bool, directory: Uuid) -> FileInfo {
+    let mut fi = FileInfo::new("object", 1, 0);
+    fi.erasure.index = 1;
+    fi.version_id = Some(Uuid::nil());
+    fi.data_dir = Some(directory);
+    fi.size = i64::try_from(data.len()).expect("test data size");
+    fi.parts = vec![ObjectPartInfo {
+        number: 1,
+        size: data.len(),
+        actual_size: fi.size,
+        ..Default::default()
+    }];
+    fi.mod_time = Some(OffsetDateTime::now_utc());
+    fi.data = inline.then_some(data);
+    if inline {
+        fi.set_inline_data();
+    }
+    fi
+}
+
+async fn put(disk: &DiskStore, staging: &str, data: Bytes, inline: bool) -> Uuid {
+    let directory = Uuid::new_v4();
+    if !inline {
+        disk.write_all(RUSTFS_META_TMP_BUCKET, &format!("{staging}/{directory}/part.1"), data.clone())
+            .await
+            .expect("stage shard");
+    }
+    disk.rename_data(RUSTFS_META_TMP_BUCKET, staging, metadata(data, inline, directory), "bucket", "object")
+        .await
+        .expect("commit object");
+    directory
+}
+
+#[tokio::test]
+async fn pinned_root_reads_and_writes_original_disk_after_path_replacement() {
+    let temp = tempfile::tempdir().expect("test directory");
+    let root = temp.path().join("disk");
+    std::fs::create_dir(&root).expect("disk root");
+    let disk = disk_at(&root).await;
+    disk.make_volume("bucket").await.expect("bucket");
+    disk.write_all("bucket", "metadata", Bytes::from_static(b"original"))
+        .await
+        .expect("initial write");
+    let moved = temp.path().join("original");
+    std::fs::rename(&root, &moved).expect("move mount path");
+    std::fs::create_dir_all(root.join("bucket")).expect("replacement disk");
+    std::fs::write(root.join("bucket/metadata"), b"replacement").expect("replacement sentinel");
+    assert_eq!(
+        disk.read_all("bucket", "metadata").await.expect("pinned read"),
+        Bytes::from_static(b"original")
+    );
+    disk.write_all("bucket", "metadata", Bytes::from_static(b"updated"))
+        .await
+        .expect("pinned write");
+    assert_eq!(std::fs::read(moved.join("bucket/metadata")).expect("original disk"), b"updated");
+    assert_eq!(std::fs::read(root.join("bucket/metadata")).expect("replacement disk"), b"replacement");
+}
+
+#[tokio::test]
+async fn relative_reads_and_writes_reject_symlink_escape() {
+    let temp = tempfile::tempdir().expect("test directory");
+    let disk = disk_at(temp.path()).await;
+    disk.make_volume("bucket").await.expect("bucket");
+    let outside = tempfile::tempdir().expect("outside directory");
+    std::fs::write(outside.path().join("sentinel"), b"preserved").expect("sentinel");
+    std::os::unix::fs::symlink(outside.path(), temp.path().join("bucket/link")).expect("directory link");
+    std::os::unix::fs::symlink(outside.path().join("sentinel"), temp.path().join("bucket/file-link")).expect("file link");
+    for path in ["link/sentinel", "file-link"] {
+        assert!(matches!(disk.read_all("bucket", path).await, Err(DiskError::InvalidPath)));
+        assert!(matches!(
+            disk.write_all("bucket", path, Bytes::from_static(b"unsafe")).await,
+            Err(DiskError::InvalidPath)
+        ));
+    }
+    assert_eq!(std::fs::read(outside.path().join("sentinel")).expect("outside bytes"), b"preserved");
+}
+
+#[tokio::test]
+async fn inline_and_non_inline_overwrites_remain_readable_after_reopen() {
+    for inline in [true, false] {
+        let temp = tempfile::tempdir().expect("test directory");
+        let disk = disk_at(temp.path()).await;
+        disk.make_volume("bucket").await.expect("bucket");
+        put(&disk, "first", Bytes::from_static(b"old bytes"), inline).await;
+        let directory = put(&disk, "second", Bytes::from_static(b"new bytes"), inline).await;
+        drop(disk);
+        let disk = disk_at(temp.path()).await;
+        let fi = disk
+            .read_version(
+                "bucket",
+                "bucket",
+                "object",
+                "",
+                &ReadOptions {
+                    read_data: true,
+                    ..Default::default()
+                },
+            )
+            .await
+            .expect("committed metadata after reopen");
+        assert_eq!(fi.data_dir, Some(directory));
+        if inline {
+            assert_eq!(fi.data, Some(Bytes::from_static(b"new bytes")));
+        } else {
+            assert_eq!(
+                disk.read_all("bucket", &format!("object/{directory}/part.1"))
+                    .await
+                    .expect("shard bytes"),
+                Bytes::from_static(b"new bytes")
+            );
+        }
+    }
+}
+
+#[tokio::test]
+async fn rename_rejects_symlinked_destination_metadata() {
+    for inline in [true, false] {
+        let temp = tempfile::tempdir().expect("disk directory");
+        let outside = tempfile::tempdir().expect("outside directory");
+        let disk = disk_at(temp.path()).await;
+        disk.make_volume("bucket").await.expect("bucket");
+        put(&disk, "seed", Bytes::from_static(b"old object"), inline).await;
+        let destination = temp.path().join("bucket/object/xl.meta");
+        let original = std::fs::read(&destination).expect("valid existing metadata");
+        let external_metadata = outside.path().join("xl.meta");
+        std::fs::write(&external_metadata, &original).expect("external valid metadata");
+        std::fs::remove_file(&destination).expect("replace destination metadata");
+        std::os::unix::fs::symlink(&external_metadata, &destination).expect("metadata escape link");
+
+        let directory = Uuid::new_v4();
+        let data = Bytes::from_static(b"replacement");
+        if !inline {
+            disk.write_all(RUSTFS_META_TMP_BUCKET, &format!("replacement/{directory}/part.1"), data.clone())
+                .await
+                .expect("stage non-inline data");
+        }
+        let result = disk
+            .rename_data(
+                RUSTFS_META_TMP_BUCKET,
+                "replacement",
+                metadata(data, inline, directory),
+                "bucket",
+                "object",
+            )
+            .await;
+        assert!(
+            matches!(result, Err(DiskError::InvalidPath)),
+            "symlinked metadata must be rejected: {result:?}"
+        );
+        assert_eq!(std::fs::read_link(&destination).expect("link must remain"), external_metadata);
+        assert_eq!(std::fs::read(&external_metadata).expect("external metadata must survive"), original);
+    }
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn prepared_publication_rejects_source_or_directory_replacement() {
+    for replace_directory in [false, true] {
+        let temp = tempfile::tempdir().expect("test directory");
+        let disk = disk_at(temp.path()).await;
+        disk.make_volume("bucket").await.expect("bucket");
+        put(&disk, "first", Bytes::from_static(b"old bytes"), true).await;
+        let old = std::fs::read(temp.path().join("bucket/object/xl.meta")).expect("old metadata");
+        let mut pause = LocalPublicationPause::install(&disk, "bucket", "object/xl.meta", LocalPublicationStage::PreparedRename)
+            .expect("publication pause");
+        let writer = disk.clone();
+        let task = tokio::spawn(async move {
+            writer
+                .rename_data(
+                    RUSTFS_META_TMP_BUCKET,
+                    "second",
+                    metadata(Bytes::from_static(b"new bytes"), true, Uuid::new_v4()),
+                    "bucket",
+                    "object",
+                )
+                .await
+        });
+        tokio::time::timeout(Duration::from_secs(30), pause.entered())
+            .await
+            .expect("publication reached pause")
+            .expect("pause entered");
+        if replace_directory {
+            std::fs::rename(temp.path().join("bucket/object"), temp.path().join("detached")).expect("detach object");
+            std::fs::create_dir(temp.path().join("bucket/object")).expect("replacement directory");
+            std::fs::write(temp.path().join("bucket/object/xl.meta"), &old).expect("replacement metadata");
+        } else {
+            let src = temp.path().join(RUSTFS_META_TMP_BUCKET).join("second/xl.meta");
+            std::fs::rename(&src, src.with_extension("detached")).expect("detach staged metadata");
+            std::fs::write(src, b"unrelated").expect("replacement staged file");
+        }
+        drop(pause);
+        assert!(task.await.expect("writer task").is_err());
+        assert_eq!(
+            std::fs::read(temp.path().join("bucket/object/xl.meta")).expect("old metadata survives"),
+            old
+        );
+    }
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn cancelled_waiter_keeps_publication_serialized_until_physical_completion() {
+    let temp = tempfile::tempdir().expect("test directory");
+    let disk = disk_at(temp.path()).await;
+    disk.make_volume("bucket").await.expect("bucket");
+    put(&disk, "first", Bytes::from_static(b"old bytes"), true).await;
+    let mut pause = LocalPublicationPause::install(&disk, "bucket", "object/xl.meta", LocalPublicationStage::PreparedRename)
+        .expect("publication pause");
+    let writer = disk.clone();
+    let pending = tokio::spawn(async move { put(&writer, "shared-stage", Bytes::from_static(b"cancelled waiter"), true).await });
+    tokio::time::timeout(Duration::from_secs(30), pause.entered())
+        .await
+        .expect("publication reached pause")
+        .expect("pause entered");
+    pending.abort();
+    assert!(pending.await.expect_err("waiter aborted").is_cancelled());
+    let writer = disk.clone();
+    let (started_tx, started_rx) = tokio::sync::oneshot::channel();
+    let mut retry = tokio::spawn(async move {
+        let _ = started_tx.send(());
+        put(&writer, "shared-stage", Bytes::from_static(b"latest bytes"), true).await
+    });
+    started_rx.await.expect("retry started");
+    assert!(
+        tokio::time::timeout(Duration::from_millis(200), &mut retry).await.is_err(),
+        "a retry must not publish while the cancelled writer still owns physical IO"
+    );
+    drop(pause);
+    tokio::time::timeout(Duration::from_secs(30), retry)
+        .await
+        .expect("retry completed")
+        .expect("retry task");
+    let fi = disk
+        .read_version(
+            "bucket",
+            "bucket",
+            "object",
+            "",
+            &ReadOptions {
+                read_data: true,
+                ..Default::default()
+            },
+        )
+        .await
+        .expect("latest metadata");
+    assert_eq!(fi.data, Some(Bytes::from_static(b"latest bytes")));
+}