ci(s3tests): omit proxy query strings

ci(s3tests): add targeted proxy diagnostics
fix(app): wait for peer bucket metadata reload
2026-08-22 20:36:38 +00:00 · 2026-08-22 19:15:55 +08:00 · 2026-08-22 18:53:25 +08:00 · 2026-08-22 18:52:24 +08:00
135 changed files with 1935 additions and 13827 deletions
@@ -1,20 +0,0 @@
-# Report-only calibration baseline from https://github.com/rustfs/rustfs/actions/runs/29394996173.
-# Update counts only with a linked coverage run and a reviewed explanation.
-phase = "report-only"
-allowed_drop_percentage_points = 1.0
-
-[crates."crates/iam"]
-covered = 5149
-count = 8131
-
-[crates."crates/kms"]
-covered = 2950
-count = 4200
-
-[crates."crates/policy"]
-covered = 4636
-count = 5464
-
-[crates."crates/crypto"]
-covered = 469
-count = 494
@@ -1,2 +1,2 @@
-sha256-darwin=9f767b37ed8b1c82da62ea441462d75487785c8086e56f08fb6f6cd89c6e2e52
-sha256-linux=fbdaf42b220958d4b1e8880e0f8b5a7992d38e21051bb60596dd4538424757d6
+sha256-darwin=b4ae71aa894e5c7795ae3eb8116f1777a7601d0f5db3898be2e48faf3329bd9b
+sha256-linux=433debd9d9defa832986269abdf0f1d131597b2d7a417ce930e17c1fd47d85ba
@@ -36,8 +36,6 @@ script-tests: ## Run shell script tests
 	./scripts/test_manual_transition_runbooks.sh
 	./scripts/check_embedded_secrets.sh --self-test
 	python3 ./scripts/check_test_wiring.py --self-test
-	python3 ./scripts/check_security_coverage.py --self-test
-	python3 ./scripts/check_scheduled_validation_freshness.py --self-test
 	python3 ./scripts/s3-tests/test_report_compat.py
 	bash -n ./scripts/validate_object_data_cache_cold_stampede.sh
 	python3 ./scripts/check_object_data_cache_follower_samples.py --self-test
@@ -14,10 +14,9 @@

 name: "Schedule Failure Issue"
 description: >-
-  Open (or update) a tracking issue when a scheduled workflow run fails or
-  does not complete normally.
+  Open (or update) a tracking issue when a scheduled workflow run fails.
  Dedupes by workflow name: if an open issue titled
-  "[scheduled-failure] <workflow name>" already exists, the result is
+  "[scheduled-failure] <workflow name>" already exists, the failure is
  appended as a comment; otherwise a new issue is created. This is the
  single alerting mechanism for all scheduled pipelines (backlog#1149 ci-8).

@@ -39,30 +38,6 @@ inputs:
      Set to an empty string to skip labeling.
    required: false
    default: "infrastructure"
-  source-run-id:
-    description: "Run ID to report. Defaults to the current workflow run."
-    required: false
-    default: ${{ github.run_id }}
-  source-run-attempt:
-    description: "Run attempt to report. Defaults to the current attempt."
-    required: false
-    default: ${{ github.run_attempt }}
-  source-event:
-    description: "Trigger event of the run being reported."
-    required: false
-    default: ${{ github.event_name }}
-  source-ref-name:
-    description: "Ref name of the run being reported."
-    required: false
-    default: ${{ github.ref_name }}
-  source-sha:
-    description: "Commit SHA of the run being reported."
-    required: false
-    default: ${{ github.sha }}
-  details-file:
-    description: "Optional Markdown file appended to the issue body."
-    required: false
-    default: ""

 runs:
  using: "composite"
@@ -73,22 +48,17 @@ runs:
        GH_TOKEN: ${{ inputs.github-token }}
        WORKFLOW_NAME: ${{ inputs.workflow-name }}
        ISSUE_LABEL: ${{ inputs.label }}
-        SOURCE_RUN_ID: ${{ inputs.source-run-id }}
-        SOURCE_RUN_ATTEMPT: ${{ inputs.source-run-attempt }}
-        SOURCE_EVENT: ${{ inputs.source-event }}
-        SOURCE_REF_NAME: ${{ inputs.source-ref-name }}
-        SOURCE_SHA: ${{ inputs.source-sha }}
-        DETAILS_FILE: ${{ inputs.details-file }}
      run: |
        set -euo pipefail

        title="[scheduled-failure] ${WORKFLOW_NAME}"
-        run_url="${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${SOURCE_RUN_ID}"
+        run_url="${GITHUB_SERVER_URL}/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}"

-        # Inspect the reported run attempt. It can be the current in-workflow
-        # failure or a completed run observed by the external watchdog.
+        # Failed job names for this run attempt. The alert job runs while the
+        # run as a whole is still in progress, so inspect the jobs that have
+        # already completed with a non-success conclusion.
        failed_jobs="$(gh api \
-          "repos/${GITHUB_REPOSITORY}/actions/runs/${SOURCE_RUN_ID}/attempts/${SOURCE_RUN_ATTEMPT}/jobs" \
+          "repos/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}/attempts/${GITHUB_RUN_ATTEMPT}/jobs" \
          --paginate \
          --jq '.jobs[]
                | select(.conclusion == "failure" or .conclusion == "timed_out" or .conclusion == "cancelled")
@@ -97,26 +67,15 @@ runs:
          failed_jobs="- (failed job not recorded yet — see the run page)"
        fi

-        details=""
-        if [ -n "${DETAILS_FILE}" ]; then
-          if [ -f "${DETAILS_FILE}" ]; then
-            details="$(cat "${DETAILS_FILE}")"
-          else
-            details="Details file was not available: \`${DETAILS_FILE}\`"
-          fi
-        fi
-
        body="$(cat <<EOF
-        Run of **${WORKFLOW_NAME}** did not complete successfully.
+        Scheduled run of **${WORKFLOW_NAME}** failed.

-        - Run: ${run_url} (attempt ${SOURCE_RUN_ATTEMPT})
-        - Event: \`${SOURCE_EVENT}\`
-        - Ref: \`${SOURCE_REF_NAME}\` @ \`${SOURCE_SHA}\`
+        - Run: ${run_url} (attempt ${GITHUB_RUN_ATTEMPT})
+        - Event: \`${GITHUB_EVENT_NAME}\`
+        - Ref: \`${GITHUB_REF_NAME}\` @ \`${GITHUB_SHA}\`

-        Non-success jobs:
+        Failed jobs:
        ${failed_jobs}
-
-        ${details}
        EOF
        )"

@@ -1,14 +0,0 @@
-[
-  { "workflow": ".github/workflows/audit.yml", "max_age_hours": 36 },
-  { "workflow": ".github/workflows/build.yml", "max_age_hours": 192 },
-  { "workflow": ".github/workflows/ci.yml", "max_age_hours": 192 },
-  { "workflow": ".github/workflows/coverage.yml", "max_age_hours": 192 },
-  { "workflow": ".github/workflows/e2e-replication-nightly.yml", "max_age_hours": 36 },
-  { "workflow": ".github/workflows/e2e-s3tests.yml", "max_age_hours": 192 },
-  { "workflow": ".github/workflows/fuzz.yml", "max_age_hours": 36 },
-  { "workflow": ".github/workflows/mint.yml", "max_age_hours": 192 },
-  { "workflow": ".github/workflows/minio-interop.yml", "max_age_hours": 36 },
-  { "workflow": ".github/workflows/nightly-gnu.yml", "max_age_hours": 36 },
-  { "workflow": ".github/workflows/performance-ab.yml", "max_age_hours": 36 },
-  { "workflow": ".github/workflows/runner-hygiene.yml", "max_age_hours": 792 }
-]
@@ -46,7 +46,7 @@ on:
    # advisory could sit unnoticed for seven days. The check list is unchanged —
    # splitting it into a light daily advisories-only run and a weekly full run
    # would create runs where sources/bans/licenses go unverified.
-    - cron: '23 3 * * *' # Daily 03:23 UTC
+    - cron: '0 3 * * *' # Daily 03:00 UTC (staggered after the midnight ci/build crons)
  workflow_dispatch:

 permissions:
@@ -52,7 +52,7 @@ on:
      - ".dockerignore"
      - "flake.lock"
  schedule:
-    - cron: "13 1 * * 0" # Weekly on Sunday 01:13 UTC
+    - cron: "0 1 * * 0" # Weekly on Sunday 01:00 UTC (staggered after the ci.yml midnight cron)
  workflow_dispatch:
    inputs:
      build_docker:
@@ -1032,23 +1032,3 @@ jobs:

          echo "🎉 Released $TAG successfully!"
          echo "📄 Release URL: ${{ needs.create-release.outputs.release_url }}"
-
-  alert-on-failure:
-    name: Alert on scheduled failure
-    needs: [build-check, prepare-platform-matrix, build-rustfs, build-summary]
-    if: >-
-      always() && github.event_name == 'schedule' &&
-      (contains(needs.*.result, 'failure') || contains(needs.*.result, 'cancelled'))
-    runs-on: ubuntu-latest
-    timeout-minutes: 10
-    permissions:
-      contents: read
-      issues: write
-    steps:
-      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
-        with:
-          persist-credentials: false
-      - name: Open or update failure-tracking issue
-        uses: ./.github/actions/schedule-failure-issue
-        with:
-          github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -94,9 +94,6 @@ concurrency:

 env:
  CARGO_TERM_COLOR: always
-  # Swatinem/rust-cache hashes every RUST* variable. Keep this aligned with
-  # ci.yml or the writer and readers use disjoint cache keys.
-  RUST_BACKTRACE: 1

 jobs:
  # Readers: test-and-lint, test-ilm-integration-serial, build-rustfs-debug-binary,
@@ -104,7 +101,7 @@ jobs:
  warm-ci-dev:
    name: Warm ci-dev
    runs-on: sm-standard-4
-    timeout-minutes: 120
+    timeout-minutes: 90
    env:
      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
    steps:
@@ -194,7 +191,7 @@ jobs:
  warm-ci-feat-rio:
    name: Warm ci-feat-rio
    runs-on: sm-standard-4
-    timeout-minutes: 120
+    timeout-minutes: 90
    env:
      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
    steps:
@@ -222,7 +219,7 @@ jobs:
  warm-ci-feat-proto:
    name: Warm ci-feat-proto
    runs-on: sm-standard-4
-    timeout-minutes: 120
+    timeout-minutes: 90
    env:
      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
    steps:
@@ -126,10 +126,7 @@ jobs:
        run: ./scripts/check_embedded_secrets.sh

      - name: Check test wiring
-        run: |
-          python3 ./scripts/check_test_wiring.py --self-test
-          python3 ./scripts/check_scheduled_validation_freshness.py --self-test
-          python3 ./scripts/check_test_wiring.py
+        run: python3 ./scripts/check_test_wiring.py

      - name: Check no planning docs committed
        run: ./scripts/check_no_planning_docs.sh
@@ -59,7 +59,7 @@ on:
  merge_group:
    types: [ checks_requested ]
  schedule:
-    - cron: "11 0 * * 0" # Weekly on Sunday 00:11 UTC
+    - cron: "0 0 * * 0" # Weekly on Sunday at midnight UTC
  workflow_dispatch:

 permissions:
@@ -161,10 +161,7 @@ jobs:
        run: ./scripts/check_embedded_secrets.sh

      - name: Check test wiring
-        run: |
-          python3 ./scripts/check_test_wiring.py --self-test
-          python3 ./scripts/check_scheduled_validation_freshness.py --self-test
-          python3 ./scripts/check_test_wiring.py
+        run: python3 ./scripts/check_test_wiring.py

      - name: Check no planning docs committed
        run: ./scripts/check_no_planning_docs.sh
@@ -1035,37 +1032,3 @@ jobs:
          path: artifacts/s3tests-single/**
          if-no-files-found: ignore
          retention-days: 3
-
-  alert-on-failure:
-    name: Alert on scheduled failure
-    needs:
-      - typos
-      - quick-checks
-      - test-and-lint
-      - test-ilm-integration-serial
-      - test-and-lint-rio-v2
-      - test-and-lint-protocols
-      - build-rustfs-debug-binary
-      - build-rustfs-debug-binary-rio-v2
-      - uring-integration
-      - e2e-tests
-      - e2e-full
-      - e2e-tests-rio-v2
-      - s3-implemented-tests
-      - s3-lifecycle-behavior-tests
-    if: >-
-      always() && github.event_name == 'schedule' &&
-      (contains(needs.*.result, 'failure') || contains(needs.*.result, 'cancelled'))
-    runs-on: ubuntu-latest
-    timeout-minutes: 10
-    permissions:
-      contents: read
-      issues: write
-    steps:
-      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
-        with:
-          persist-credentials: false
-      - name: Open or update failure-tracking issue
-        uses: ./.github/actions/schedule-failure-issue
-        with:
-          github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -12,12 +12,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.

-# Workspace line-coverage baseline and security-crate calibration
-# (backlog#1153 infra-5/infra-6).
+# Weekly workspace line-coverage baseline (backlog#1153 infra-5).
 #
-# NON-BLOCKING by design: the weekly job gives coverage a visible baseline and
-# trend, while relevant pull requests run a report-only security-crate
-# comparison. Neither job is a required check during calibration.
+# NON-BLOCKING by design: this workflow only runs on schedule and manual
+# dispatch, so it never attaches a status to a PR and must never be made a
+# required check. It exists to give coverage a visible baseline and trend
+# (per-crate table in the job summary, lcov artifact kept 90 days) — the
+# per-crate ratchet for the security-critical crates builds on it later
+# (backlog#1153 infra-6, report-only first per the ci-11 ladder).
 #
 # Measurement scope matches the PR test gate (ci.yml "Run tests"):
 # `--workspace --exclude e2e_test` with the `ci` nextest profile. Doctests are
@@ -29,28 +31,13 @@
 name: coverage

 on:
-  pull_request:
-    branches: [main]
-    paths:
-      - "crates/iam/**"
-      - "crates/kms/**"
-      - "crates/policy/**"
-      - "crates/crypto/**"
-      - ".config/coverage-baselines.toml"
-      - "scripts/coverage_per_crate.py"
-      - "scripts/check_security_coverage.py"
-      - ".github/workflows/coverage.yml"
  workflow_dispatch:
  schedule:
    # 07:00 UTC Sunday — staggered clear of the other Sunday crons: ci (00:00),
    # build (01:00), e2e-s3tests (02:00), audit (03:00), nix-flake-update
    # (05:00), mint (06:00), and the daily fuzz (02:00), minio-interop (03:17),
    # e2e-replication-nightly (04:00) and performance-ab (06:00) lanes.
-    - cron: "43 7 * * 0"
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.event_name }}-${{ github.event.pull_request.number || github.ref }}
-  cancel-in-progress: ${{ github.event_name != 'schedule' }}
+    - cron: "0 7 * * 0"

 # Only alert-on-failure needs more than read access; it declares its own
 # job-level `issues: write`.
@@ -59,13 +46,12 @@ permissions:

 jobs:
  coverage:
-    name: Workspace line coverage
+    name: Workspace coverage (weekly)
    runs-on: sm-standard-4
    # The instrumented build cannot reuse the regular CI cache (different
-    # RUSTFLAGS), so a cold run rebuilds the workspace before running the
-    # full suite. Exact-head run 32573798257 needed 119m42s including reports
-    # and artifact upload, so keep a bounded 30-minute publication margin.
-    timeout-minutes: 150
+    # RUSTFLAGS), so a cold week rebuilds the workspace before running the
+    # full suite; give it double the test job's 60-minute budget.
+    timeout-minutes: 120
    env:
      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
      # Match the PR gate's nextest semantics (ci.yml runs `--profile ci`):
@@ -105,9 +91,7 @@ jobs:
          cargo llvm-cov report --json --output-path target/llvm-cov/coverage.json

      - name: Write per-crate summary
-        run: |
-          python3 scripts/coverage_per_crate.py target/llvm-cov/coverage.json >> "$GITHUB_STEP_SUMMARY"
-          python3 scripts/check_security_coverage.py target/llvm-cov/coverage.json >> "$GITHUB_STEP_SUMMARY"
+        run: python3 scripts/coverage_per_crate.py target/llvm-cov/coverage.json >> "$GITHUB_STEP_SUMMARY"

      - name: Upload coverage artifact
        if: always()
@@ -40,7 +40,7 @@ on:
  schedule:
    # 04:00 UTC nightly — staggered clear of fuzz/e2e-s3tests (02:00),
    # stale (01:30) and performance-ab (06:00).
-    - cron: "29 4 * * *"
+    - cron: "0 4 * * *"

 # Only alert-on-failure needs more than read access; it declares its own
 # job-level `issues: write`.
@@ -196,9 +196,6 @@ jobs:
          cache-save-if: 'false'
          install-build-packaging-tools: 'false'

-      - name: Verify protocol socket oracle
-        run: ss -tn state CLOSE-WAIT >/dev/null
-
      # The suite owns fixed protocol ports and serializes its internal cases.
      - name: Verify protocol e2e membership
        env:
@@ -97,7 +97,7 @@ on:
  schedule:
    # Weekly full sweep (Sunday 02:00 UTC): full suite, run against BOTH the
    # single-node and the 4-node distributed topologies (matrix below).
-    - cron: "19 2 * * 0"
+    - cron: "0 2 * * 0"

 env:
  # main user
@@ -287,7 +287,6 @@ jobs:

          frontend fe_s3
            bind *:9000
-            option http-buffer-request
            default_backend be_s3

          backend be_s3
@@ -30,7 +30,7 @@ on:
      - "Cargo.lock"
      - ".github/workflows/fuzz.yml"
  schedule:
-    - cron: "17 2 * * *"
+    - cron: "0 2 * * *"
  workflow_dispatch:
    inputs:
      profile:
@@ -121,21 +121,3 @@ jobs:
          cargo nextest run --run-ignored ignored-only --no-tests=fail \
            -p "$INTEROP_PACKAGE" --features "$INTEROP_FEATURES" \
            -E "$INTEROP_FILTER"
-
-  alert-on-failure:
-    name: Alert on scheduled failure
-    needs: [minio-interop]
-    if: always() && github.event_name == 'schedule' && contains(needs.*.result, 'failure')
-    runs-on: ubuntu-latest
-    timeout-minutes: 10
-    permissions:
-      contents: read
-      issues: write
-    steps:
-      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
-        with:
-          persist-credentials: false
-      - name: Open or update failure-tracking issue
-        uses: ./.github/actions/schedule-failure-issue
-        with:
-          github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -45,6 +45,13 @@
 # docker-capable self-hosted `dind-sm-standard-2` label was the alternative but
 # has fewer cores and reintroduces fleet-state risk for no reliability gain.

+# DISABLED. This workflow is switched off in the repository's Actions settings
+# (state: disabled_manually) and does not run on any trigger, including its cron
+# and workflow_dispatch. That state lives in GitHub's UI and is invisible when
+# reading this file, which has already misled at least one audit — hence this
+# banner. Re-enabling is a UI action; anyone doing so should first check that the
+# workflow still matches the current CI layout. See rustfs/backlog#1603.
+#
 name: mint

 on:
@@ -63,13 +70,13 @@ on:
          - core
          - full
      mint-image:
-        description: "Mint image reference (empty = pinned default)"
+        description: "Mint image reference"
        required: false
-        default: ""
+        default: "minio/mint:edge"
  schedule:
    # Weekly, after the Sunday s3-tests full sweep (starts 02:00 UTC, up to
    # 3h) has finished, so the two never contend for the same runner pool.
-    - cron: "41 6 * * 0"
+    - cron: "0 6 * * 0"

 env:
  S3_ACCESS_KEY: rustfsadmin-ci
@@ -16,7 +16,7 @@ name: Nightly GNU Build

 on:
  schedule:
-    - cron: "7 0 * * *"
+    - cron: "0 0 * * *"
      timezone: "Asia/Shanghai"
  workflow_dispatch:

@@ -194,23 +194,3 @@ jobs:

      - name: Run HA leader failover live checks (three-node Raft cluster in Docker)
        run: bash scripts/test/vault_ha_kms_live.sh
-
-  alert-on-failure:
-    name: Alert on scheduled failure
-    needs: [build, kms-vault-lane, kms-vault-ha-failover]
-    if: >-
-      always() && github.event_name == 'schedule' &&
-      (contains(needs.*.result, 'failure') || contains(needs.*.result, 'cancelled'))
-    runs-on: ubuntu-latest
-    timeout-minutes: 10
-    permissions:
-      contents: read
-      issues: write
-    steps:
-      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
-        with:
-          persist-credentials: false
-      - name: Open or update failure-tracking issue
-        uses: ./.github/actions/schedule-failure-issue
-        with:
-          github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -22,15 +22,22 @@
 # correctness cost (e.g. the #4221 fsync durability fix) is recorded, not
 # blocked (rustfs/backlog#935 correction 1).

+# DISABLED. This workflow is switched off in the repository's Actions settings
+# (state: disabled_manually) and does not run on any trigger, including its cron
+# and workflow_dispatch. That state lives in GitHub's UI and is invisible when
+# reading this file, which has already misled at least one audit — hence this
+# banner. Re-enabling is a UI action; anyone doing so should first check that the
+# workflow still matches the current CI layout. See rustfs/backlog#1603.
+#
 name: Performance A/B

 on:
  schedule:
-    - cron: "31 6 * * *" # 06:31 UTC nightly, against main
+    - cron: "0 6 * * *" # 06:00 UTC nightly, against main
  workflow_dispatch:
    inputs:
      duration:
-        description: "warp duration per round"
+        description: "warp duration per round (short by default to fit the double-build budget)"
        required: false
        default: "12s"
        type: string
@@ -39,8 +46,12 @@ on:
        required: false
        default: false
        type: boolean
+  push:
+    # Every main commit pre-builds and caches its release binary (perf-3) so the
+    # nightly A/B restores a ready baseline instead of paying the double build.
+    branches: [main]
+
 permissions:
-  actions: read
  contents: read

 env:
@@ -48,19 +59,83 @@ env:
  RUST_BACKTRACE: 1

 jobs:
-  warp-ab:
-    name: Warp A/B budget gate
+  # perf-3: on every push to main, build the release binary once and cache it
+  # keyed by commit SHA (rustfs-baseline-<sha>). The warp-ab measurements
+  # restore this instead of paying the ~32min-per-side source
+  # build. That double build is what pushed the expanded 24-cell nightly past its
+  # ceiling — 2026-07-11..07-14 all cancelled on the 120min timeout. Incremental
+  # builds off the shared cargo cache keep each push cheap, and building on the
+  # same sm-standard-2 runner the A/B measures on guarantees the cached binary is
+  # ABI-identical. Do NOT source this from build.yml's per-merge artifact: those
+  # are cancelled ~7/8 of the time and are not a reliable baseline.
+  build-baseline-cache:
+    name: Build + cache baseline binary
+    if: github.event_name == 'push'
    runs-on: sm-standard-2
-    # A normal nightly restores the last successful binary and builds only the
-    # candidate; daily access keeps that cache warm. A cache miss may build both
-    # and needs room for the A/B run plus artifact and cache publication.
-    timeout-minutes: 180
+    # Latest-wins: consumers only ever restore the binary for the *current*
+    # origin/main tip, so when pushes land faster than the ~65min build, a
+    # superseded build's output is dead weight — cancel it instead of stacking
+    # hour-long jobs on the shared runner pool. A skipped intermediate SHA at
+    # most costs one same-commit self-heal in the A/B job.
+    concurrency:
+      group: perf-baseline-build-main
+      cancel-in-progress: true
+    # #4806 put thin LTO + codegen-units=1 on [profile.release], pushing a
+    # single release build past 60min on this runner — every cache build on
+    # 2026-07-15 died on the old 60min ceiling ("exceeded the maximum execution
+    # time of 1h0m0s") and the cache never populated. The measured binary must
+    # keep the production profile, so the budget absorbs the build instead.
+    timeout-minutes: 100
    steps:
      - name: Checkout repository
        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
        with:
          persist-credentials: false
-          fetch-depth: 0 # baseline may be an earlier successful scheduled head
+
+      - name: Setup Rust environment
+        uses: ./.github/actions/setup
+        with:
+          rust-version: stable
+          cache-shared-key: warp-ab-${{ hashFiles('**/Cargo.lock') }}
+          cache-save-if: ${{ github.ref == 'refs/heads/main' }}
+
+      - name: Build release rustfs
+        run: cargo build --release --bin rustfs
+
+      - name: Stage binary for cache
+        run: |
+          set -euo pipefail
+          mkdir -p baseline-bin
+          cp target/release/rustfs baseline-bin/rustfs
+
+      - name: Cache baseline binary by SHA
+        uses: actions/cache/save@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6
+        with:
+          path: baseline-bin/rustfs
+          key: rustfs-baseline-${{ github.sha }}
+
+  warp-ab:
+    name: Warp A/B budget gate
+    # Always run on schedule / manual dispatch. Never on push — that event only
+    # feeds build-baseline-cache above.
+    if: >-
+      github.event_name == 'schedule' ||
+      github.event_name == 'workflow_dispatch'
+    runs-on: sm-standard-2
+    # With perf-3's cached baseline binary the common (cache-hit) nightly is
+    # measurement-only and finishes well under 50min. This ceiling stays
+    # generous only to absorb the same-commit cache-miss self-heal (~65min
+    # single build with the post-#4806 LTO profile + measurement). A timeout
+    # surfaces via the alert-on-failure job (it fires on cancelled/timed-out,
+    # not just failure). perf-6 recalibrates the budget once the noise study
+    # lands.
+    timeout-minutes: 120
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
+        with:
+          persist-credentials: false
+          fetch-depth: 0 # baseline is built from origin/main

      - name: Setup Rust environment
        uses: ./.github/actions/setup
@@ -88,55 +163,24 @@ jobs:
          fi
          echo "allow_regression=$allow" >> "$GITHUB_OUTPUT"

-      # A failed regression run must keep comparing against the last known-good
-      # scheduled head. Otherwise the next nightly would absorb the regression
-      # into its baseline and turn green without a fix.
-      - name: Find last successful scheduled baseline
-        id: scheduled_baseline
-        if: github.event_name == 'schedule'
-        uses: actions/github-script@ed597411d8f924073f98dfc5c65a23a2325f34cd # v8
-        with:
-          result-encoding: string
-          script: |
-            const { data } = await github.rest.actions.listWorkflowRuns({
-              owner: context.repo.owner,
-              repo: context.repo.repo,
-              workflow_id: "performance-ab.yml",
-              event: "schedule",
-              status: "success",
-              per_page: 1,
-            });
-            return data.workflow_runs[0]?.head_sha ?? "";
-
-      # Manual runs compare a selected ref with current main. Scheduled runs
-      # compare current main with the last successful scheduled head. With no
-      # history, the first run measures the candidate against itself and seeds
-      # that head only if the complete rig succeeds.
+      # perf-3: resolve the commits so the cache can be keyed by SHA. The
+      # baseline is origin/main; the candidate is the checked-out ref. On the
+      # nightly (checkout == main) they are the same commit, so one cached binary
+      # serves both phases and the run does zero source builds.
      - name: Resolve baseline / candidate commits
        id: commits
-        env:
-          SCHEDULED_BASELINE_SHA: ${{ steps.scheduled_baseline.outputs.result }}
        run: |
          set -euo pipefail
+          baseline_sha="$(git rev-parse origin/main)"
          candidate_sha="$(git rev-parse HEAD)"
-          if [[ "${{ github.event_name }}" == "schedule" ]]; then
-            baseline_sha="${SCHEDULED_BASELINE_SHA:-$candidate_sha}"
-            if ! git merge-base --is-ancestor "$baseline_sha" "$candidate_sha"; then
-              echo "::error::scheduled baseline $baseline_sha is not an ancestor of candidate $candidate_sha" >&2
-              exit 1
-            fi
-          else
-            baseline_sha="$(git rev-parse origin/main)"
-          fi
-          git cat-file -e "${baseline_sha}^{commit}"
          echo "baseline_sha=$baseline_sha" >> "$GITHUB_OUTPUT"
          echo "candidate_sha=$candidate_sha" >> "$GITHUB_OUTPUT"
          echo "baseline commit: $baseline_sha"
          echo "candidate commit: $candidate_sha"

-      # Exact-key restore of the candidate binary saved by its successful
-      # scheduled run. A miss leaves cache-hit unset and falls back to a source
-      # build of that known-good head.
+      # Exact-key restore of the baseline binary built by build-baseline-cache
+      # when origin/main last landed. A miss (binary evicted or not built yet)
+      # leaves cache-hit unset and the rig falls back to a source build.
      - name: Restore cached baseline binary
        id: baseline_cache
        uses: actions/cache/restore@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6
@@ -226,11 +270,11 @@ jobs:
            elif [[ "$selfheal_built" == "true" ]]; then
              base_src="source build (cache self-heal, saved as rustfs-baseline-$baseline_sha)"
            else
-              base_src="isolated baseline source build (saved as rustfs-baseline-$baseline_sha)"
+              base_src="isolated origin/main source build (saved as rustfs-baseline-$baseline_sha)"
            fi
            if [[ "$candidate_sha" == "$baseline_sha" ]]; then
-              # No commits landed since the last successful baseline, so reuse
-              # the one binary for both phases and measure only rig drift.
+              # Nightly on main: the candidate is the same commit as the baseline,
+              # so reuse the one binary for both phases and skip all builds.
              args+=(--candidate-bin "$base_bin")
              cand_src="same binary as baseline (same commit)"
            elif [[ "$candidate_built" == "true" ]]; then
@@ -318,23 +362,6 @@ jobs:
            fi
          } >> "$GITHUB_STEP_SUMMARY"

-      - name: Stage successful candidate baseline
-        if: >-
-          steps.ab.outputs.status == '0' &&
-          steps.commits.outputs.baseline_sha != steps.commits.outputs.candidate_sha
-        run: |
-          set -euo pipefail
-          cp candidate-bin/rustfs baseline-bin/rustfs
-
-      - name: Cache successful candidate baseline
-        if: >-
-          steps.ab.outputs.status == '0' &&
-          steps.commits.outputs.baseline_sha != steps.commits.outputs.candidate_sha
-        uses: actions/cache/save@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6
-        with:
-          path: baseline-bin/rustfs
-          key: rustfs-baseline-${{ steps.commits.outputs.candidate_sha }}
-
      # Scheduled failure alerting is handled by the alert-on-failure job below
      # (perf-2 consuming ci-8's schedule-failure-issue composite action).

@@ -30,7 +30,7 @@ name: Runner Hygiene

 on:
  schedule:
-    - cron: "37 6 1 * *" # Monthly, 1st at 06:37 UTC
+    - cron: "0 6 1 * *" # Monthly, 1st at 06:00 UTC (after the daily audit cron)
  workflow_dispatch:

 permissions:
@@ -1,57 +0,0 @@
-# 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: Scheduled Validation Freshness
-
-on:
-  schedule:
-    - cron: "47 23 * * *"
-  workflow_dispatch:
-
-permissions:
-  contents: read
-
-concurrency:
-  group: scheduled-validation-freshness
-  cancel-in-progress: false
-
-jobs:
-  check-freshness:
-    name: Check scheduled validation freshness
-    runs-on: ubuntu-latest
-    timeout-minutes: 10
-    permissions:
-      actions: read
-      contents: read
-      issues: write
-    steps:
-      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
-        with:
-          persist-credentials: false
-      - name: Check latest scheduled runs
-        env:
-          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          set +e
-          python3 scripts/check_scheduled_validation_freshness.py \
-            --report "${RUNNER_TEMP}/scheduled-validation-freshness.md"
-          status=$?
-          cat "${RUNNER_TEMP}/scheduled-validation-freshness.md" >> "${GITHUB_STEP_SUMMARY}"
-          exit "${status}"
-      - name: Open or update freshness issue
-        if: failure()
-        uses: ./.github/actions/schedule-failure-issue
-        with:
-          github-token: ${{ secrets.GITHUB_TOKEN }}
-          details-file: ${{ runner.temp }}/scheduled-validation-freshness.md
@@ -1,63 +0,0 @@
-# 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: Scheduled Validation Watchdog
-
-on:
-  workflow_run:
-    workflows:
-      - "Security Audit"
-      - "Build and Release"
-      - "Continuous Integration"
-      - "coverage"
-      - "e2e-nightly"
-      - "e2e-s3tests"
-      - "Fuzz"
-      - "mint"
-      - "minio-interop"
-      - "Nightly GNU Build"
-      - "Performance A/B"
-      - "Runner Hygiene"
-    types: [completed]
-
-permissions:
-  contents: read
-
-jobs:
-  alert-on-incomplete-run:
-    name: Alert on incomplete scheduled run
-    if: >-
-      github.event.workflow_run.event == 'schedule' &&
-      github.event.workflow_run.conclusion != 'success' &&
-      github.event.workflow_run.conclusion != 'failure'
-    runs-on: ubuntu-latest
-    timeout-minutes: 10
-    permissions:
-      actions: read
-      contents: read
-      issues: write
-    steps:
-      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
-        with:
-          persist-credentials: false
-      - name: Open or update incomplete-run issue
-        uses: ./.github/actions/schedule-failure-issue
-        with:
-          github-token: ${{ secrets.GITHUB_TOKEN }}
-          workflow-name: ${{ github.event.workflow_run.name }}
-          source-run-id: ${{ github.event.workflow_run.id }}
-          source-run-attempt: ${{ github.event.workflow_run.run_attempt }}
-          source-event: ${{ github.event.workflow_run.event }}
-          source-ref-name: ${{ github.event.workflow_run.head_branch }}
-          source-sha: ${{ github.event.workflow_run.head_sha }}
@@ -901,10 +901,6 @@ pub struct Metrics {
    scanner_cycle_max_duration_millis: AtomicU64,
    scanner_cycle_max_objects: AtomicU64,
    scanner_cycle_max_directories: AtomicU64,
-    scanner_cycle_timeout_total: AtomicU64,
-    scanner_cycle_recovery_required_total: AtomicU64,
-    scanner_cycle_last_progress_age_seconds: AtomicU64,
-    scanner_leader_lease_without_progress: AtomicBool,
    scanner_bitrot_cycle_enabled: AtomicBool,
    scanner_bitrot_cycle_millis: AtomicU64,
    scanner_checkpoint: Mutex<Option<ScannerCheckpointReport>>,
@@ -1374,14 +1370,6 @@ pub struct ScannerMetricsReport {
    #[serde(default)]
    pub cycle_max_directories: u64,
    #[serde(default)]
-    pub cycle_timeout_total: u64,
-    #[serde(default)]
-    pub cycle_recovery_required_total: u64,
-    #[serde(default)]
-    pub cycle_last_progress_age: u64,
-    #[serde(default)]
-    pub leader_lease_without_progress: bool,
-    #[serde(default)]
    pub bitrot_cycle_enabled: bool,
    #[serde(default)]
    pub bitrot_cycle_seconds: f64,
@@ -1442,9 +1430,6 @@ const OTEL_SCANNER_BUCKETS_SCANNED: &str = "rustfs_scanner_buckets_scanned_total
 const OTEL_SCANNER_CYCLES: &str = "rustfs_scanner_cycles_total";
 const OTEL_SCANNER_CYCLE_DURATION_SECONDS: &str = "rustfs_scanner_cycle_duration_seconds";
 const OTEL_SCANNER_BUCKET_DRIVE_DURATION_SECONDS: &str = "rustfs_scanner_bucket_drive_duration_seconds";
-const OTEL_SCANNER_CYCLE_TIMEOUT_TOTAL: &str = "rustfs_scanner_cycle_timeout_total";
-const OTEL_SCANNER_CYCLE_LAST_PROGRESS_AGE: &str = "rustfs_scanner_cycle_last_progress_age";
-const OTEL_SCANNER_LEADER_LEASE_WITHOUT_PROGRESS: &str = "rustfs_scanner_leader_lease_without_progress";

 fn scan_cycle_result_label(result: u8) -> &'static str {
    match result {
@@ -1928,10 +1913,6 @@ impl Metrics {
            scanner_cycle_max_duration_millis: AtomicU64::new(0),
            scanner_cycle_max_objects: AtomicU64::new(0),
            scanner_cycle_max_directories: AtomicU64::new(0),
-            scanner_cycle_timeout_total: AtomicU64::new(0),
-            scanner_cycle_recovery_required_total: AtomicU64::new(0),
-            scanner_cycle_last_progress_age_seconds: AtomicU64::new(0),
-            scanner_leader_lease_without_progress: AtomicBool::new(false),
            scanner_bitrot_cycle_enabled: AtomicBool::new(false),
            scanner_bitrot_cycle_millis: AtomicU64::new(0),
            scanner_checkpoint: Mutex::new(None),
@@ -2431,29 +2412,12 @@ impl Metrics {
            .store(cycle_max_objects.unwrap_or_default(), Ordering::Relaxed);
        self.scanner_cycle_max_directories
            .store(cycle_max_directories.unwrap_or_default(), Ordering::Relaxed);
-        self.scanner_leader_lease_without_progress.store(false, Ordering::Relaxed);
-        self.scanner_cycle_last_progress_age_seconds.store(0, Ordering::Relaxed);
-        metrics::gauge!(OTEL_SCANNER_LEADER_LEASE_WITHOUT_PROGRESS).set(0.0);
-        metrics::gauge!(OTEL_SCANNER_CYCLE_LAST_PROGRESS_AGE).set(0.0);
        self.scanner_bitrot_cycle_enabled
            .store(bitrot_cycle.is_some(), Ordering::Relaxed);
        self.scanner_bitrot_cycle_millis
            .store(bitrot_cycle.map(duration_millis_saturated).unwrap_or_default(), Ordering::Relaxed);
    }

-    pub fn record_scanner_cycle_timeout(&self, recovery_required: bool, progress_age: Duration) {
-        self.scanner_cycle_timeout_total.fetch_add(1, Ordering::Relaxed);
-        if recovery_required {
-            self.scanner_cycle_recovery_required_total.fetch_add(1, Ordering::Relaxed);
-        }
-        self.scanner_cycle_last_progress_age_seconds
-            .store(progress_age.as_secs(), Ordering::Relaxed);
-        self.scanner_leader_lease_without_progress.store(true, Ordering::Relaxed);
-        metrics::counter!(OTEL_SCANNER_CYCLE_TIMEOUT_TOTAL).increment(1);
-        metrics::gauge!(OTEL_SCANNER_CYCLE_LAST_PROGRESS_AGE).set(progress_age.as_secs_f64());
-        metrics::gauge!(OTEL_SCANNER_LEADER_LEASE_WITHOUT_PROGRESS).set(1.0);
-    }
-
    pub fn record_scanner_set_scan_state(&self, concurrency_limit: Option<usize>, queued: Option<usize>, active: Option<usize>) {
        if let Some(concurrency_limit) = concurrency_limit {
            self.scanner_set_scan_concurrency_limit
@@ -3301,10 +3265,6 @@ impl Metrics {
        m.cycle_max_duration_seconds = self.scanner_cycle_max_duration_millis.load(Ordering::Relaxed) as f64 / 1000.0;
        m.cycle_max_objects = self.scanner_cycle_max_objects.load(Ordering::Relaxed);
        m.cycle_max_directories = self.scanner_cycle_max_directories.load(Ordering::Relaxed);
-        m.cycle_timeout_total = self.scanner_cycle_timeout_total.load(Ordering::Relaxed);
-        m.cycle_recovery_required_total = self.scanner_cycle_recovery_required_total.load(Ordering::Relaxed);
-        m.cycle_last_progress_age = self.scanner_cycle_last_progress_age_seconds.load(Ordering::Relaxed);
-        m.leader_lease_without_progress = self.scanner_leader_lease_without_progress.load(Ordering::Relaxed);
        m.bitrot_cycle_enabled = self.scanner_bitrot_cycle_enabled.load(Ordering::Relaxed);
        m.bitrot_cycle_seconds = self.scanner_bitrot_cycle_millis.load(Ordering::Relaxed) as f64 / 1000.0;
        m.scan_checkpoint = match self.scanner_checkpoint.lock() {
@@ -4966,20 +4926,4 @@ mod tests {
        assert!(!report.bitrot_cycle_enabled);
        assert_eq!(report.bitrot_cycle_seconds, 0.0);
    }
-
-    #[tokio::test]
-    async fn scanner_cycle_timeout_metrics_reset_for_a_new_cycle() {
-        let metrics = Metrics::new();
-        metrics.record_scanner_cycle_timeout(true, Duration::from_secs(17));
-        let timed_out = metrics.report().await;
-        assert_eq!(timed_out.cycle_timeout_total, 1);
-        assert_eq!(timed_out.cycle_last_progress_age, 17);
-        assert!(timed_out.leader_lease_without_progress);
-
-        metrics.record_scanner_cycle_config(Duration::from_secs(60), None, Some(Duration::from_secs(1)), None, None);
-        let current = metrics.report().await;
-        assert_eq!(current.cycle_timeout_total, 1);
-        assert_eq!(current.cycle_last_progress_age, 0);
-        assert!(!current.leader_lease_without_progress);
-    }
 }
@@ -84,12 +84,6 @@ Current guidance:
 - `RUSTFS_SCANNER_CYCLE_MAX_OBJECTS` (canonical)
 - `RUSTFS_SCANNER_CYCLE_MAX_DIRECTORIES` (canonical)

-Scanner cycle budget controls:
-
- When `RUSTFS_SCANNER_CYCLE_MAX_DURATION_SECS` is unset, the finite default is 1800 seconds (30 minutes), matching the scanner benchmark guidance.
- An explicit `0` preserves the compatibility behavior of an unbounded runtime budget. Object and directory budgets likewise remain unbounded when explicitly set to `0`.
- A timed-out cycle cancels cooperative scanner work, then fences its leader epoch before releasing the lease. An uncooperative I/O operation is dropped after the bounded shutdown window; its cursor is not claimed to be durable and the scanner reports `recovery-required` when the worker cannot stop cooperatively, the cycle state was not confirmed durable, or epoch fencing cannot be persisted.
-
 ## Mmap read environment aliases

 - `RUSTFS_OBJECT_MMAP_READ_ENABLE` (canonical)
@@ -143,12 +143,9 @@ pub const ENV_SCANNER_MAX_WAIT_SECS: &str = "RUSTFS_SCANNER_MAX_WAIT_SECS";
 /// Default scanner speed preset.
 pub const DEFAULT_SCANNER_SPEED: &str = "default";

-/// Default scanner cycle runtime budget when no override is configured.
-///
-/// An explicit `0` remains the compatibility escape hatch for an unbounded
-/// cycle. Keeping the unset default finite prevents a stalled scanner I/O
-/// operation from holding the leader lease forever.
-pub const DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS: u64 = 30 * 60;
+/// Default scanner cycle runtime budget.
+/// `0` keeps the existing unbounded per-cycle behavior.
+pub const DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS: u64 = 0;

 /// Default scanner per-cycle object budget.
 /// `0` keeps the existing unbounded per-cycle behavior.
@@ -585,12 +585,9 @@ impl VersionsHistogram {
    }
 }

-/// Replication statistics for a single target.
-///
-/// Renamed from `ReplicationStats`; serde field names are preserved
-/// byte-identically to maintain wire compatibility with existing snapshots.
-#[derive(Debug, Default, Clone, Serialize, Deserialize, PartialEq, Eq)]
-pub struct ReplicationTargetUsage {
+/// Replication statistics for a single target
+#[derive(Debug, Default, Clone, Serialize, Deserialize)]
+pub struct ReplicationStats {
    pub pending_size: u64,
    pub replicated_size: u64,
    pub failed_size: u64,
@@ -603,7 +600,7 @@ pub struct ReplicationTargetUsage {
    pub replicated_count: u64,
 }

-impl ReplicationTargetUsage {
+impl ReplicationStats {
    pub fn is_empty(&self) -> bool {
        let Self {
            pending_size,
@@ -639,7 +636,7 @@ impl ReplicationTargetUsage {
 /// Replication statistics for all targets
 #[derive(Debug, Default, Clone, Serialize, Deserialize)]
 pub struct ReplicationAllStats {
-    pub targets: HashMap<String, ReplicationTargetUsage>,
+    pub targets: HashMap<String, ReplicationStats>,
    pub replica_size: u64,
    pub replica_count: u64,
 }
@@ -652,7 +649,7 @@ impl ReplicationAllStats {
            targets,
        } = self;

-        *replica_size == 0 && *replica_count == 0 && targets.values().all(ReplicationTargetUsage::is_empty)
+        *replica_size == 0 && *replica_count == 0 && targets.values().all(ReplicationStats::is_empty)
    }

    #[deprecated(note = "use is_empty instead")]
@@ -2469,7 +2466,7 @@ mod tests {

    #[test]
    fn replication_stats_empty_checks_every_field() {
-        type SetField = fn(&mut ReplicationTargetUsage);
+        type SetField = fn(&mut ReplicationStats);

        let cases: [(&str, SetField); 10] = [
            ("pending_size", |stats| stats.pending_size = 1),
@@ -2484,9 +2481,9 @@ mod tests {
            ("replicated_count", |stats| stats.replicated_count = 1),
        ];

-        assert!(ReplicationTargetUsage::default().is_empty());
+        assert!(ReplicationStats::default().is_empty());
        for (field, set_nonzero) in cases {
-            let mut stats = ReplicationTargetUsage::default();
+            let mut stats = ReplicationStats::default();
            set_nonzero(&mut stats);
            assert!(!stats.is_empty(), "{field} must make replication stats non-empty");
        }
@@ -2517,17 +2514,17 @@ mod tests {
        }

        let empty_targets = ReplicationAllStats {
-            targets: HashMap::from([("arn:test:empty".to_string(), ReplicationTargetUsage::default())]),
+            targets: HashMap::from([("arn:test:empty".to_string(), ReplicationStats::default())]),
            ..Default::default()
        };
        assert!(empty_targets.is_empty(), "all-empty targets must keep aggregate stats empty");

        let stats = ReplicationAllStats {
            targets: HashMap::from([
-                ("arn:test:empty".to_string(), ReplicationTargetUsage::default()),
+                ("arn:test:empty".to_string(), ReplicationStats::default()),
                (
                    "arn:test:non-empty".to_string(),
-                    ReplicationTargetUsage {
+                    ReplicationStats {
                        pending_count: 1,
                        ..Default::default()
                    },
@@ -2568,7 +2565,7 @@ mod tests {
                replication_stats: Some(ReplicationAllStats {
                    targets: HashMap::from([(
                        "arn:test:pending".to_string(),
-                        ReplicationTargetUsage {
+                        ReplicationStats {
                            pending_count: 1,
                            ..Default::default()
                        },
@@ -2717,7 +2714,7 @@ mod tests {
                targets: HashMap::from([
                    (
                        "arn:self-only".to_string(),
-                        ReplicationTargetUsage {
+                        ReplicationStats {
                            pending_size: 7,
                            pending_count: 1,
                            ..Default::default()
@@ -2725,7 +2722,7 @@ mod tests {
                    ),
                    (
                        "arn:shared".to_string(),
-                        ReplicationTargetUsage {
+                        ReplicationStats {
                            failed_size: 3,
                            failed_count: 1,
                            missed_threshold_size: 2,
@@ -2744,7 +2741,7 @@ mod tests {
                targets: HashMap::from([
                    (
                        "arn:shared".to_string(),
-                        ReplicationTargetUsage {
+                        ReplicationStats {
                            failed_size: 5,
                            failed_count: 2,
                            after_threshold_size: 4,
@@ -2754,7 +2751,7 @@ mod tests {
                    ),
                    (
                        "arn:other-only".to_string(),
-                        ReplicationTargetUsage {
+                        ReplicationStats {
                            replicated_size: 11,
                            replicated_count: 3,
                            ..Default::default()
@@ -2996,9 +2993,7 @@ mod tests {
    fn replication_target_deserialization_preserves_large_historical_maps() {
        let mut stats = ReplicationAllStats::default();
        for index in 0..=1024 {
-            stats
-                .targets
-                .insert(format!("target-{index}"), ReplicationTargetUsage::default());
+            stats.targets.insert(format!("target-{index}"), ReplicationStats::default());
        }
        let encoded = rmp_serde::to_vec_named(&stats).expect("large replication target fixture should encode");
        let decoded = rmp_serde::from_slice::<ReplicationAllStats>(&encoded)
@@ -3007,47 +3002,6 @@ mod tests {
        assert_eq!(decoded.targets.len(), stats.targets.len());
    }

-    /// Round-trip test: encoding a [`ReplicationTargetUsage`] and decoding it back
-    /// must produce the exact same value.  This guards against accidental serde
-    /// field-name drift during the `ReplicationStats` -> `ReplicationTargetUsage`
-    /// rename.  Wire-level field names are the serialized Rust field identifiers,
-    /// which must remain byte-identical.
-    #[test]
-    fn replication_target_usage_rmp_round_trip() {
-        let original = ReplicationTargetUsage {
-            pending_size: 100,
-            replicated_size: 2_000,
-            failed_size: 50,
-            failed_count: 3,
-            pending_count: 7,
-            missed_threshold_size: 11,
-            after_threshold_size: 22,
-            missed_threshold_count: 1,
-            after_threshold_count: 2,
-            replicated_count: 99,
-        };
-
-        let buf = rmp_serde::to_vec_named(&original).expect("encode ReplicationTargetUsage to msgpack");
-        let decoded: ReplicationTargetUsage = rmp_serde::from_slice(&buf).expect("decode ReplicationTargetUsage from msgpack");
-        assert_eq!(original, decoded, "round-trip through rmp must preserve every field");
-
-        // Also verify that encoding as an unnamed sequence and then decoding
-        // with named fields produces the correct mapping (this catches reordering).
-        let named_buf = rmp_serde::to_vec_named(&original).expect("re-encode for field-name pinning");
-        // Spot-check that known field names appear in the named encoding.
-        let named_str = String::from_utf8_lossy(&named_buf);
-        assert!(named_str.contains("pending_size"), "field 'pending_size' must survive the rename");
-        assert!(named_str.contains("replicated_size"), "field 'replicated_size' must survive the rename");
-        assert!(
-            named_str.contains("missed_threshold_size"),
-            "field 'missed_threshold_size' must survive the rename"
-        );
-        assert!(
-            named_str.contains("after_threshold_count"),
-            "field 'after_threshold_count' must survive the rename"
-        );
-    }
-
    #[test]
    fn checked_merge_rejects_noncanonical_histograms_without_mutation() {
        let mut entry = DataUsageEntry {
@@ -15,12 +15,14 @@
 use crate::common::RustFSTestClusterEnvironment;
 use aws_sdk_s3::Client;
 use aws_sdk_s3::error::SdkError;
+use aws_sdk_s3::types::{CorsConfiguration, CorsRule};
 use bytes::Bytes;
 use std::sync::Arc;
 use tokio::sync::Barrier;
 use tracing::{info, warn};

 const BUCKET: &str = "conditional-put-race-bucket";
+const BUCKET_METADATA_RELOAD_BUCKET: &str = "bucket-metadata-reload-barrier";

 async fn cleanup_object(client: &Client, key: &str) {
    if let Err(e) = client.delete_object().bucket(BUCKET).key(key).send().await {
@@ -28,6 +30,16 @@ async fn cleanup_object(client: &Client, key: &str) {
    }
 }

+async fn assert_bucket_cors_missing(client: &Client) {
+    let result = client.get_bucket_cors().bucket(BUCKET_METADATA_RELOAD_BUCKET).send().await;
+    match result {
+        Err(SdkError::ServiceError(error)) => {
+            assert_eq!(error.err().meta().code(), Some("NoSuchCORSConfiguration"));
+        }
+        result => panic!("expected the peer to report a missing CORS configuration: {result:?}"),
+    }
+}
+
 async fn conditional_put(
    client: &Client,
    key: &str,
@@ -236,3 +248,48 @@ async fn test_conditional_put_basic_cluster() -> Result<(), Box<dyn std::error::
    cleanup_object(&client, test_key).await;
    Ok(())
 }
+
+#[tokio::test]
+async fn test_bucket_cors_write_is_visible_on_peer_before_response() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    crate::common::init_logging();
+
+    let mut cluster = RustFSTestClusterEnvironment::new(2).await?;
+    cluster.start().await?;
+    cluster.create_test_bucket(BUCKET_METADATA_RELOAD_BUCKET).await?;
+
+    let writer = cluster.create_s3_client(0)?;
+    let reader = cluster.create_s3_client(1)?;
+    assert_bucket_cors_missing(&reader).await;
+
+    let rule = CorsRule::builder()
+        .allowed_methods("GET")
+        .allowed_origins("https://example.com")
+        .build()?;
+    let configuration = CorsConfiguration::builder().cors_rules(rule).build()?;
+
+    writer
+        .put_bucket_cors()
+        .bucket(BUCKET_METADATA_RELOAD_BUCKET)
+        .cors_configuration(configuration)
+        .send()
+        .await?;
+
+    let response = reader.get_bucket_cors().bucket(BUCKET_METADATA_RELOAD_BUCKET).send().await?;
+    let rules = response.cors_rules();
+    assert_eq!(
+        rules.len(),
+        1,
+        "peer should observe the committed CORS rule before the write response returns"
+    );
+    assert_eq!(rules[0].allowed_methods(), ["GET"]);
+    assert_eq!(rules[0].allowed_origins(), ["https://example.com"]);
+
+    writer
+        .delete_bucket_cors()
+        .bucket(BUCKET_METADATA_RELOAD_BUCKET)
+        .send()
+        .await?;
+    assert_bucket_cors_missing(&reader).await;
+    writer.delete_bucket().bucket(BUCKET_METADATA_RELOAD_BUCKET).send().await?;
+    Ok(())
+}
@@ -380,24 +380,10 @@ mod tests {
        cluster.start_node(1).await?;

        let status_url = format!("{}/rustfs/admin/v3/background-heal/status", cluster.nodes[0].url);
-        let mut recovered = serde_json::Value::Null;
-        for _ in 0..60 {
-            let status_body = signed_admin_post(&status_url, None, &cluster.access_key, &cluster.secret_key).await?;
-            assert!(
-                !status_body.contains("MissingContentLength"),
-                "background heal status should not fail without an explicit Content-Length: {status_body}"
-            );
-            recovered = serde_json::from_str(&status_body)
-                .map_err(|err| format!("background heal status is not JSON ({err}): {status_body}"))?;
-            if recovered["clusterStatusComplete"] == serde_json::Value::Bool(true) {
-                break;
-            }
-            sleep(Duration::from_secs(1)).await;
-        }
-        assert_eq!(
-            recovered["clusterStatusComplete"],
-            serde_json::Value::Bool(true),
-            "cluster heal status should recover before root heal starts: {recovered}"
+        let status_body = signed_admin_post(&status_url, None, &cluster.access_key, &cluster.secret_key).await?;
+        assert!(
+            !status_body.contains("MissingContentLength"),
+            "background heal status should not fail without an explicit Content-Length: {status_body}"
        );

        let heal_body = r#"{"recursive":true,"dryRun":false,"remove":false,"recreate":true,"scanMode":2,"updateParity":false,"nolock":false}"#;
@@ -37,7 +37,7 @@ async fn test_bucket_default_sse_s3_put_object() -> Result<(), Box<dyn std::erro

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -159,7 +159,7 @@ async fn test_bucket_default_sse_kms_put_object() -> Result<(), Box<dyn std::err

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -278,7 +278,7 @@ async fn test_bucket_default_sse_kms_multipart_crc32() -> Result<(), Box<dyn std

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -475,7 +475,7 @@ async fn test_explicit_encryption_overrides_bucket_default() -> Result<(), Box<d

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -570,7 +570,7 @@ async fn test_sse_kms_without_key_id_populates_default() -> Result<(), Box<dyn s

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -189,121 +189,34 @@ pub async fn wait_for_kms_ready(
    access_key: &str,
    secret_key: &str,
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
-    wait_for_kms_ready_with_timeout(base_url, access_key, secret_key, Duration::from_secs(5)).await
-}
-
-async fn wait_for_kms_ready_with_timeout(
-    base_url: &str,
-    access_key: &str,
-    secret_key: &str,
-    total_deadline: Duration,
-) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let total_deadline = Duration::from_secs(5);
    let start = tokio::time::Instant::now();
-    let deadline = start + total_deadline;
    let mut backoff = Duration::from_millis(200);
    let max_backoff = Duration::from_secs(1);
+    let mut first_attempt = true;

    loop {
-        match tokio::time::timeout_at(deadline, get_kms_status(base_url, access_key, secret_key)).await {
-            Ok(Ok(status)) => {
-                let backend_status = serde_json::from_str::<serde_json::Value>(&status)
-                    .ok()
-                    .and_then(|value| value.get("backend_status")?.as_str().map(str::to_owned));
-                if backend_status.as_deref() == Some("healthy") {
-                    info!("KMS is ready (status: {})", status);
-                    return Ok(());
-                }
-                warn!(
-                    backend_status = backend_status.as_deref().unwrap_or("missing"),
-                    elapsed_ms = u64::try_from(start.elapsed().as_millis()).unwrap_or(u64::MAX),
-                    "KMS not ready yet, retrying…"
-                );
+        if !first_attempt {
+            if start.elapsed() >= total_deadline {
+                return Err("KMS failed to become ready within 5 seconds".into());
            }
-            Ok(Err(e)) => {
-                let elapsed_ms = u64::try_from(start.elapsed().as_millis()).unwrap_or(u64::MAX);
-                warn!(error = %e, elapsed_ms, "KMS not ready yet, retrying…");
-            }
-            Err(_) => return Err(format!("KMS failed to become ready within {} ms", total_deadline.as_millis()).into()),
+            sleep(backoff).await;
+            backoff = (backoff * 2).min(max_backoff);
        }
+        first_attempt = false;

-        let now = tokio::time::Instant::now();
-        if now >= deadline {
-            return Err(format!("KMS failed to become ready within {} ms", total_deadline.as_millis()).into());
-        }
-        sleep((now + backoff).min(deadline) - now).await;
-        backoff = (backoff * 2).min(max_backoff);
-    }
-}
-
-#[cfg(test)]
-mod readiness_tests {
-    use super::{wait_for_kms_ready, wait_for_kms_ready_with_timeout};
-    use std::sync::{
-        Arc,
-        atomic::{AtomicUsize, Ordering},
-    };
-    use std::time::Duration;
-    use tokio::io::{AsyncReadExt, AsyncWriteExt};
-    use tokio::net::TcpListener;
-
-    #[tokio::test]
-    async fn kms_readiness_retries_http_success_until_backend_is_healthy() {
-        let listener = TcpListener::bind("127.0.0.1:0").await.expect("bind readiness test server");
-        let address = listener.local_addr().expect("read readiness test server address");
-        let requests = Arc::new(AtomicUsize::new(0));
-        let server_requests = Arc::clone(&requests);
-        let server = tokio::spawn(async move {
-            for backend_status in ["error", "healthy"] {
-                let (mut socket, _) = listener.accept().await.expect("accept readiness request");
-                let mut request = Vec::new();
-                let mut chunk = [0_u8; 1024];
-                while !request.windows(4).any(|window| window == b"\r\n\r\n") {
-                    let read = socket.read(&mut chunk).await.expect("read readiness request");
-                    if read == 0 {
-                        break;
-                    }
-                    request.extend_from_slice(&chunk[..read]);
-                }
-                server_requests.fetch_add(1, Ordering::SeqCst);
-
-                let body = format!(r#"{{"backend_status":"{backend_status}"}}"#);
-                let response = format!(
-                    "HTTP/1.1 200 OK\r\nContent-Type: application/json\r\nContent-Length: {}\r\nConnection: close\r\n\r\n{body}",
-                    body.len()
-                );
-                socket.write_all(response.as_bytes()).await.expect("write readiness response");
+        match get_kms_status(base_url, access_key, secret_key).await {
+            Ok(status) => {
+                info!("KMS is ready (status: {})", status);
+                return Ok(());
            }
-        });
-
-        wait_for_kms_ready(&format!("http://{address}"), "access-key", "secret-key")
-            .await
-            .expect("KMS should become ready after the healthy response");
-
-        let observed_requests = requests.load(Ordering::SeqCst);
-        server.abort();
-        assert_eq!(observed_requests, 2, "an HTTP 200 unhealthy status must be retried");
-    }
-
-    #[tokio::test]
-    async fn kms_readiness_deadline_covers_a_stalled_status_request() {
-        let listener = TcpListener::bind("127.0.0.1:0").await.expect("bind readiness test server");
-        let address = listener.local_addr().expect("read readiness test server address");
-        let server = tokio::spawn(async move {
-            let (mut socket, _) = listener.accept().await.expect("accept readiness request");
-            let mut request = [0_u8; 1024];
-            let _ = socket.read(&mut request).await.expect("read readiness request");
-            std::future::pending::<()>().await;
-        });
-
-        let result = tokio::time::timeout(
-            Duration::from_secs(1),
-            wait_for_kms_ready_with_timeout(&format!("http://{address}"), "access-key", "secret-key", Duration::from_millis(50)),
-        )
-        .await
-        .expect("readiness helper must enforce its own deadline");
-
-        server.abort();
-        assert!(result.is_err(), "a stalled status request must not outlive the readiness deadline");
+            Err(e) => {
+                if start.elapsed() >= total_deadline {
+                    return Err(format!("KMS did not become ready within 5 s: last error: {e}").into());
+                }
+                warn!(error = %e, elapsed_ms = start.elapsed().as_millis() as u64, "KMS not ready yet, retrying…");
+            }
+        }
    }
 }

@@ -61,7 +61,7 @@ async fn test_metadata_replace_self_copy_of_sse_object_stays_decryptable() {
        )
        .await
        .expect("failed to start RustFS with local KMS");
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let client = kms_env.base_env.create_s3_client();
    // Deliberately an UNVERSIONED bucket: that is the branch where the store layer can service
@@ -160,7 +160,7 @@ async fn test_metadata_replace_self_copy_dropping_sse_rewrites_plaintext() {
        )
        .await
        .expect("failed to start RustFS with local KMS");
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let client = kms_env.base_env.create_s3_client();
    // Unversioned, and deliberately WITHOUT a bucket default-encryption rule, so the copy below
@@ -256,7 +256,7 @@ async fn test_metadata_replace_self_copy_under_bucket_default_sse_stays_decrypta
        )
        .await
        .expect("failed to start RustFS with local KMS");
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let client = kms_env.base_env.create_s3_client();
    let bucket = "copy-object-self-copy-bucket-default-sse-test";
@@ -56,7 +56,7 @@ async fn test_self_copy_of_historical_sse_s3_version_is_readable() {
        )
        .await
        .expect("failed to start RustFS with local KMS");
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let client = kms_env.base_env.create_s3_client();
    let bucket = "copy-object-version-restore-sse-test";
@@ -87,7 +87,7 @@ async fn test_head_reports_managed_metadata_for_sse_s3() -> Result<(), Box<dyn s

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -147,7 +147,7 @@ async fn test_head_reports_managed_metadata_for_sse_kms_and_copy() -> Result<(),

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -250,7 +250,7 @@ async fn test_multipart_upload_writes_encrypted_data() -> Result<(), Box<dyn std

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -24,6 +24,7 @@ use super::common::{
    test_sse_kms_encryption, test_sse_s3_encryption,
 };
 use crate::common::{TEST_BUCKET, init_logging};
+use tokio::time::{Duration, sleep};
 use tracing::info;

 /// Comprehensive test: Full KMS workflow with all encryption types
@@ -34,7 +35,7 @@ async fn test_comprehensive_kms_full_workflow() -> Result<(), Box<dyn std::error

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -102,7 +103,7 @@ async fn test_comprehensive_stress_test() -> Result<(), Box<dyn std::error::Erro

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -136,7 +137,7 @@ async fn test_comprehensive_key_isolation() -> Result<(), Box<dyn std::error::Er

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -207,7 +208,7 @@ async fn test_comprehensive_concurrent_operations() -> Result<(), Box<dyn std::e

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -252,7 +253,7 @@ async fn test_comprehensive_performance_benchmark() -> Result<(), Box<dyn std::e

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -44,7 +44,7 @@ async fn test_kms_zero_byte_file_encryption() -> Result<(), Box<dyn std::error::

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -117,7 +117,7 @@ async fn test_kms_single_byte_file_encryption() -> Result<(), Box<dyn std::error

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -209,7 +209,7 @@ async fn test_kms_multipart_boundary_conditions() -> Result<(), Box<dyn std::err

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -284,7 +284,7 @@ async fn test_kms_invalid_key_scenarios() -> Result<(), Box<dyn std::error::Erro

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -371,7 +371,7 @@ async fn test_kms_concurrent_encryption() -> Result<(), Box<dyn std::error::Erro

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = Arc::new(kms_env.base_env.create_s3_client());
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -478,7 +478,7 @@ async fn test_kms_key_validation_security() -> Result<(), Box<dyn std::error::Er

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -37,7 +37,7 @@ async fn test_kms_key_directory_unavailable() -> Result<(), Box<dyn std::error::

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -127,7 +127,7 @@ async fn test_kms_corrupted_key_files() -> Result<(), Box<dyn std::error::Error

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -218,7 +218,7 @@ async fn test_kms_multipart_upload_interruption() -> Result<(), Box<dyn std::err

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -401,7 +401,7 @@ async fn test_kms_resource_constraints() -> Result<(), Box<dyn std::error::Error

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -46,7 +46,7 @@ async fn test_local_kms_end_to_end() -> Result<(), Box<dyn std::error::Error + S
        .expect("Failed to start RustFS with Local KMS");

    // Wait a moment for RustFS to fully start up and initialize KMS
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    info!("RustFS started with KMS auto-configuration, default_key_id: {}", default_key_id);

@@ -127,7 +127,7 @@ async fn test_local_kms_key_isolation() {
        .expect("Failed to start RustFS with Local KMS");

    // Wait a moment for RustFS to fully start up and initialize KMS
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    info!("RustFS started with KMS auto-configuration, default_key_id: {}", default_key_id);

@@ -227,7 +227,7 @@ async fn test_local_kms_large_file() {
        .expect("Failed to start RustFS with Local KMS");

    // Wait a moment for RustFS to fully start up and initialize KMS
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    info!("RustFS started with KMS auto-configuration, default_key_id: {}", default_key_id);

@@ -309,7 +309,7 @@ async fn test_local_kms_multipart_upload() {
        .expect("Failed to start RustFS with Local KMS");

    // Wait for KMS initialization
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    info!("RustFS started with KMS auto-configuration, default_key_id: {}", default_key_id);

@@ -19,6 +19,7 @@
 //! multipart upload behaviour.

 use crate::common::{TEST_BUCKET, init_logging};
+use tokio::time::{Duration, sleep};
 use tracing::{error, info};

 use super::common::{
@@ -44,8 +45,8 @@ impl VaultKmsTestContext {

        start_kms(&env.base_env.url, &env.base_env.access_key, &env.base_env.secret_key).await?;

-        // Wait for KMS to finish initialising.
-        super::common::wait_for_kms_ready(&env.base_env.url, &env.base_env.access_key, &env.base_env.secret_key).await?;
+        // Allow Vault to finish initialising token auth and transit engine.
+        sleep(Duration::from_secs(2)).await;

        Ok(Self { env })
    }
@@ -39,6 +39,9 @@ mod kms_edge_cases_test;
 #[cfg(test)]
 mod kms_fault_recovery_test;

+#[cfg(test)]
+mod test_runner;
+
 #[cfg(test)]
 mod bucket_default_encryption_test;

@@ -33,7 +33,7 @@ async fn test_step1_basic_single_file_encryption() -> Result<(), Box<dyn std::er

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -89,7 +89,7 @@ async fn test_step2_basic_multipart_upload_without_encryption() -> Result<(), Bo

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -187,7 +187,7 @@ async fn test_step3_multipart_upload_with_sse_s3() -> Result<(), Box<dyn std::er

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -310,7 +310,7 @@ async fn test_step4_large_multipart_upload_with_encryption() -> Result<(), Box<d

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -435,7 +435,7 @@ async fn test_step5_all_encryption_types_multipart() -> Result<(), Box<dyn std::

    let mut kms_env = LocalKMSTestEnvironment::new().await?;
    let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;

    let s3_client = kms_env.base_env.create_s3_client();
    kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -0,0 +1,499 @@
+// 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
+//
+#![allow(dead_code)]
+// 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.
+
+//! Unified KMS test suite runner
+//!
+//! This module provides a unified interface for running KMS tests with categorization,
+//! filtering, and comprehensive reporting capabilities.
+
+use crate::common::init_logging;
+use std::time::Instant;
+use tokio::time::{Duration, sleep};
+use tracing::{debug, error, info, warn};
+
+/// Test category for organization and filtering
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+pub enum TestCategory {
+    CoreFunctionality,
+    MultipartEncryption,
+    EdgeCases,
+    FaultRecovery,
+    Comprehensive,
+    Performance,
+}
+
+impl TestCategory {
+    pub fn as_str(&self) -> &'static str {
+        match self {
+            TestCategory::CoreFunctionality => "core-functionality",
+            TestCategory::MultipartEncryption => "multipart-encryption",
+            TestCategory::EdgeCases => "edge-cases",
+            TestCategory::FaultRecovery => "fault-recovery",
+            TestCategory::Comprehensive => "comprehensive",
+            TestCategory::Performance => "performance",
+        }
+    }
+}
+
+/// Test definition with metadata
+#[derive(Debug, Clone)]
+pub struct TestDefinition {
+    pub name: String,
+    pub description: String,
+    pub category: TestCategory,
+    pub estimated_duration: Duration,
+    pub is_critical: bool,
+}
+
+impl TestDefinition {
+    pub fn new(
+        name: impl Into<String>,
+        description: impl Into<String>,
+        category: TestCategory,
+        estimated_duration: Duration,
+        is_critical: bool,
+    ) -> Self {
+        Self {
+            name: name.into(),
+            description: description.into(),
+            category,
+            estimated_duration,
+            is_critical,
+        }
+    }
+}
+
+/// Test execution result
+#[derive(Debug, Clone)]
+pub struct TestResult {
+    pub test_name: String,
+    pub category: TestCategory,
+    pub success: bool,
+    pub duration: Duration,
+    pub error_message: Option<String>,
+}
+
+impl TestResult {
+    pub fn success(test_name: String, category: TestCategory, duration: Duration) -> Self {
+        Self {
+            test_name,
+            category,
+            success: true,
+            duration,
+            error_message: None,
+        }
+    }
+
+    pub fn failure(test_name: String, category: TestCategory, duration: Duration, error: String) -> Self {
+        Self {
+            test_name,
+            category,
+            success: false,
+            duration,
+            error_message: Some(error),
+        }
+    }
+}
+
+/// Comprehensive test suite configuration
+#[derive(Debug, Clone)]
+pub struct TestSuiteConfig {
+    pub categories: Vec<TestCategory>,
+    pub include_critical_only: bool,
+    pub max_duration: Option<Duration>,
+    pub parallel_execution: bool,
+}
+
+impl Default for TestSuiteConfig {
+    fn default() -> Self {
+        Self {
+            categories: vec![
+                TestCategory::CoreFunctionality,
+                TestCategory::MultipartEncryption,
+                TestCategory::EdgeCases,
+                TestCategory::FaultRecovery,
+                TestCategory::Comprehensive,
+            ],
+            include_critical_only: false,
+            max_duration: None,
+            parallel_execution: false,
+        }
+    }
+}
+
+/// Unified KMS test suite runner
+pub struct KMSTestSuite {
+    tests: Vec<TestDefinition>,
+    config: TestSuiteConfig,
+}
+
+impl KMSTestSuite {
+    /// Create a new test suite with default configuration
+    pub fn new() -> Self {
+        let tests = vec![
+            // Core Functionality Tests
+            TestDefinition::new(
+                "test_local_kms_end_to_end",
+                "End-to-end KMS test with all encryption types",
+                TestCategory::CoreFunctionality,
+                Duration::from_secs(60),
+                true,
+            ),
+            TestDefinition::new(
+                "test_local_kms_key_isolation",
+                "Test KMS key isolation and security",
+                TestCategory::CoreFunctionality,
+                Duration::from_secs(45),
+                true,
+            ),
+            // Multipart Encryption Tests
+            TestDefinition::new(
+                "test_local_kms_multipart_upload",
+                "Test large file multipart upload with encryption",
+                TestCategory::MultipartEncryption,
+                Duration::from_secs(120),
+                true,
+            ),
+            TestDefinition::new(
+                "test_step1_basic_single_file_encryption",
+                "Basic single file encryption test",
+                TestCategory::MultipartEncryption,
+                Duration::from_secs(30),
+                false,
+            ),
+            TestDefinition::new(
+                "test_step2_basic_multipart_upload_without_encryption",
+                "Basic multipart upload without encryption",
+                TestCategory::MultipartEncryption,
+                Duration::from_secs(45),
+                false,
+            ),
+            TestDefinition::new(
+                "test_step3_multipart_upload_with_sse_s3",
+                "Multipart upload with SSE-S3 encryption",
+                TestCategory::MultipartEncryption,
+                Duration::from_secs(60),
+                true,
+            ),
+            TestDefinition::new(
+                "test_step4_large_multipart_upload_with_encryption",
+                "Large file multipart upload with encryption",
+                TestCategory::MultipartEncryption,
+                Duration::from_secs(90),
+                false,
+            ),
+            TestDefinition::new(
+                "test_step5_all_encryption_types_multipart",
+                "All encryption types multipart test",
+                TestCategory::MultipartEncryption,
+                Duration::from_secs(120),
+                true,
+            ),
+            // Edge Cases Tests
+            TestDefinition::new(
+                "test_kms_zero_byte_file_encryption",
+                "Test encryption of zero-byte files",
+                TestCategory::EdgeCases,
+                Duration::from_secs(20),
+                false,
+            ),
+            TestDefinition::new(
+                "test_kms_single_byte_file_encryption",
+                "Test encryption of single-byte files",
+                TestCategory::EdgeCases,
+                Duration::from_secs(20),
+                false,
+            ),
+            TestDefinition::new(
+                "test_kms_multipart_boundary_conditions",
+                "Test multipart upload boundary conditions",
+                TestCategory::EdgeCases,
+                Duration::from_secs(45),
+                false,
+            ),
+            TestDefinition::new(
+                "test_kms_invalid_key_scenarios",
+                "Test invalid key scenarios",
+                TestCategory::EdgeCases,
+                Duration::from_secs(30),
+                false,
+            ),
+            TestDefinition::new(
+                "test_kms_concurrent_encryption",
+                "Test concurrent encryption operations",
+                TestCategory::EdgeCases,
+                Duration::from_secs(60),
+                false,
+            ),
+            TestDefinition::new(
+                "test_kms_key_validation_security",
+                "Test key validation security",
+                TestCategory::EdgeCases,
+                Duration::from_secs(30),
+                false,
+            ),
+            // Fault Recovery Tests
+            TestDefinition::new(
+                "test_kms_key_directory_unavailable",
+                "Test KMS when key directory is unavailable",
+                TestCategory::FaultRecovery,
+                Duration::from_secs(45),
+                false,
+            ),
+            TestDefinition::new(
+                "test_kms_corrupted_key_files",
+                "Test KMS with corrupted key files",
+                TestCategory::FaultRecovery,
+                Duration::from_secs(30),
+                false,
+            ),
+            TestDefinition::new(
+                "test_kms_multipart_upload_interruption",
+                "Test multipart upload interruption recovery",
+                TestCategory::FaultRecovery,
+                Duration::from_secs(60),
+                false,
+            ),
+            TestDefinition::new(
+                "test_kms_resource_constraints",
+                "Test KMS under resource constraints",
+                TestCategory::FaultRecovery,
+                Duration::from_secs(90),
+                false,
+            ),
+            // Comprehensive Tests
+            TestDefinition::new(
+                "test_comprehensive_kms_full_workflow",
+                "Full KMS workflow comprehensive test",
+                TestCategory::Comprehensive,
+                Duration::from_secs(300),
+                true,
+            ),
+            TestDefinition::new(
+                "test_comprehensive_stress_test",
+                "KMS stress test with large datasets",
+                TestCategory::Comprehensive,
+                Duration::from_secs(400),
+                false,
+            ),
+            TestDefinition::new(
+                "test_comprehensive_key_isolation",
+                "Comprehensive key isolation test",
+                TestCategory::Comprehensive,
+                Duration::from_secs(180),
+                false,
+            ),
+            TestDefinition::new(
+                "test_comprehensive_concurrent_operations",
+                "Comprehensive concurrent operations test",
+                TestCategory::Comprehensive,
+                Duration::from_secs(240),
+                false,
+            ),
+            TestDefinition::new(
+                "test_comprehensive_performance_benchmark",
+                "KMS performance benchmark test",
+                TestCategory::Comprehensive,
+                Duration::from_secs(360),
+                false,
+            ),
+        ];
+
+        Self {
+            tests,
+            config: TestSuiteConfig::default(),
+        }
+    }
+
+    /// Configure the test suite
+    pub fn with_config(mut self, config: TestSuiteConfig) -> Self {
+        self.config = config;
+        self
+    }
+
+    /// Filter tests based on category
+    pub fn filter_by_category(&self, category: &TestCategory) -> Vec<&TestDefinition> {
+        self.tests.iter().filter(|test| &test.category == category).collect()
+    }
+
+    /// Filter tests based on criticality
+    pub fn filter_critical_tests(&self) -> Vec<&TestDefinition> {
+        self.tests.iter().filter(|test| test.is_critical).collect()
+    }
+
+    /// Get test summary by category
+    pub fn get_category_summary(&self) -> std::collections::HashMap<TestCategory, Vec<&TestDefinition>> {
+        let mut summary = std::collections::HashMap::new();
+        for test in &self.tests {
+            summary.entry(test.category.clone()).or_insert_with(Vec::new).push(test);
+        }
+        summary
+    }
+
+    /// Run the complete test suite
+    pub async fn run_test_suite(&self) -> Vec<TestResult> {
+        init_logging();
+        info!("🚀 Starting unified KMS test suite");
+
+        let start_time = Instant::now();
+        let mut results = Vec::new();
+
+        // Filter tests based on configuration
+        let tests_to_run: Vec<&TestDefinition> = self
+            .tests
+            .iter()
+            .filter(|test| self.config.categories.contains(&test.category))
+            .filter(|test| !self.config.include_critical_only || test.is_critical)
+            .collect();
+
+        info!("📊 Test plan: {} test(s) scheduled", tests_to_run.len());
+        for (i, test) in tests_to_run.iter().enumerate() {
+            info!("  {}. {} ({})", i + 1, test.name, test.category.as_str());
+        }
+
+        // Execute tests
+        for (i, test_def) in tests_to_run.iter().enumerate() {
+            info!("🧪 Running test {}/{}: {}", i + 1, tests_to_run.len(), test_def.name);
+            info!("   📝 Description: {}", test_def.description);
+            info!("   🏷️  Category: {}", test_def.category.as_str());
+            info!("   ⏱️  Estimated duration: {:?}", test_def.estimated_duration);
+
+            let test_start = Instant::now();
+            let result = self.run_single_test(test_def).await;
+            let test_duration = test_start.elapsed();
+
+            match result {
+                Ok(_) => {
+                    info!("✅ Test passed: {} ({:.2}s)", test_def.name, test_duration.as_secs_f64());
+                    results.push(TestResult::success(test_def.name.clone(), test_def.category.clone(), test_duration));
+                }
+                Err(e) => {
+                    error!("❌ Test failed: {} ({:.2}s): {}", test_def.name, test_duration.as_secs_f64(), e);
+                    results.push(TestResult::failure(
+                        test_def.name.clone(),
+                        test_def.category.clone(),
+                        test_duration,
+                        e.to_string(),
+                    ));
+                }
+            }
+
+            // Add delay between tests to avoid resource conflicts
+            if i < tests_to_run.len() - 1 {
+                debug!("⏸️  Waiting two seconds before the next test...");
+                sleep(Duration::from_secs(2)).await;
+            }
+        }
+
+        let total_duration = start_time.elapsed();
+        self.print_test_summary(&results, total_duration);
+
+        results
+    }
+
+    /// Run a single test by dispatching to the appropriate test function
+    async fn run_single_test(&self, test_def: &TestDefinition) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+        // This is a placeholder for test dispatch logic
+        // In a real implementation, this would dispatch to actual test functions
+        warn!("⚠️  Test '{}' is not implemented in the unified runner; skipping", test_def.name);
+        Ok(())
+    }
+
+    /// Print comprehensive test summary
+    fn print_test_summary(&self, results: &[TestResult], total_duration: Duration) {
+        info!("📊 KMS test suite summary");
+        info!("⏱️  Total duration: {:.2} seconds", total_duration.as_secs_f64());
+        info!("📈 Total tests: {}", results.len());
+
+        let passed = results.iter().filter(|r| r.success).count();
+        let failed = results.iter().filter(|r| !r.success).count();
+
+        info!("✅ Passed: {}", passed);
+        info!("❌ Failed: {}", failed);
+        info!("📊 Success rate: {:.1}%", (passed as f64 / results.len() as f64) * 100.0);
+
+        // Summary by category
+        let mut category_summary: std::collections::HashMap<TestCategory, (usize, usize)> = std::collections::HashMap::new();
+        for result in results {
+            let (total, passed_count) = category_summary.entry(result.category.clone()).or_insert((0, 0));
+            *total += 1;
+            if result.success {
+                *passed_count += 1;
+            }
+        }
+
+        info!("📊 Category summary:");
+        for (category, (total, passed_count)) in category_summary {
+            info!(
+                "  🏷️  {}: {}/{} ({:.1}%)",
+                category.as_str(),
+                passed_count,
+                total,
+                (passed_count as f64 / total as f64) * 100.0
+            );
+        }
+
+        // List failed tests
+        if failed > 0 {
+            warn!("❌ Failing tests:");
+            for result in results.iter().filter(|r| !r.success) {
+                warn!("  - {}: {}", result.test_name, result.error_message.as_deref().unwrap_or("Unknown error"));
+            }
+        }
+    }
+}
+
+/// Quick test suite for critical tests only
+#[tokio::test]
+async fn test_kms_critical_suite() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let config = TestSuiteConfig {
+        categories: vec![TestCategory::CoreFunctionality, TestCategory::MultipartEncryption],
+        include_critical_only: true,
+        max_duration: Some(Duration::from_secs(600)), // 10 minutes max
+        parallel_execution: false,
+    };
+
+    let suite = KMSTestSuite::new().with_config(config);
+    let results = suite.run_test_suite().await;
+
+    let failed_count = results.iter().filter(|r| !r.success).count();
+    if failed_count > 0 {
+        return Err(format!("Critical test suite failed: {failed_count} tests failed").into());
+    }
+
+    info!("✅ All critical tests passed");
+    Ok(())
+}
+
+/// Full comprehensive test suite
+#[tokio::test]
+async fn test_kms_full_suite() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let suite = KMSTestSuite::new();
+    let results = suite.run_test_suite().await;
+
+    let total_tests = results.len();
+    let failed_count = results.iter().filter(|r| !r.success).count();
+    let success_rate = ((total_tests - failed_count) as f64 / total_tests as f64) * 100.0;
+
+    info!("📊 Full suite success rate: {:.1}%", success_rate);
+
+    // Allow up to 10% failure rate for non-critical tests
+    if success_rate < 90.0 {
+        return Err(format!("Test suite success rate too low: {success_rate:.1}%").into());
+    }
+
+    info!("✅ Full test suite succeeded");
+    Ok(())
+}
@@ -16,7 +16,6 @@ use crate::common::{RustFSTestClusterEnvironment, RustFSTestEnvironment, init_lo
 use aws_sdk_s3::primitives::ByteStream;
 use http::header::{CONTENT_TYPE, HOST};
 use reqwest::StatusCode;
-use rustfs_config::{ENV_DRIVE_ACTIVE_CHECK_INTERVAL_SECS, ENV_NOTIFY_ENABLE};
 use rustfs_signer::pre_sign_v4;
 use rustfs_utils::egress::ENV_OUTBOUND_ALLOW_ORIGINS;
 use s3s::Body;
@@ -977,8 +976,7 @@ async fn test_get_object_lambda_rejects_disabled_target() -> Result<(), Box<dyn
    init_logging();

    let mut env = RustFSTestEnvironment::new().await?;
-    env.start_rustfs_server_with_env(vec![], &[(ENV_NOTIFY_ENABLE, "true")])
-        .await?;
+    env.start_rustfs_server(vec![]).await?;

    let bucket = "object-lambda-e2e-disabled-target";
    let key = "input.txt";
@@ -994,24 +992,17 @@ async fn test_get_object_lambda_rejects_disabled_target() -> Result<(), Box<dyn
        .send()
        .await?;

-    let queue_dir = format!("{}/disabled-target-queue", env.temp_dir);
-    tokio::fs::create_dir_all(&queue_dir).await?;
-    let config_url = format!("{}/rustfs/admin/v3/set-config-kv", env.url);
-    let directive = format!(
-        "notify_webhook:transformer enable=off endpoint=\"http://127.0.0.1:9/transform\" auth_token=\"secret-token\" queue_dir=\"{queue_dir}\""
-    );
-    let disable_response = signed_request(
-        http::Method::PUT,
-        &config_url,
-        &env.access_key,
-        &env.secret_key,
-        Some(directive.into_bytes()),
-        Some("text/plain"),
+    configure_webhook_target_with_key_values(
+        &env,
+        "transformer",
+        vec![
+            ("endpoint", "http://127.0.0.1:9/transform".to_string()),
+            ("auth_token", "secret-token".to_string()),
+            ("enable", "off".to_string()),
+        ],
    )
    .await?;
-    let disable_status = disable_response.status();
-    let disable_body = disable_response.text().await?;
-    assert_eq!(disable_status, StatusCode::OK, "failed to disable target: {disable_body}");
+    wait_for_target_visibility(&env, "transformer").await?;

    let lambda_url = format!("{}/{}/{}?lambdaArn={}", env.url, bucket, key, urlencoding::encode(lambda_arn));
    let response = signed_request(http::Method::GET, &lambda_url, &env.access_key, &env.secret_key, None, None).await?;
@@ -1030,8 +1021,7 @@ async fn test_configure_object_lambda_target_rejects_invalid_endpoint() -> Resul
    init_logging();

    let mut env = RustFSTestEnvironment::new().await?;
-    env.start_rustfs_server_with_env(vec![], &[(ENV_NOTIFY_ENABLE, "true")])
-        .await?;
+    env.start_rustfs_server(vec![]).await?;

    let bucket = "object-lambda-e2e-invalid-endpoint";

@@ -1074,8 +1064,7 @@ async fn test_configure_object_lambda_notify_webhook_rejects_response_header_tim
    init_logging();

    let mut env = RustFSTestEnvironment::new().await?;
-    env.start_rustfs_server_with_env(vec![], &[(ENV_NOTIFY_ENABLE, "true")])
-        .await?;
+    env.start_rustfs_server(vec![]).await?;

    let response = send_configure_webhook_target_request(
        &env,
@@ -1184,8 +1173,6 @@ async fn test_listen_notification_fans_in_remote_node_events() -> Result<(), Box
    init_logging();

    let mut cluster = RustFSTestClusterEnvironment::new(2).await?;
-    cluster.set_env(ENV_NOTIFY_ENABLE, "true");
-    cluster.set_env(ENV_DRIVE_ACTIVE_CHECK_INTERVAL_SECS, "1");
    cluster.start().await?;

    let bucket = "listen-notification-cluster";
@@ -15,6 +15,7 @@
 use crate::common::{RustFSTestClusterEnvironment, init_logging};
 use aws_sdk_s3::error::SdkError;
 use aws_sdk_s3::primitives::ByteStream;
+use aws_sdk_s3::types::CompletedMultipartUpload;
 use tokio::time::{Duration, sleep};
 use tracing::info;
 use uuid::Uuid;
@@ -42,18 +43,32 @@ async fn list_parts_reports_missing_upload(
    }
 }

-async fn multipart_listing_reports_missing_upload(
+async fn complete_reports_missing_upload(
    client: &aws_sdk_s3::Client,
    bucket: &str,
    key: &str,
    upload_id: &str,
 ) -> Result<bool, Box<dyn std::error::Error + Send + Sync>> {
-    let result = client.list_multipart_uploads().bucket(bucket).prefix(key).send().await?;
-
-    Ok(!result
-        .uploads()
-        .iter()
-        .any(|upload| upload.key() == Some(key) && upload.upload_id() == Some(upload_id)))
+    let result = client
+        .complete_multipart_upload()
+        .bucket(bucket)
+        .key(key)
+        .upload_id(upload_id)
+        .multipart_upload(CompletedMultipartUpload::builder().build())
+        .send()
+        .await;
+    match result {
+        Ok(_) => Ok(false),
+        Err(SdkError::ServiceError(err)) => {
+            let code = err.err().meta().code().unwrap_or("");
+            if code == "NoSuchUpload" {
+                Ok(true)
+            } else {
+                Err(format!("unexpected complete_multipart_upload service error: code={code}, err={err:?}").into())
+            }
+        }
+        Err(err) => Err(format!("unexpected complete_multipart_upload error: {err:?}").into()),
+    }
 }

 async fn wait_for_cleanup_on_all_nodes(
@@ -66,8 +81,8 @@ async fn wait_for_cleanup_on_all_nodes(
        let mut all_cleaned = true;
        for (idx, client) in clients.iter().enumerate() {
            let list_parts_missing = list_parts_reports_missing_upload(client, bucket, key, upload_id).await?;
-            let listing_missing = multipart_listing_reports_missing_upload(client, bucket, key, upload_id).await?;
-            if !(list_parts_missing && listing_missing) {
+            let complete_missing = complete_reports_missing_upload(client, bucket, key, upload_id).await?;
+            if !(list_parts_missing && complete_missing) {
                info!("stale multipart still visible on node {} at attempt {}", idx, attempt + 1);
                all_cleaned = false;
                break;
@@ -131,10 +146,6 @@ async fn test_stale_multipart_cleanup_removes_incomplete_upload_across_cluster()
        1,
        "multipart upload should be visible before background cleanup"
    );
-    assert!(
-        !multipart_listing_reports_missing_upload(&clients[2], CLEANUP_BUCKET, &key, &upload_id).await?,
-        "multipart upload listing should contain the upload before background cleanup"
-    );

    wait_for_cleanup_on_all_nodes(&clients, CLEANUP_BUCKET, &key, &upload_id).await?;

@@ -317,6 +317,8 @@ pub mod config {
 }

 pub mod data_usage {
+    #[cfg(feature = "test-util")]
+    pub use crate::data_usage::seed_bucket_usage_memory_for_test;
    pub use crate::data_usage::{
        DATA_USAGE_CACHE_NAME, apply_bucket_usage_memory_overlay, compute_bucket_usage,
        init_compression_total_memory_from_backend, invalidate_admin_data_usage_snapshot_cache,
@@ -328,8 +330,6 @@ pub mod data_usage {
        remove_bucket_usage_from_backend, replace_bucket_usage_memory_from_info, store_compression_total_in_backend,
        store_data_usage_in_backend,
    };
-    #[cfg(feature = "test-util")]
-    pub use crate::data_usage::{get_bucket_usage_memory, seed_bucket_usage_memory_for_test};
 }

 pub mod disk {
@@ -3425,44 +3425,6 @@ mod tests {
        assert!(mutexes.contains_key("second"));
    }

-    #[tokio::test]
-    async fn update_all_targets_publishes_disable_proxy_on_target_client() {
-        // The read-proxy selector (replication_proxy::get_proxy_targets) skips
-        // targets whose TargetClient carries disable_proxy — the persisted
-        // per-target opt-out must survive client publication.
-        let sys = BucketTargetSys::default();
-        let target = |arn: &str, disable_proxy: bool| BucketTarget {
-            arn: arn.to_string(),
-            endpoint: "192.168.1.10:9000".to_string(),
-            target_bucket: "target-bucket".to_string(),
-            region: "us-east-1".to_string(),
-            disable_proxy,
-            credentials: Some(Credentials {
-                access_key: "access".to_string(),
-                secret_key: "secret".to_string(),
-                session_token: None,
-                expiration: None,
-            }),
-            ..Default::default()
-        };
-        let targets = BucketTargets {
-            targets: vec![target("arn:proxied", false), target("arn:opted-out", true)],
-        };
-
-        sys.update_all_targets("bucket", Some(&targets)).await;
-
-        let proxied = sys
-            .get_remote_target_client("bucket", "arn:proxied")
-            .await
-            .expect("client should be published");
-        assert!(!proxied.disable_proxy);
-        let opted_out = sys
-            .get_remote_target_client("bucket", "arn:opted-out")
-            .await
-            .expect("client should be published");
-        assert!(opted_out.disable_proxy, "disable_proxy must reach the published TargetClient");
-    }
-
    #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
    async fn target_updates_serialize_client_build_through_publication_per_bucket() {
        let sys = Arc::new(BucketTargetSys::default());
@@ -2855,7 +2855,7 @@ fn replicate_object_info_from_object_info(
        .map(|v| OffsetDateTime::parse(&v, &Rfc3339).unwrap_or(OffsetDateTime::UNIX_EPOCH));
    let mut rstate = oi.replication_state();
    rstate.replicate_decision_str = dsc.to_string();
-    let asz = oi.get_actual_size_or_physical();
+    let asz = oi.get_actual_size().unwrap_or_default();
    let ssec = replication_object_is_ssec_encrypted(&oi.user_defined);
    let checksum = if ssec { oi.checksum.clone() } else { None };

@@ -1412,7 +1412,7 @@ pub async fn get_heal_replicate_object_info(oi: &ObjectInfo, rcfg: &ReplicationC
    };
    let mut replication_state = oi.replication_state();
    replication_state.replicate_decision_str = dsc.to_string();
-    let actual_size = oi.get_actual_size_or_physical();
+    let actual_size = oi.get_actual_size().unwrap_or_default();

    Ok(ReplicateObjectInfo {
        name: oi.name.clone(),
@@ -389,7 +389,7 @@ fn replication_source_object(object_info: &ObjectInfo) -> ReplicationSourceObjec
            .map(|mod_time| OffsetDateTime::from_unix_timestamp(mod_time.unix_timestamp()).unwrap_or(mod_time)),
        version_id: object_info.version_id.map(|version_id| version_id.to_string()),
        etag: object_info.etag.as_deref(),
-        actual_size: object_info.get_actual_size_or_physical(),
+        actual_size: object_info.get_actual_size().unwrap_or_default(),
        delete_marker: object_info.delete_marker,
        content_type: object_info.content_type.as_deref(),
        content_encoding: object_info.content_encoding.as_deref(),
@@ -542,20 +542,6 @@ mod tests {
        assert!(replication_target_head_is_newer_null_version(&source, &target));
    }

-    #[test]
-    fn replication_source_uses_physical_size_for_unknown_compressed_object() {
-        let mut metadata = HashMap::new();
-        rustfs_utils::http::insert_str(&mut metadata, rustfs_utils::http::SUFFIX_COMPRESSION, "zstd".to_string());
-        let source = ObjectInfo {
-            size: 128,
-            actual_size: -1,
-            user_defined: Arc::new(metadata),
-            ..Default::default()
-        };
-
-        assert_eq!(replication_source_object(&source).actual_size, 128);
-    }
-
    #[test]
    fn replication_target_head_content_matches_compare_etag_only() {
        let source = ObjectInfo {
@@ -85,6 +85,7 @@ const PEER_REST_RECOVERY_MAX_ATTEMPTS: u32 = 60;
 const PEER_REST_RECOVERY_MAX_BACKOFF: Duration = Duration::from_secs(30);
 const SCANNER_ACTIVITY_MAX_MESSAGE_SIZE: usize = 1024;
 const REPLICATION_STATS_MAX_MESSAGE_SIZE: usize = 8 * 1024 * 1024;
+const BUCKET_METADATA_RELOAD_TIMEOUT: Duration = Duration::from_secs(5);

 /// Error for a peer that reported `success = false` without an `error_info` payload.
 ///
@@ -1328,27 +1329,38 @@ impl PeerRestClient {
    }

    pub async fn load_bucket_metadata(&self, bucket: &str, scanner_maintenance_change: bool) -> Result<()> {
-        self.finalize_result(
-            async {
-                let mut client = self.get_client().await?;
-                let mut request = Request::new(LoadBucketMetadataRequest {
-                    bucket: bucket.to_string(),
-                    scanner_maintenance_change,
-                });
-                set_tonic_mutation_body_digest(&mut request)?;
-
-                let response = client.load_bucket_metadata(request).await?.into_inner();
-                if !response.success {
-                    if let Some(msg) = response.error_info {
-                        return Err(Error::other(msg));
-                    }
-                    return Err(peer_failure_without_details("load_bucket_metadata", Some(bucket)));
-                }
-                Ok(())
+        let result = tokio::time::timeout(BUCKET_METADATA_RELOAD_TIMEOUT, async {
+            let result = self.load_bucket_metadata_once(bucket, scanner_maintenance_change).await;
+            if let Err(err) = &result
+                && Self::is_network_like_error(err)
+            {
+                self.prepare_retry().await;
+                return self.load_bucket_metadata_once(bucket, scanner_maintenance_change).await;
            }
-            .await,
-        )
+            result
+        })
        .await
+        .unwrap_or_else(|_| Err(Error::other(format!("load_bucket_metadata({bucket}) timed out"))));
+        self.finalize_result(result).await
+    }
+
+    async fn load_bucket_metadata_once(&self, bucket: &str, scanner_maintenance_change: bool) -> Result<()> {
+        let mut client = self.get_client().await?;
+        let mut request = Request::new(LoadBucketMetadataRequest {
+            bucket: bucket.to_string(),
+            scanner_maintenance_change,
+        });
+        set_tonic_mutation_body_digest(&mut request)?;
+        request.set_timeout(BUCKET_METADATA_RELOAD_TIMEOUT);
+
+        let response = client.load_bucket_metadata(request).await?.into_inner();
+        if !response.success {
+            if let Some(msg) = response.error_info {
+                return Err(Error::other(msg));
+            }
+            return Err(peer_failure_without_details("load_bucket_metadata", Some(bucket)));
+        }
+        Ok(())
    }

    pub async fn delete_bucket_metadata(&self, bucket: &str) -> Result<()> {
@@ -42,9 +42,8 @@ use futures::lock::Mutex;
 use metrics::counter;
 use rustfs_filemeta::{FileInfo, ObjectPartInfo, RawFileInfo};
 use rustfs_io_metrics::internode_metrics::{
-    INTERNODE_STAGE_BATCH_READ_VERSION_REQUEST_ENCODE, INTERNODE_STAGE_BATCH_READ_VERSION_RESPONSE_DECODE,
-    INTERNODE_STAGE_BATCH_READ_VERSION_RPC_ROUNDTRIP, INTERNODE_STAGE_READ_VERSION_REQUEST_ENCODE,
-    INTERNODE_STAGE_READ_VERSION_RESPONSE_DECODE, INTERNODE_STAGE_READ_VERSION_RPC_ROUNDTRIP,
+    INTERNODE_STAGE_READ_VERSION_REQUEST_ENCODE, INTERNODE_STAGE_READ_VERSION_RESPONSE_DECODE,
+    INTERNODE_STAGE_READ_VERSION_RPC_ROUNDTRIP,
 };
 use rustfs_protos::ChannelClass;
 use rustfs_protos::evict_failed_connection;
@@ -99,7 +98,6 @@ const NS_SCANNER_CAPABILITY_PROBE_TIMEOUT: Duration = Duration::from_secs(5);
 const REMOTE_DISK_READ_RETRY_BASE_BACKOFF: Duration = Duration::from_millis(50);
 const ENV_RUSTFS_METADATA_BATCH_READ: &str = "RUSTFS_METADATA_BATCH_READ";
 const LEGACY_ENV_RUSTFS_BATCH_METADATA_RPC: &str = "RUSTFS_BATCH_METADATA_RPC";
-const ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE: &str = "RUSTFS_GET_METADATA_READ_VERSION_COALESCE";
 const BATCH_METADATA_RPC_OFF: &str = "off";
 const BATCH_METADATA_RPC_AUTO: &str = "auto";
 const BATCH_METADATA_RPC_ON: &str = "on";
@@ -204,8 +202,7 @@ fn parse_batch_metadata_rpc_mode(raw: &str) -> BatchMetadataRpcMode {
 }

 fn batch_metadata_rpc_mode_from_env() -> BatchMetadataRpcMode {
-    rustfs_utils::get_env_opt_str(ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE)
-        .or_else(|| rustfs_utils::get_env_opt_str(ENV_RUSTFS_METADATA_BATCH_READ))
+    rustfs_utils::get_env_opt_str(ENV_RUSTFS_METADATA_BATCH_READ)
        .or_else(|| rustfs_utils::get_env_opt_str(LEGACY_ENV_RUSTFS_BATCH_METADATA_RPC))
        .as_deref()
        .map(parse_batch_metadata_rpc_mode)
@@ -1829,12 +1826,6 @@ fn record_read_version_stage(stage: &'static str, started_at: Option<Instant>) {
    }
 }

-fn record_batch_read_version_stage(stage: &'static str, started_at: Option<Instant>) {
-    if let Some(started_at) = started_at {
-        crate::cluster::rpc::runtime_sources::record_remote_disk_grpc_batch_read_version_stage(stage, started_at.elapsed());
-    }
-}
-
 /// Aggregate encoded size (bytes) of a `ReadMultiple` response, preferring the msgpack payloads
 /// and falling back to the JSON compatibility strings. Used to size the RPC for the payload
 /// histogram / large-payload alerting (grpc-optimization P0 instrumentation).
@@ -1945,27 +1936,6 @@ fn decode_batch_read_version_response_items(
    Ok(batch_read_version_resps)
 }

-fn batch_read_version_request_payload_len(req: &BatchReadVersionReq, req_json: &str, req_bin: &[u8]) -> usize {
-    req.items
-        .iter()
-        .fold(req_json.len().saturating_add(req_bin.len()), |total, item| {
-            total
-                .saturating_add(item.org_volume.len())
-                .saturating_add(item.volume.len())
-                .saturating_add(item.path.len())
-                .saturating_add(item.version_id.len())
-        })
-}
-
-fn batch_read_version_response_payload_len(response: &BatchReadVersionResponse) -> usize {
-    response
-        .batch_read_version_resps
-        .iter()
-        .map(String::len)
-        .sum::<usize>()
-        .saturating_add(response.batch_read_version_resps_bin.iter().map(Bytes::len).sum::<usize>())
-}
-
 fn validate_decoded_file_info(file_info: &FileInfo) -> Result<()> {
    file_info.validate_for_metadata_read().map_err(Into::into)
 }
@@ -2867,19 +2837,14 @@ impl DiskAPI for RemoteDisk {
            state = "started",
            "Remote disk RPC started"
        );
-        let batch_read_version_attribution_enabled = rustfs_io_metrics::get_stage_metrics_enabled();
-        let encode_started = read_version_stage_timer(batch_read_version_attribution_enabled);
        let batch_read_version_req = compat_json(&req)?;
        let batch_read_version_req_bin = encode_msgpack(&req)?;
-        record_batch_read_version_stage(INTERNODE_STAGE_BATCH_READ_VERSION_REQUEST_ENCODE, encode_started);
-        let request_payload_bytes = batch_read_version_attribution_enabled
-            .then(|| batch_read_version_request_payload_len(&req, &batch_read_version_req, &batch_read_version_req_bin));
+
        let batch_result = self
            .execute_with_timeout_for_op(
                "batch_read_version",
                move || async move {
                    let disk = self.disk_ref().await;
-                    let disk_len = disk.len();
                    let mut client = self
                        .get_bulk_client()
                        .await
@@ -2890,20 +2855,9 @@ impl DiskAPI for RemoteDisk {
                        batch_read_version_req_bin: batch_read_version_req_bin.into(),
                    });

-                    crate::cluster::rpc::runtime_sources::record_remote_disk_grpc_batch_read_version_request();
-                    if let Some(request_payload_bytes) = request_payload_bytes {
-                        crate::cluster::rpc::runtime_sources::record_remote_disk_grpc_batch_read_version_sent_bytes(
-                            request_payload_bytes.saturating_add(disk_len),
-                        );
-                    }
-                    let rpc_started = read_version_stage_timer(batch_read_version_attribution_enabled);
                    let response = match client.batch_read_version(request).await {
-                        Ok(response) => {
-                            record_batch_read_version_stage(INTERNODE_STAGE_BATCH_READ_VERSION_RPC_ROUNDTRIP, rpc_started);
-                            response.into_inner()
-                        }
+                        Ok(response) => response.into_inner(),
                        Err(status) if status.code() == Code::Unimplemented => {
-                            record_batch_read_version_stage(INTERNODE_STAGE_BATCH_READ_VERSION_RPC_ROUNDTRIP, rpc_started);
                            if mode.should_fallback_on_unimplemented() {
                                record_batch_read_version_gate_decision(mode, BATCH_READ_VERSION_GATE_FALLBACK_UNIMPLEMENTED);
                                warn!(
@@ -2920,7 +2874,6 @@ impl DiskAPI for RemoteDisk {
                            }

                            record_batch_read_version_gate_decision(mode, BATCH_READ_VERSION_GATE_UNSUPPORTED_NO_FALLBACK);
-                            crate::cluster::rpc::runtime_sources::record_remote_disk_grpc_batch_read_version_error();
                            warn!(
                                event = EVENT_REMOTE_DISK_RPC,
                                component = LOG_COMPONENT_ECSTORE,
@@ -2933,33 +2886,14 @@ impl DiskAPI for RemoteDisk {
                            );
                            return Err(Error::from(status));
                        }
-                        Err(status) => {
-                            record_batch_read_version_stage(INTERNODE_STAGE_BATCH_READ_VERSION_RPC_ROUNDTRIP, rpc_started);
-                            crate::cluster::rpc::runtime_sources::record_remote_disk_grpc_batch_read_version_error();
-                            return Err(Error::from(status));
-                        }
+                        Err(status) => return Err(Error::from(status)),
                    };

                    if !response.success {
-                        crate::cluster::rpc::runtime_sources::record_remote_disk_grpc_batch_read_version_error();
                        return Err(response.error.unwrap_or_default().into());
                    }

-                    crate::cluster::rpc::runtime_sources::record_remote_disk_grpc_batch_read_version_recv_bytes(
-                        batch_read_version_response_payload_len(&response),
-                    );
-                    let decode_started = read_version_stage_timer(batch_read_version_attribution_enabled);
-                    match decode_batch_read_version_response_items(response, &self.endpoint) {
-                        Ok(batch_read_version_resps) => {
-                            record_batch_read_version_stage(INTERNODE_STAGE_BATCH_READ_VERSION_RESPONSE_DECODE, decode_started);
-                            Ok(Some(batch_read_version_resps))
-                        }
-                        Err(err) => {
-                            record_batch_read_version_stage(INTERNODE_STAGE_BATCH_READ_VERSION_RESPONSE_DECODE, decode_started);
-                            crate::cluster::rpc::runtime_sources::record_remote_disk_grpc_batch_read_version_error();
-                            Err(err)
-                        }
-                    }
+                    decode_batch_read_version_response_items(response, &self.endpoint).map(Some)
                },
                get_max_timeout_duration(),
            )
@@ -4687,7 +4621,6 @@ mod tests {
            } else {
                "file version not found".to_string()
            },
-            error_code: if success { 0 } else { DiskError::FileVersionNotFound.to_u32() },
        }
    }

@@ -4807,7 +4740,6 @@ mod tests {
    fn batch_metadata_rpc_mode_uses_documented_env_before_legacy_alias() {
        temp_env::with_vars(
            [
-                (ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE, None::<&str>),
                (ENV_RUSTFS_METADATA_BATCH_READ, Some("auto")),
                (LEGACY_ENV_RUSTFS_BATCH_METADATA_RPC, Some("on")),
            ],
@@ -4817,25 +4749,10 @@ mod tests {
        );
    }

-    #[test]
-    fn batch_metadata_rpc_mode_uses_get_coalescer_env_before_batch_env() {
-        temp_env::with_vars(
-            [
-                (ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE, Some("on")),
-                (ENV_RUSTFS_METADATA_BATCH_READ, Some("off")),
-                (LEGACY_ENV_RUSTFS_BATCH_METADATA_RPC, Some("off")),
-            ],
-            || {
-                assert_eq!(batch_metadata_rpc_mode_from_env(), BatchMetadataRpcMode::On);
-            },
-        );
-    }
-
    #[test]
    fn batch_metadata_rpc_mode_falls_back_to_legacy_env_alias() {
        temp_env::with_vars(
            [
-                (ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE, None::<&str>),
                (ENV_RUSTFS_METADATA_BATCH_READ, None::<&str>),
                (LEGACY_ENV_RUSTFS_BATCH_METADATA_RPC, Some("on")),
            ],
@@ -14,10 +14,9 @@

 use rustfs_io_metrics::internode_metrics::{
    INTERNODE_MSGPACK_CODEC_JSON, INTERNODE_MSGPACK_CODEC_MSGPACK, INTERNODE_MSGPACK_DIRECTION_RESPONSE,
-    INTERNODE_OPERATION_GRPC_BATCH_READ_VERSION, INTERNODE_OPERATION_GRPC_READ_ALL, INTERNODE_OPERATION_GRPC_READ_MULTIPLE,
-    INTERNODE_OPERATION_GRPC_READ_VERSION, INTERNODE_OPERATION_GRPC_WRITE_ALL, INTERNODE_OPERATION_PUT_FILE_STREAM,
-    INTERNODE_OPERATION_READ_FILE_STREAM, INTERNODE_TRANSPORT_BACKEND_GRPC, INTERNODE_TRANSPORT_BACKEND_TCP_HTTP,
-    global_internode_metrics,
+    INTERNODE_OPERATION_GRPC_READ_ALL, INTERNODE_OPERATION_GRPC_READ_MULTIPLE, INTERNODE_OPERATION_GRPC_READ_VERSION,
+    INTERNODE_OPERATION_GRPC_WRITE_ALL, INTERNODE_OPERATION_PUT_FILE_STREAM, INTERNODE_OPERATION_READ_FILE_STREAM,
+    INTERNODE_TRANSPORT_BACKEND_GRPC, INTERNODE_TRANSPORT_BACKEND_TCP_HTTP, global_internode_metrics,
 };
 use std::time::Duration;

@@ -94,59 +93,6 @@ pub(crate) fn record_remote_disk_grpc_read_version_request() {
    );
 }

-pub(crate) fn record_remote_disk_grpc_batch_read_version_request() {
-    if !rustfs_io_metrics::get_stage_metrics_enabled() {
-        return;
-    }
-    global_internode_metrics().record_outgoing_request_for_operation_and_backend(
-        INTERNODE_OPERATION_GRPC_BATCH_READ_VERSION,
-        INTERNODE_TRANSPORT_BACKEND_GRPC,
-    );
-}
-
-pub(crate) fn record_remote_disk_grpc_batch_read_version_stage(stage: &'static str, duration: Duration) {
-    if !rustfs_io_metrics::get_stage_metrics_enabled() {
-        return;
-    }
-    global_internode_metrics().record_stage_duration_for_operation_and_backend(
-        INTERNODE_OPERATION_GRPC_BATCH_READ_VERSION,
-        INTERNODE_TRANSPORT_BACKEND_GRPC,
-        stage,
-        duration,
-    );
-}
-
-pub(crate) fn record_remote_disk_grpc_batch_read_version_error() {
-    if !rustfs_io_metrics::get_stage_metrics_enabled() {
-        return;
-    }
-    global_internode_metrics()
-        .record_error_for_operation_and_backend(INTERNODE_OPERATION_GRPC_BATCH_READ_VERSION, INTERNODE_TRANSPORT_BACKEND_GRPC);
-}
-
-pub(crate) fn record_remote_disk_grpc_batch_read_version_sent_bytes(bytes: usize) {
-    if !rustfs_io_metrics::get_stage_metrics_enabled() {
-        return;
-    }
-    global_internode_metrics().record_sent_bytes_for_operation_and_backend(
-        INTERNODE_OPERATION_GRPC_BATCH_READ_VERSION,
-        INTERNODE_TRANSPORT_BACKEND_GRPC,
-        bytes,
-    );
-}
-
-pub(crate) fn record_remote_disk_grpc_batch_read_version_recv_bytes(bytes: usize) {
-    if !rustfs_io_metrics::get_stage_metrics_enabled() {
-        return;
-    }
-    global_internode_metrics().record_recv_bytes_for_operation_and_backend(
-        INTERNODE_OPERATION_GRPC_BATCH_READ_VERSION,
-        INTERNODE_TRANSPORT_BACKEND_GRPC,
-        bytes,
-    );
-    record_grpc_payload_size(INTERNODE_OPERATION_GRPC_BATCH_READ_VERSION, bytes);
-}
-
 pub(crate) fn record_remote_disk_grpc_read_version_error() {
    if !rustfs_io_metrics::get_stage_metrics_enabled() {
        return;
@@ -248,13 +248,10 @@ impl Config {
        let shard_size = shard_size as usize;
        // Keep the historical two-data-shard object budget while preventing
        // wider EC layouts from multiplying the maximum inline object size.
-        // Use div_ceil to match the shard_file_size calculation (which also uses
-        // div_ceil), avoiding a 1-byte rounding discrepancy that prevents inline
-        // for objects right at the threshold.
        let inline_block = if self.initialized && self.inline_block_explicit {
            self.inline_block
        } else {
-            DEFAULT_INLINE_OBJECT_BUDGET.div_ceil(data_shards).min(DEFAULT_INLINE_BLOCK)
+            (DEFAULT_INLINE_OBJECT_BUDGET / data_shards).min(DEFAULT_INLINE_BLOCK)
        };

        if versioned {
@@ -21,7 +21,7 @@ use crate::storage_api_contracts::{
    bucket::{BucketInfo, BucketOperations, BucketOptions, DeleteBucketOptions, MakeBucketOptions},
    list::{StorageListObjectVersionsInfo, StorageListObjectsV2Info, StorageObjectInfoOrErr, StorageWalkOptions},
    multipart::{CompletePart, ListMultipartsInfo, ListPartsInfo, MultipartInfo, MultipartUploadResult, PartInfo},
-    object::{DeleteAccounting, DeletedObject, ObjectIO as _, ObjectOperations as _, ObjectToDelete},
+    object::{DeletedObject, ObjectIO as _, ObjectOperations as _, ObjectToDelete},
    range::HTTPRangeSpec,
 };
 use crate::{
@@ -414,66 +414,6 @@ fn apply_delete_objects_results(
    }
 }

-fn apply_delete_accounting_results(
-    accounting: &mut [Option<DeleteAccounting>],
-    set_objects: &[DelObj],
-    set_accounting: &[Option<DeleteAccounting>],
-) {
-    for (obj, value) in set_objects.iter().zip(set_accounting.iter()) {
-        accounting[obj.orig_idx] = value.clone();
-    }
-}
-
-impl Sets {
-    pub(crate) async fn delete_objects_with_accounting(
-        &self,
-        bucket: &str,
-        objects: Vec<ObjectToDelete>,
-        opts: ObjectOptions,
-    ) -> (Vec<DeletedObject>, Vec<Option<Error>>, Vec<Option<DeleteAccounting>>) {
-        let mut del_objects = vec![DeletedObject::default(); objects.len()];
-        let mut del_errs = vec![None; objects.len()];
-        let mut accounting = vec![None; objects.len()];
-        let mut set_obj_map = HashMap::new();
-
-        for (i, obj) in objects.iter().enumerate() {
-            let idx = self.get_hashed_set_index(obj.object_name.as_str());
-            set_obj_map.entry(idx).or_insert_with(Vec::new).push(DelObj {
-                orig_idx: i,
-                obj: obj.clone(),
-            });
-        }
-
-        let max_concurrent = set_obj_map.len().min(num_cpus::get()).max(1);
-        let semaphore = Arc::new(tokio::sync::Semaphore::new(max_concurrent));
-        let mut futures = FuturesUnordered::new();
-        let bucket = bucket.to_owned();
-
-        for (set_index, set_objects) in set_obj_map {
-            let disks = self.get_disks(set_index);
-            let objects = set_objects.iter().map(|entry| entry.obj.clone()).collect::<Vec<_>>();
-            let bucket = bucket.clone();
-            let opts = opts.clone();
-            let semaphore = semaphore.clone();
-            futures.push(async move {
-                let _permit = semaphore
-                    .acquire_owned()
-                    .await
-                    .expect("delete_objects semaphore should remain open");
-                let (deleted, errors, accounting) = disks.delete_objects_with_accounting(&bucket, objects, opts).await;
-                (set_objects, deleted, errors, accounting)
-            });
-        }
-
-        while let Some((set_objects, deleted, errors, set_accounting)) = futures.next().await {
-            apply_delete_objects_results(&mut del_objects, &mut del_errs, &set_objects, &deleted, errors);
-            apply_delete_accounting_results(&mut accounting, &set_objects, &set_accounting);
-        }
-
-        (del_objects, del_errs, accounting)
-    }
-}
-
 #[async_trait::async_trait]
 impl crate::storage_api_contracts::object::ObjectIO for Sets {
    type Error = Error;
@@ -715,8 +655,65 @@ impl crate::storage_api_contracts::object::ObjectOperations for Sets {
        objects: Vec<ObjectToDelete>,
        opts: ObjectOptions,
    ) -> (Vec<DeletedObject>, Vec<Option<Error>>) {
-        let (deleted, errors, _) = self.delete_objects_with_accounting(bucket, objects, opts).await;
-        (deleted, errors)
+        // Default return value
+        let mut del_objects = vec![DeletedObject::default(); objects.len()];
+
+        let mut del_errs = Vec::with_capacity(objects.len());
+        for _ in 0..objects.len() {
+            del_errs.push(None)
+        }
+
+        let mut set_obj_map = HashMap::new();
+
+        // hash key
+        for (i, obj) in objects.iter().enumerate() {
+            let idx = self.get_hashed_set_index(obj.object_name.as_str());
+
+            if !set_obj_map.contains_key(&idx) {
+                set_obj_map.insert(
+                    idx,
+                    vec![DelObj {
+                        // set_idx: idx,
+                        orig_idx: i,
+                        obj: obj.clone(),
+                    }],
+                );
+            } else if let Some(val) = set_obj_map.get_mut(&idx) {
+                val.push(DelObj {
+                    // set_idx: idx,
+                    orig_idx: i,
+                    obj: obj.clone(),
+                });
+            }
+        }
+
+        let max_concurrent = set_obj_map.len().min(num_cpus::get()).max(1);
+        let semaphore = Arc::new(tokio::sync::Semaphore::new(max_concurrent));
+        let mut futures = FuturesUnordered::new();
+        let bucket = bucket.to_string();
+
+        for (k, v) in set_obj_map {
+            let disks = self.get_disks(k);
+            let objs: Vec<ObjectToDelete> = v.iter().map(|v| v.obj.clone()).collect();
+            let bucket = bucket.clone();
+            let opts = opts.clone();
+            let semaphore = semaphore.clone();
+
+            futures.push(async move {
+                let _permit = semaphore
+                    .acquire_owned()
+                    .await
+                    .expect("delete_objects semaphore should remain open");
+                let (dobjects, errs) = disks.delete_objects(&bucket, objs, opts).await;
+                (v, dobjects, errs)
+            });
+        }
+
+        while let Some((v, dobjects, errs)) = futures.next().await {
+            apply_delete_objects_results(&mut del_objects, &mut del_errs, &v, &dobjects, errs);
+        }
+
+        (del_objects, del_errs)
    }

    #[tracing::instrument(skip(self))]
@@ -26,7 +26,7 @@ use crate::storage_api_contracts::{
    namespace::NamespaceLocking as _,
    object::{HTTPPreconditions, ObjectOperations as _},
 };
-use crate::store::{ECStore, ObjectLockDiagGuard, SourceCleanupMutationFence};
+use crate::store::ECStore;
 use bytes::Bytes;
 use rustfs_filemeta::{FileInfo, FileInfoVersions, ObjectPartInfo};
 use rustfs_rio::{EtagResolvable, HashReader, HashReaderDetector, Index, TryGetIndex};
@@ -856,6 +856,7 @@ fn is_equivalent_data_movement_object(source: &ObjectInfo, target: &ObjectInfo)
 fn is_superseding_unversioned_data_movement_object(source: &ObjectInfo, target: &ObjectInfo) -> bool {
    is_unversioned_data_movement_object(source)
        && is_unversioned_data_movement_object(target)
+        && !target.delete_marker
        && source
            .mod_time
            .zip(target.mod_time)
@@ -1027,7 +1028,6 @@ pub(crate) enum SourceCleanupError {
 pub(crate) struct SourceCleanupBucketFence<'a> {
    pub(crate) expected_incarnation_id: Option<uuid::Uuid>,
    pub(crate) lifecycle_guard: Option<&'a rustfs_lock::NamespaceLockGuard>,
-    pub(crate) object_mutation_fence: Option<&'a SourceCleanupMutationFence>,
 }

 fn ensure_source_cleanup_versions_match(
@@ -1065,9 +1065,7 @@ pub(crate) async fn ensure_source_cleanup_versions_unchanged(
 struct SourceCleanupDeleteBarrierState {
    bucket: String,
    object: String,
-    fence_pending: tokio::sync::Notify,
    arrived: tokio::sync::Notify,
-    is_paused: AtomicBool,
    release: tokio::sync::Notify,
 }

@@ -1081,7 +1079,7 @@ pub(crate) struct SourceCleanupDeleteBarrier {
 }

 #[cfg(test)]
-static SOURCE_CLEANUP_DELETE_BARRIERS: std::sync::OnceLock<std::sync::Mutex<Vec<Arc<SourceCleanupDeleteBarrierState>>>> =
+static SOURCE_CLEANUP_DELETE_BARRIER: std::sync::OnceLock<std::sync::Mutex<Option<Arc<SourceCleanupDeleteBarrierState>>>> =
    std::sync::OnceLock::new();

 #[cfg(test)]
@@ -1094,22 +1092,15 @@ impl SourceCleanupDeleteBarrier {
        let state = Arc::new(SourceCleanupDeleteBarrierState {
            bucket: bucket.to_string(),
            object: object.to_string(),
-            fence_pending: tokio::sync::Notify::new(),
            arrived: tokio::sync::Notify::new(),
-            is_paused: AtomicBool::new(false),
            release: tokio::sync::Notify::new(),
        });
-        let mut barriers = SOURCE_CLEANUP_DELETE_BARRIERS
-            .get_or_init(|| std::sync::Mutex::new(Vec::new()))
+        let mut slot = SOURCE_CLEANUP_DELETE_BARRIER
+            .get_or_init(|| std::sync::Mutex::new(None))
            .lock()
            .expect("source cleanup delete barrier mutex should not poison");
-        assert!(
-            !barriers
-                .iter()
-                .any(|barrier| barrier.bucket == bucket && barrier.object == object),
-            "source cleanup delete barrier must be unique per object"
-        );
-        barriers.push(Arc::clone(&state));
+        assert!(slot.is_none(), "source cleanup delete barrier must be unique");
+        *slot = Some(Arc::clone(&state));
        Self { state }
    }

@@ -1119,58 +1110,35 @@ impl SourceCleanupDeleteBarrier {
            .expect("source cleanup should reach the pre-delete barrier");
    }

-    pub(crate) async fn wait_until_fence_pending(&self) {
-        tokio::time::timeout(StdDuration::from_secs(30), self.state.fence_pending.notified())
-            .await
-            .expect("source cleanup should attempt the fixed mutation fence");
-    }
-
-    pub(crate) fn is_paused(&self) -> bool {
-        self.state.is_paused.load(Ordering::Acquire)
-    }
-
    pub(crate) fn release(&self) {
        self.state.release.notify_one();
    }
 }

-#[cfg(test)]
-pub(crate) fn notify_source_cleanup_mutation_fence_pending(bucket: &str, object: &str) {
-    let barrier = SOURCE_CLEANUP_DELETE_BARRIERS
-        .get_or_init(|| std::sync::Mutex::new(Vec::new()))
-        .lock()
-        .expect("source cleanup delete barrier mutex should not poison")
-        .iter()
-        .find(|barrier| barrier.bucket == bucket && barrier.object == object)
-        .cloned();
-    if let Some(barrier) = barrier {
-        barrier.fence_pending.notify_one();
-    }
-}
-
 #[cfg(test)]
 impl Drop for SourceCleanupDeleteBarrier {
    fn drop(&mut self) {
        self.state.release.notify_one();
-        let mut barriers = SOURCE_CLEANUP_DELETE_BARRIERS
-            .get_or_init(|| std::sync::Mutex::new(Vec::new()))
+        let mut slot = SOURCE_CLEANUP_DELETE_BARRIER
+            .get_or_init(|| std::sync::Mutex::new(None))
            .lock()
            .expect("source cleanup delete barrier mutex should not poison");
-        barriers.retain(|state| !Arc::ptr_eq(state, &self.state));
+        if slot.as_ref().is_some_and(|state| Arc::ptr_eq(state, &self.state)) {
+            *slot = None;
+        }
    }
 }

 #[cfg(test)]
 async fn pause_source_cleanup_before_delete(bucket: &str, object: &str) {
-    let barrier = SOURCE_CLEANUP_DELETE_BARRIERS
-        .get_or_init(|| std::sync::Mutex::new(Vec::new()))
+    let barrier = SOURCE_CLEANUP_DELETE_BARRIER
+        .get_or_init(|| std::sync::Mutex::new(None))
        .lock()
        .expect("source cleanup delete barrier mutex should not poison")
-        .iter()
-        .find(|barrier| barrier.bucket == bucket && barrier.object == object)
+        .as_ref()
+        .filter(|barrier| barrier.bucket == bucket && barrier.object == object)
        .cloned();
    if let Some(barrier) = barrier {
-        barrier.is_paused.store(true, Ordering::Release);
        barrier.arrived.notify_one();
        barrier.release.notified().await;
    }
@@ -1186,20 +1154,11 @@ pub(crate) async fn cleanup_source_entry_if_unchanged(
    op_label: &str,
 ) -> std::result::Result<ObjectInfo, SourceCleanupError> {
    let cleanup_key = encode_dir_object(object);
-    let source_guard = if bucket_fence
-        .object_mutation_fence
-        .is_some_and(SourceCleanupMutationFence::source_lock_covered)
-    {
-        None
-    } else {
-        let ns_lock = set.new_ns_lock(bucket, cleanup_key.as_str()).await?;
-        Some(
-            ns_lock
-                .get_write_lock(get_lock_acquire_timeout())
-                .await
-                .map_err(Error::from)?,
-        )
-    };
+    let ns_lock = set.new_ns_lock(bucket, cleanup_key.as_str()).await?;
+    let _guard = ns_lock
+        .get_write_lock(get_lock_acquire_timeout())
+        .await
+        .map_err(Error::from)?;

    if bucket_fence
        .lifecycle_guard
@@ -1209,14 +1168,6 @@ pub(crate) async fn cleanup_source_entry_if_unchanged(
            "{op_label}: bucket incarnation fence was lost before source cleanup"
        ))));
    }
-    if bucket_fence
-        .object_mutation_fence
-        .is_some_and(SourceCleanupMutationFence::is_lock_lost)
-    {
-        return Err(SourceCleanupError::Storage(Error::other(format!(
-            "{op_label}: object mutation fence was lost before source cleanup"
-        ))));
-    }

    ensure_source_cleanup_versions_unchanged(set.clone(), bucket, object, expected, allowed_missing, op_label).await?;

@@ -1231,12 +1182,7 @@ pub(crate) async fn cleanup_source_entry_if_unchanged(
        expected_bucket_incarnation_id: bucket_fence.expected_incarnation_id,
        ..Default::default()
    };
-    if let Some(source_guard) = source_guard.as_ref() {
-        opts.add_namespace_lock_guard(source_guard);
-    }
-    if let Some(object_mutation_fence) = bucket_fence.object_mutation_fence {
-        object_mutation_fence.add_namespace_lock_fence(&mut opts);
-    }
+    opts.add_namespace_lock_guard(&_guard);
    if let Some(bucket_lifecycle_guard) = bucket_fence.lifecycle_guard {
        opts.add_bucket_lifecycle_lock_guard(bucket_lifecycle_guard);
    }
@@ -1384,37 +1330,6 @@ fn data_movement_part_upload_failure_stage(err: &Error) -> &'static str {
    }
 }

-pub(crate) async fn migrate_decommission_object(
-    store: Arc<ECStore>,
-    pool_idx: usize,
-    bucket: String,
-    rd: GetObjectReader,
-    source_bucket_incarnation_id: Option<uuid::Uuid>,
-    op_label: &str,
-) -> Result<()> {
-    let source = rd.object_info.clone();
-    let _mutation_fence = store
-        .acquire_decommission_object_mutation_fence(&bucket, &source.name)
-        .await?;
-    let current = find_data_movement_target_info(store.as_ref(), pool_idx, &bucket, &source)
-        .await?
-        .ok_or(Error::FileNotFound)?;
-    if !is_equivalent_data_movement_object_identity(&source, &current, true, false) {
-        return Err(Error::FileNotFound);
-    }
-
-    migrate_object_inner(
-        store,
-        pool_idx,
-        bucket,
-        rd,
-        source_bucket_incarnation_id,
-        op_label,
-        Some(&_mutation_fence),
-    )
-    .await
-}
-
 pub(crate) async fn migrate_object(
    store: Arc<ECStore>,
    pool_idx: usize,
@@ -1422,18 +1337,6 @@ pub(crate) async fn migrate_object(
    rd: GetObjectReader,
    source_bucket_incarnation_id: Option<uuid::Uuid>,
    op_label: &str,
-) -> Result<()> {
-    migrate_object_inner(store, pool_idx, bucket, rd, source_bucket_incarnation_id, op_label, None).await
-}
-
-async fn migrate_object_inner(
-    store: Arc<ECStore>,
-    pool_idx: usize,
-    bucket: String,
-    rd: GetObjectReader,
-    source_bucket_incarnation_id: Option<uuid::Uuid>,
-    op_label: &str,
-    mutation_fence: Option<&ObjectLockDiagGuard>,
 ) -> Result<()> {
    let object_info = rd.object_info.clone();
    let has_part_checksums = object_info
@@ -1447,7 +1350,7 @@ async fn migrate_object_inner(
        let mut new_multipart_opts = data_movement_new_multipart_opts(&object_info, pool_idx);
        new_multipart_opts.expected_bucket_incarnation_id = source_bucket_incarnation_id;
        let (res, target_pool_idx, expected_bucket_incarnation_id) = match store
-            .handle_new_multipart_upload_with_pool_idx(&bucket, &object_info.name, &new_multipart_opts, mutation_fence)
+            .handle_new_multipart_upload_with_pool_idx(&bucket, &object_info.name, &new_multipart_opts)
            .await
        {
            Ok(res) => res,
@@ -1545,7 +1448,7 @@ async fn migrate_object_inner(
            if let Err(err) = store
                .clone()
                .complete_multipart_upload_for_data_movement(
-                    (target_pool_idx, mutation_fence),
+                    target_pool_idx,
                    &bucket,
                    &object_info.name,
                    &res.upload_id,
@@ -1706,7 +1609,7 @@ async fn migrate_object_inner(
    let mut put_opts = data_movement_put_object_opts(&object_info, pool_idx);
    put_opts.expected_bucket_incarnation_id = source_bucket_incarnation_id;
    let (target_pool_idx, put_result) = store
-        .put_object_for_data_movement(&bucket, &object_info.name, &mut data, &put_opts, mutation_fence)
+        .put_object_for_data_movement(&bucket, &object_info.name, &mut data, &put_opts)
        .await
        .map_err(|err| data_movement_stage_error(op_label, "prepare_put_object", &bucket, &object_info.name, err))?;
    if let Err(err) = put_result {
@@ -3638,47 +3541,25 @@ mod tests {
    }

    #[test]
-    fn test_precondition_conflict_accepts_only_newer_null_delete_marker() {
-        for version_id in [None, Some(Uuid::nil())] {
-            let source = ObjectInfo {
-                version_id,
-                size: 128,
-                etag: Some("etag-source".to_string()),
-                mod_time: Some(OffsetDateTime::UNIX_EPOCH),
-                ..Default::default()
-            };
-            let target = ObjectInfo {
-                delete_marker: true,
-                etag: None,
-                mod_time: OffsetDateTime::UNIX_EPOCH.checked_add(time::Duration::SECOND),
-                ..source.clone()
-            };
+    fn test_precondition_conflict_rejects_newer_delete_marker() {
+        let source = ObjectInfo {
+            size: 128,
+            etag: Some("etag-source".to_string()),
+            mod_time: Some(OffsetDateTime::UNIX_EPOCH),
+            ..Default::default()
+        };
+        let target = ObjectInfo {
+            delete_marker: true,
+            etag: None,
+            mod_time: OffsetDateTime::UNIX_EPOCH.checked_add(time::Duration::SECOND),
+            ..source.clone()
+        };

-            assert!(
-                resolve_data_movement_overwrite_resume_result(
-                    &Error::PreconditionFailed,
-                    Ok(Some(target.clone())),
-                    &source,
-                    0,
-                    1,
-                )
-                .expect("newer null delete marker should be evaluated")
-            );
+        let should_resume =
+            resolve_data_movement_overwrite_resume_result(&Error::PreconditionFailed, Ok(Some(target)), &source, 0, 1)
+                .expect("delete marker conflict should be evaluated");

-            let mut same_time = target.clone();
-            same_time.mod_time = source.mod_time;
-            assert!(
-                !resolve_data_movement_overwrite_resume_result(&Error::PreconditionFailed, Ok(Some(same_time)), &source, 0, 1,)
-                    .expect("same-generation null delete marker should be rejected")
-            );
-
-            let mut versioned = target;
-            versioned.version_id = Some(Uuid::new_v4());
-            assert!(
-                !resolve_data_movement_overwrite_resume_result(&Error::PreconditionFailed, Ok(Some(versioned)), &source, 0, 1,)
-                    .expect("a UUID delete marker must not erase a null source version")
-            );
-        }
+        assert!(!should_resume);
    }

    #[test]
@@ -1391,37 +1391,7 @@ impl BucketUsageAccumulator {
 }

 pub fn quota_object_size(object: &ObjectInfo) -> Result<u64, Error> {
-    // A compressed object may carry -1 while the transformed size is unknown
-    // (legacy streaming sentinel). In that case the persisted physical size
-    // is still a valid accounting floor; every other negative value is corrupt.
-    // An explicit negative `actual-size` metadata value is corrupt, however:
-    // the sentinel is only valid in the in-memory/object-part field written by
-    // the legacy streaming path, not as a persisted declared size.
-    let compressed = object.is_compressed();
-    if object.actual_size < -1 || (object.actual_size == -1 && !compressed) {
-        return Err(Error::PartMissingOrCorrupt);
-    }
-    if object
-        .parts
-        .iter()
-        .any(|part| part.actual_size < -1 || (part.actual_size < 0 && !compressed))
-    {
-        return Err(Error::PartMissingOrCorrupt);
-    }
-    let declared_actual_size = rustfs_utils::http::get_str(&object.user_defined, rustfs_utils::http::SUFFIX_ACTUAL_SIZE)
-        .filter(|value| !value.is_empty());
-    if declared_actual_size
-        .as_deref()
-        .and_then(|value| value.parse::<i64>().ok())
-        .is_some_and(|size| size < 0)
-    {
-        return Err(Error::PartMissingOrCorrupt);
-    }
-    let logical_size = match object.get_actual_size().map_err(Error::other)? {
-        size if size == -1 && compressed && declared_actual_size.is_none() => None,
-        size if size >= 0 => Some(u64::try_from(size).map_err(|_| Error::PartMissingOrCorrupt)?),
-        _ => return Err(Error::PartMissingOrCorrupt),
-    };
+    let logical_size = u64::try_from(object.get_actual_size().map_err(Error::other)?).map_err(|_| Error::PartMissingOrCorrupt)?;
    let persisted_part_size = if object.parts.is_empty() {
        u64::try_from(object.size).map_err(|_| Error::PartMissingOrCorrupt)?
    } else {
@@ -1429,8 +1399,12 @@ pub fn quota_object_size(object: &ObjectInfo) -> Result<u64, Error> {
            // Compressed streaming objects persist -1 when the transformed
            // part size is unknown. The physical part size remains a valid
            // quota floor; reject only non-negative values that overflow.
-            let actual_size = if part.actual_size == -1 {
-                0
+            let actual_size = if part.actual_size < 0 {
+                if object.is_compressed() {
+                    0
+                } else {
+                    return Err(Error::PartMissingOrCorrupt);
+                }
            } else {
                u64::try_from(part.actual_size).map_err(|_| Error::PartMissingOrCorrupt)?
            };
@@ -1438,7 +1412,7 @@ pub fn quota_object_size(object: &ObjectInfo) -> Result<u64, Error> {
            total.checked_add(part_size).ok_or(Error::PartMissingOrCorrupt)
        })?
    };
-    Ok(logical_size.unwrap_or(0).max(persisted_part_size))
+    Ok(logical_size.max(persisted_part_size))
 }

 type UsageVersionPage = StorageListObjectVersionsInfo<ObjectInfo>;
@@ -3346,80 +3320,6 @@ mod tests {
        );
    }

-    #[test]
-    fn quota_object_size_accepts_compressed_unknown_actual_size_sentinel() {
-        let mut metadata = HashMap::new();
-        rustfs_utils::http::insert_str(
-            &mut metadata,
-            rustfs_utils::http::SUFFIX_COMPRESSION,
-            "klauspost/compress/s2".to_string(),
-        );
-        let object = ObjectInfo {
-            size: 400,
-            actual_size: -1,
-            user_defined: Arc::new(metadata),
-            ..Default::default()
-        };
-
-        assert_eq!(quota_object_size(&object).expect("compressed sentinel is valid"), 400);
-    }
-
-    #[test]
-    fn quota_object_size_rejects_compressed_part_sum_overflow() {
-        let mut metadata = HashMap::new();
-        rustfs_utils::http::insert_str(
-            &mut metadata,
-            rustfs_utils::http::SUFFIX_COMPRESSION,
-            "klauspost/compress/s2".to_string(),
-        );
-        let object = ObjectInfo {
-            size: 1,
-            user_defined: Arc::new(metadata),
-            parts: Arc::new(vec![
-                rustfs_filemeta::ObjectPartInfo {
-                    actual_size: i64::MAX,
-                    ..Default::default()
-                },
-                rustfs_filemeta::ObjectPartInfo {
-                    actual_size: 1,
-                    ..Default::default()
-                },
-            ]),
-            ..Default::default()
-        };
-
-        assert!(matches!(quota_object_size(&object), Err(Error::Io(_))));
-    }
-
-    #[test]
-    fn quota_object_size_rejects_negative_values_other_than_the_compressed_sentinel() {
-        let mut metadata = HashMap::new();
-        rustfs_utils::http::insert_str(
-            &mut metadata,
-            rustfs_utils::http::SUFFIX_COMPRESSION,
-            "klauspost/compress/s2".to_string(),
-        );
-        let corrupt_object = ObjectInfo {
-            size: 400,
-            actual_size: -2,
-            user_defined: Arc::new(metadata.clone()),
-            ..Default::default()
-        };
-        assert!(matches!(quota_object_size(&corrupt_object), Err(Error::PartMissingOrCorrupt)));
-
-        let corrupt_part = ObjectInfo {
-            size: 400,
-            user_defined: Arc::new(metadata),
-            parts: Arc::new(vec![rustfs_filemeta::ObjectPartInfo {
-                size: 400,
-                actual_size: -2,
-                ..Default::default()
-            }]),
-            ..Default::default()
-        };
-        assert!(matches!(quota_object_size(&corrupt_part), Err(Error::PartMissingOrCorrupt)));
-    }
-
    #[tokio::test]
    #[serial]
    async fn live_bucket_usage_refreshes_are_coalesced_only_while_in_flight() {
@@ -858,7 +858,6 @@ const EVENT_DISK_LOCAL_DIRECT_IO_FALLBACK: &str = "disk_local_direct_io_fallback
 #[cfg(target_os = "linux")]
 const EVENT_DISK_LOCAL_URING_LATCH_OFF: &str = "disk_local_uring_latch_off";
 const EVENT_DISK_LOCAL_DELETE_FAILED: &str = "disk_local_delete_failed";
-const EVENT_DISK_LOCAL_DELETE_ROLLBACK_FAILED: &str = "disk_local_delete_rollback_failed";
 const EVENT_DISK_LOCAL_CHECK_PARTS: &str = "disk_local_check_parts";
 const EVENT_DISK_LOCAL_ACCESS_FAILED: &str = "disk_local_access_failed";
 const EVENT_DISK_LOCAL_VOLUME_SETUP_FAILED: &str = "disk_local_volume_setup_failed";
@@ -6107,43 +6106,6 @@ impl LocalDisk {
        Ok((bytes, modtime))
    }

-    async fn write_missing_delete_marker(
-        &self,
-        volume: &str,
-        path: &str,
-        fi: FileInfo,
-        object_dir: &Path,
-        xl_path: &Path,
-        rollback_dir: Option<Uuid>,
-    ) -> Result<()> {
-        if let Some(rollback_dir) = rollback_dir {
-            let rollback_path = object_dir.join(rollback_dir.to_string());
-            fs::create_dir_all(&rollback_path).await.map_err(to_file_error)?;
-            fs::write(rollback_path.join(DELETE_MARKER_ROLLBACK_FILE), [])
-                .await
-                .map_err(to_file_error)?;
-        }
-        if let Err(err) = self.write_metadata("", volume, path, fi).await {
-            if let Some(rollback_dir) = rollback_dir
-                && let Err(restore_err) = restore_delete_rollback(object_dir, xl_path, rollback_dir, &self.publication_root).await
-            {
-                warn!(
-                    event = EVENT_DISK_LOCAL_DELETE_ROLLBACK_FAILED,
-                    component = LOG_COMPONENT_ECSTORE,
-                    subsystem = LOG_SUBSYSTEM_DISK_LOCAL,
-                    result = "failed",
-                    volume,
-                    path,
-                    rollback_dir = %rollback_dir,
-                    error = ?restore_err,
-                    "Disk local delete rollback failed"
-                );
-            }
-            return Err(err);
-        }
-        Ok(())
-    }
-
    async fn delete_versions_internal(&self, volume: &str, path: &str, fis: &[FileInfo], opts: &DeleteOptions) -> Result<()> {
        let volume_dir = self.io_get_bucket_path(volume)?;
        let xlpath = self.io_get_object_path(volume, format!("{path}/{STORAGE_FORMAT_FILE}").as_str())?;
@@ -6161,20 +6123,7 @@ impl LocalDisk {
            return restore_metadata_backup(object_dir, &xlpath, rollback_dir, &self.publication_root).await;
        }

-        let (data, _) = match self.read_all_data_with_dmtime(volume, volume_dir.as_path(), &xlpath).await {
-            Ok(data) => data,
-            Err(DiskError::FileNotFound) => {
-                // `deleted` alone can be an explicit marker purge; only
-                // `mark_deleted` may create metadata that was not present.
-                let Some(delete_marker) = fis.iter().find(|fi| fi.deleted && fi.mark_deleted).cloned() else {
-                    return Err(DiskError::FileNotFound);
-                };
-                return self
-                    .write_missing_delete_marker(volume, path, delete_marker, object_dir, &xlpath, opts.old_data_dir)
-                    .await;
-            }
-            Err(err) => return Err(err),
-        };
+        let (data, _) = self.read_all_data_with_dmtime(volume, volume_dir.as_path(), &xlpath).await?;

        if data.is_empty() {
            return Err(DiskError::FileNotFound);
@@ -10473,9 +10422,29 @@ impl DiskAPI for LocalDisk {
                }

                if fi.deleted && force_del_marker {
-                    return self
-                        .write_missing_delete_marker(volume, path, fi, file_path.as_path(), &xl_path, rollback_dir)
-                        .await;
+                    if let Some(rollback_dir) = rollback_dir {
+                        let rollback_path = file_path.join(rollback_dir.to_string());
+                        fs::create_dir_all(&rollback_path).await.map_err(to_file_error)?;
+                        fs::write(rollback_path.join(DELETE_MARKER_ROLLBACK_FILE), [])
+                            .await
+                            .map_err(to_file_error)?;
+                    }
+                    if let Err(err) = self.write_metadata("", volume, path, fi).await {
+                        if let Some(rollback_dir) = rollback_dir
+                            && let Err(restore_err) =
+                                restore_delete_rollback(file_path.as_path(), &xl_path, rollback_dir, &self.publication_root).await
+                        {
+                            warn!(
+                                volume,
+                                path,
+                                rollback_dir = %rollback_dir,
+                                error = ?restore_err,
+                                "failed to restore metadata after delete marker commit error"
+                            );
+                        }
+                        return Err(err);
+                    }
+                    return Ok(());
                }

                return if fi.version_id.is_some() {
@@ -44,8 +44,6 @@ pub const PART_TRANSACTION_ROLLBACK: &str = "rollback";
 const LOG_COMPONENT_ECSTORE: &str = "ecstore";
 const LOG_SUBSYSTEM_DISK: &str = "disk";
 const EVENT_DISK_PART_ERR_UNCLASSIFIED: &str = "disk_part_err_unclassified";
-const ENV_BATCH_READ_VERSION_SERVER_PARALLELISM: &str = "RUSTFS_BATCH_READ_VERSION_SERVER_PARALLELISM";
-const BATCH_READ_VERSION_SERVER_PARALLELISM: usize = 4;

 pub fn part_transaction_path(part_path: &str) -> String {
    match part_path.rsplit_once('/') {
@@ -64,7 +62,6 @@ use bytes::Bytes;
 use endpoint::Endpoint;
 use error::DiskError;
 use error::{Error, Result};
-use futures::stream::{self, StreamExt};
 use local::LocalDisk;
 use rustfs_filemeta::{FileInfo, ObjectPartInfo, RawFileInfo};
 use rustfs_madmin::info_commands::DiskMetrics;
@@ -420,14 +417,6 @@ impl DiskAPI for Disk {
        }
    }

-    #[tracing::instrument(level = "trace", skip_all)]
-    async fn batch_read_version(&self, req: BatchReadVersionReq) -> Result<Vec<BatchReadVersionResp>> {
-        match self {
-            Disk::Local(local_disk) => local_disk.batch_read_version(req).await,
-            Disk::Remote(remote_disk) => remote_disk.batch_read_version(req).await,
-        }
-    }
-
    #[tracing::instrument(level = "trace", skip_all)]
    async fn read_xl(&self, volume: &str, path: &str, read_data: bool) -> Result<RawFileInfo> {
        match self {
@@ -1039,47 +1028,36 @@ where
    D: DiskAPI + ?Sized,
 {
    validate_batch_read_version_item_count(req.items.len())?;
-    let parallelism = batch_read_version_server_parallelism();

-    let mut responses = stream::iter(req.items.into_iter().enumerate())
-        .map(|(index, item)| async move {
-            match disk
-                .read_version(&item.org_volume, &item.volume, &item.path, &item.version_id, &req.opts)
-                .await
-            {
-                Ok(file_info) => BatchReadVersionResp {
-                    index,
-                    path: item.path,
-                    version_id: item.version_id,
-                    success: true,
-                    file_info,
-                    error: String::new(),
-                    error_code: 0,
-                },
-                Err(err) => BatchReadVersionResp {
-                    index,
-                    path: item.path,
-                    version_id: item.version_id,
-                    success: false,
-                    file_info: FileInfo::default(),
-                    error: err.to_string(),
-                    error_code: err.to_u32(),
-                },
-            }
-        })
-        .buffer_unordered(parallelism)
-        .collect::<Vec<_>>()
-        .await;
-    responses.sort_unstable_by_key(|response| response.index);
+    let mut responses = Vec::with_capacity(req.items.len());
+    for (index, item) in req.items.iter().enumerate() {
+        let response = match disk
+            .read_version(&item.org_volume, &item.volume, &item.path, &item.version_id, &req.opts)
+            .await
+        {
+            Ok(file_info) => BatchReadVersionResp {
+                index,
+                path: item.path.clone(),
+                version_id: item.version_id.clone(),
+                success: true,
+                file_info,
+                error: String::new(),
+            },
+            Err(err) => BatchReadVersionResp {
+                index,
+                path: item.path.clone(),
+                version_id: item.version_id.clone(),
+                success: false,
+                file_info: FileInfo::default(),
+                error: err.to_string(),
+            },
+        };
+        responses.push(response);
+    }

    Ok(responses)
 }

-fn batch_read_version_server_parallelism() -> usize {
-    rustfs_utils::get_env_usize(ENV_BATCH_READ_VERSION_SERVER_PARALLELISM, BATCH_READ_VERSION_SERVER_PARALLELISM)
-        .clamp(1, BATCH_READ_VERSION_MAX_ITEMS)
-}
-
 #[derive(Debug, Default, Serialize, Deserialize)]
 pub struct CheckPartsResp {
    pub results: Vec<usize>,
@@ -1344,8 +1322,6 @@ pub struct BatchReadVersionResp {
    pub success: bool,
    pub file_info: FileInfo,
    pub error: String,
-    #[serde(default)]
-    pub error_code: u32,
 }

 pub fn validate_batch_read_version_item_count(item_count: usize) -> Result<()> {
@@ -1441,26 +1417,6 @@ mod tests {
        assert!(!partial_valid_location.valid());
    }

-    #[test]
-    fn batch_read_version_server_parallelism_defaults_to_conservative_four() {
-        temp_env::with_var(ENV_BATCH_READ_VERSION_SERVER_PARALLELISM, None::<&str>, || {
-            assert_eq!(batch_read_version_server_parallelism(), 4);
-        });
-    }
-
-    #[test]
-    fn batch_read_version_server_parallelism_honors_env_with_bounds() {
-        temp_env::with_var(ENV_BATCH_READ_VERSION_SERVER_PARALLELISM, Some("8"), || {
-            assert_eq!(batch_read_version_server_parallelism(), 8);
-        });
-        temp_env::with_var(ENV_BATCH_READ_VERSION_SERVER_PARALLELISM, Some("0"), || {
-            assert_eq!(batch_read_version_server_parallelism(), 1);
-        });
-        temp_env::with_var(ENV_BATCH_READ_VERSION_SERVER_PARALLELISM, Some("9999"), || {
-            assert_eq!(batch_read_version_server_parallelism(), BATCH_READ_VERSION_MAX_ITEMS);
-        });
-    }
-
    /// Test FileInfoVersions find_version_index
    #[test]
    fn test_file_info_versions_find_version_index() {
@@ -81,14 +81,6 @@ pub fn shutdown_background_monitors() {
    cluster::rpc::shutdown_background_monitors();
 }

-/// Publish that the process is ready to serve user-object GET traffic.
-///
-/// Experimental metadata coalescing is allowed to run only after this point so
-/// startup and internal metadata reads keep the original per-disk path.
-pub fn mark_get_metadata_read_version_coalescing_service_ready() {
-    runtime::global::mark_get_metadata_read_version_coalescing_service_ready();
-}
-
 #[cfg(test)]
 mod rio_tests {
    #[test]
@@ -24,7 +24,7 @@ use crate::storage_api_contracts::{
 pub struct NamespaceLockFence {
    signals: Arc<Vec<Arc<rustfs_lock::distributed_lock::LockLostSignal>>>,
    #[cfg(test)]
-    forced_lost: Arc<Vec<Arc<std::sync::atomic::AtomicBool>>>,
+    forced_lost: Arc<std::sync::atomic::AtomicBool>,
 }

 impl Debug for NamespaceLockFence {
@@ -40,17 +40,13 @@ impl NamespaceLockFence {
        Self {
            signals: Arc::default(),
            #[cfg(test)]
-            forced_lost: Arc::new(vec![Arc::new(std::sync::atomic::AtomicBool::new(false))]),
+            forced_lost: Arc::new(std::sync::atomic::AtomicBool::new(false)),
        }
    }

    pub(crate) fn is_lock_lost(&self) -> bool {
        #[cfg(test)]
-        if self
-            .forced_lost
-            .iter()
-            .any(|lost| lost.load(std::sync::atomic::Ordering::Acquire))
-        {
+        if self.forced_lost.load(std::sync::atomic::Ordering::Acquire) {
            return true;
        }
        self.signals.iter().any(|signal| signal.is_lost())
@@ -61,26 +57,27 @@ impl NamespaceLockFence {
    }

    fn extend(&mut self, other: &Self) {
-        if !Arc::ptr_eq(&self.signals, &other.signals) {
-            Arc::make_mut(&mut self.signals).extend(other.signals.iter().cloned());
+        if Arc::ptr_eq(&self.signals, &other.signals) {
+            return;
        }
+        Arc::make_mut(&mut self.signals).extend(other.signals.iter().cloned());
        #[cfg(test)]
-        if !Arc::ptr_eq(&self.forced_lost, &other.forced_lost) {
-            Arc::make_mut(&mut self.forced_lost).extend(other.forced_lost.iter().cloned());
+        if other.forced_lost.load(std::sync::atomic::Ordering::Acquire) {
+            self.forced_lost.store(true, std::sync::atomic::Ordering::Release);
        }
    }

    #[cfg(test)]
    pub(crate) fn lost_for_test() -> Self {
        let fence = Self::new();
-        fence.forced_lost[0].store(true, std::sync::atomic::Ordering::Release);
+        fence.forced_lost.store(true, std::sync::atomic::Ordering::Release);
        fence
    }

    #[cfg(test)]
    pub(crate) fn loss_handle_for_test() -> (Self, Arc<std::sync::atomic::AtomicBool>) {
        let fence = Self::new();
-        (fence.clone(), Arc::clone(&fence.forced_lost[0]))
+        (fence.clone(), Arc::clone(&fence.forced_lost))
    }
 }

@@ -414,13 +411,6 @@ impl ObjectOptions {
        self.namespace_lock_fence.get_or_insert_with(NamespaceLockFence::new);
    }

-    #[cfg(test)]
-    pub(crate) fn add_namespace_lock_fence_for_test(&mut self, fence: &NamespaceLockFence) {
-        self.namespace_lock_fence
-            .get_or_insert_with(NamespaceLockFence::new)
-            .extend(fence);
-    }
-
    pub(crate) fn ensure_lifecycle_delete_all_journal(&mut self) {
        self.lifecycle_delete_all_journal
            .get_or_insert_with(|| Arc::new(parking_lot::Mutex::new(LifecycleDeleteAllJournalState::default())));
@@ -699,9 +689,6 @@ impl ObjectInfo {
    }

    pub fn get_actual_size(&self) -> std::io::Result<i64> {
-        if self.actual_size < -1 || (self.actual_size == -1 && !self.is_compressed()) {
-            return Err(std::io::Error::other("invalid negative actual size"));
-        }
        if self.actual_size > 0 {
            return Ok(self.actual_size);
        }
@@ -713,25 +700,10 @@ impl ObjectInfo {
                let size = size_str.parse::<i64>().map_err(|e| std::io::Error::other(e.to_string()))?;
                return Ok(size);
            }
-            if self.actual_size == -1 && self.parts.is_empty() {
-                return Ok(-1);
-            }
-            let mut actual_size = 0_i64;
-            let mut unknown = false;
-            for part in self.parts.iter() {
-                match part.actual_size {
-                    -1 => unknown = true,
-                    size if size >= 0 => {
-                        actual_size = actual_size
-                            .checked_add(size)
-                            .ok_or_else(|| std::io::Error::other("compressed actual size overflow"))?;
-                    }
-                    _ => return Err(std::io::Error::other("invalid negative compressed part size")),
-                }
-            }
-            if unknown {
-                return Ok(-1);
-            }
+            let mut actual_size = 0;
+            self.parts.iter().for_each(|part| {
+                actual_size += part.actual_size;
+            });
            if actual_size == 0 && actual_size != self.size {
                return Err(std::io::Error::other(format!("invalid decompressed size {} {}", actual_size, self.size)));
            }
@@ -746,18 +718,6 @@ impl ObjectInfo {
        Ok(self.size)
    }

-    /// Returns a non-negative size for client and replication boundaries.
-    ///
-    /// Compressed legacy metadata can retain the internal `-1` unknown-size
-    /// sentinel. Those boundaries cannot emit a negative length, so they use
-    /// the persisted physical size while quota accounting keeps the sentinel
-    /// distinction in [`crate::data_usage::quota_object_size`].
-    pub fn get_actual_size_or_physical(&self) -> i64 {
-        self.get_actual_size()
-            .map(|size| if size >= 0 { size } else { self.size.max(0) })
-            .unwrap_or_else(|_| self.size.max(0))
-    }
-
    pub fn from_file_info(fi: &FileInfo, bucket: &str, object: &str, versioned: bool) -> ObjectInfo {
        let mut version_id = fi.version_id;

@@ -25,10 +25,7 @@ use lazy_static::lazy_static;
 use rustfs_lock::client::LockClient;
 use std::{
    collections::HashMap,
-    sync::{
-        Arc, OnceLock,
-        atomic::{AtomicBool, Ordering},
-    },
+    sync::{Arc, OnceLock},
    time::SystemTime,
 };
 use tokio::sync::{OnceCell, RwLock};
@@ -40,16 +37,6 @@ pub const DISK_MIN_INODES: u64 = 1000;
 pub const DISK_FILL_FRACTION: f64 = 0.99;
 pub const DISK_RESERVE_FRACTION: f64 = 0.15;

-static GET_METADATA_READ_VERSION_COALESCING_SERVICE_READY: AtomicBool = AtomicBool::new(false);
-
-pub(crate) fn mark_get_metadata_read_version_coalescing_service_ready() {
-    GET_METADATA_READ_VERSION_COALESCING_SERVICE_READY.store(true, Ordering::Release);
-}
-
-pub(crate) fn get_metadata_read_version_coalescing_service_ready() -> bool {
-    GET_METADATA_READ_VERSION_COALESCING_SERVICE_READY.load(Ordering::Acquire)
-}
-
 // Global singletons for backward compatibility with MinIO port.
 // These should be migrated to AppContext over time.
 // See issue #730 for migration plan.
@@ -160,10 +160,6 @@ pub struct InstanceContext {
    /// workers (scanner/heal/tier/lifecycle) without touching another instance.
    /// Replaces the process-global cancel-token static.
    background_cancel_token: OnceLock<CancellationToken>,
-    /// Serializes decommission data-movement operations with cancellation and
-    /// a subsequent restart. Readers are held across one object side effect;
-    /// the transition path takes the writer after cancelling the routine.
-    decommission_operation_gate: Arc<RwLock<()>>,
    /// Resolves object-encryption material at the application boundary.
    object_encryption_resolver: OnceLock<Arc<dyn ObjectEncryptionResolver>>,
    tier_delete_journal_recovery_stores: std::sync::Mutex<HashSet<Uuid>>,
@@ -204,7 +200,6 @@ impl InstanceContext {
            local_disk_set_drives: Arc::new(RwLock::new(Vec::new())),
            bucket_metadata_sys: std::sync::Mutex::new(None),
            background_cancel_token: OnceLock::new(),
-            decommission_operation_gate: Arc::new(RwLock::new(())),
            object_encryption_resolver: OnceLock::new(),
            tier_delete_journal_recovery_stores: std::sync::Mutex::new(HashSet::new()),
            transition_transaction_recovery_stores: std::sync::Mutex::new(HashSet::new()),
@@ -223,10 +218,6 @@ impl InstanceContext {
        self.lock_manager.clone()
    }

-    pub(crate) fn decommission_operation_gate(&self) -> Arc<RwLock<()>> {
-        Arc::clone(&self.decommission_operation_gate)
-    }
-
    /// Install the application-owned object-encryption resolver once.
    pub fn set_object_encryption_resolver(
        &self,
@@ -256,10 +256,6 @@ fn to_madmin_scanner_metrics(metrics: rustfs_common::metrics::ScannerMetricsRepo
        cycle_max_duration_seconds: metrics.cycle_max_duration_seconds,
        cycle_max_objects: metrics.cycle_max_objects,
        cycle_max_directories: metrics.cycle_max_directories,
-        cycle_timeout_total: metrics.cycle_timeout_total,
-        cycle_recovery_required_total: metrics.cycle_recovery_required_total,
-        cycle_last_progress_age: metrics.cycle_last_progress_age,
-        leader_lease_without_progress: metrics.leader_lease_without_progress,
        bitrot_cycle_enabled: metrics.bitrot_cycle_enabled,
        bitrot_cycle_seconds: metrics.bitrot_cycle_seconds,
        scan_checkpoint: metrics.scan_checkpoint.map(|checkpoint| MadminScannerCheckpointReport {
@@ -615,10 +611,6 @@ mod test {
            current_started: chrono_to_jiff_timestamp(current_started),
            last_cycle_partial_source: "usage".to_string(),
            last_cycle_partial_source_code: 1,
-            cycle_timeout_total: 3,
-            cycle_recovery_required_total: 2,
-            cycle_last_progress_age: 17,
-            leader_lease_without_progress: true,
            partial_cycles_by_source: vec![rustfs_common::metrics::ScannerSourceCycleSnapshot {
                source: "usage".to_string(),
                cycles: 2,
@@ -630,10 +622,6 @@ mod test {
        assert_eq!(scanner.current_started, chrono_to_jiff_timestamp(current_started));
        assert_eq!(scanner.last_cycle_partial_source, "usage");
        assert_eq!(scanner.last_cycle_partial_source_code, 1);
-        assert_eq!(scanner.cycle_timeout_total, 3);
-        assert_eq!(scanner.cycle_recovery_required_total, 2);
-        assert_eq!(scanner.cycle_last_progress_age, 17);
-        assert!(scanner.leader_lease_without_progress);
        let usage = scanner
            .partial_cycles_by_source
            .iter()
@@ -334,7 +334,6 @@ impl ECStore {
                            lifecycle_guard: bucket_incarnation_fence
                                .as_ref()
                                .and_then(|guard| guard.namespace_lock_guard()),
-                            ..Default::default()
                        },
                        "rebalance",
                    ),
@@ -53,12 +53,11 @@ use crate::diagnostics::get::{
    GetObjectFailureReason, classify_disk_error, get_stage_timer_if_enabled, record_get_object_pipeline_failure,
    record_get_object_pipeline_failure_for_path, record_get_stage_duration_if_enabled,
 };
-use crate::disk::disk_store::{DiskStoreRenameDataExt, get_drive_metadata_timeout};
+use crate::disk::disk_store::DiskStoreRenameDataExt;
 use crate::disk::local::DELETE_DATA_DIR_MARKER_PREFIX;
 use crate::disk::{
-    BATCH_READ_VERSION_MAX_ITEMS, BatchReadVersionItem, BatchReadVersionReq, BatchReadVersionResp, DataDirDeleteStatus, Disk,
-    OldCurrentSize, PART_TRANSACTION_NEW_META, PART_TRANSACTION_OLD_META, PART_TRANSACTION_ROLLBACK, PartTransactionAction,
-    STORAGE_FORMAT_FILE_BACKUP, part_transaction_path,
+    DataDirDeleteStatus, OldCurrentSize, PART_TRANSACTION_NEW_META, PART_TRANSACTION_OLD_META, PART_TRANSACTION_ROLLBACK,
+    PartTransactionAction, STORAGE_FORMAT_FILE_BACKUP, part_transaction_path,
 };
 use crate::erasure::coding::BitrotReader;
 use crate::io_support::bitrot::ShardReader;
@@ -76,7 +75,7 @@ use std::{
    future::Future,
    pin::Pin,
    sync::{
-        Arc, OnceLock,
+        OnceLock,
        atomic::{AtomicUsize, Ordering},
    },
    task::{Context, Poll},
@@ -95,242 +94,6 @@ fn metadata_distribution_key(bucket: &str, object: &str) -> String {
    [bucket, object].join("/")
 }

-fn read_version_coalescing_enabled() -> bool {
-    let enabled = || {
-        rustfs_utils::get_env_opt_str(ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE)
-            .is_some_and(|value| value.eq_ignore_ascii_case("auto") || value.eq_ignore_ascii_case("on"))
-    };
-
-    #[cfg(test)]
-    {
-        enabled()
-    }
-
-    #[cfg(not(test))]
-    {
-        static ENABLED: OnceLock<bool> = OnceLock::new();
-        *ENABLED.get_or_init(enabled)
-    }
-}
-
-fn read_version_coalescing_delay() -> Duration {
-    #[cfg(test)]
-    {
-        let micros = rustfs_utils::get_env_u64(
-            ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE_DELAY_MICROS,
-            DEFAULT_GET_METADATA_READ_VERSION_COALESCE_DELAY_MICROS,
-        );
-        Duration::from_micros(micros)
-    }
-
-    #[cfg(not(test))]
-    {
-        static DELAY: OnceLock<Duration> = OnceLock::new();
-        *DELAY.get_or_init(|| {
-            Duration::from_micros(rustfs_utils::get_env_u64(
-                ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE_DELAY_MICROS,
-                DEFAULT_GET_METADATA_READ_VERSION_COALESCE_DELAY_MICROS,
-            ))
-        })
-    }
-}
-
-struct CoalescedReadVersionRequest {
-    item: BatchReadVersionItem,
-    tx: oneshot::Sender<disk::error::Result<FileInfo>>,
-}
-
-#[derive(Clone, Copy, Debug, Eq, Hash, PartialEq)]
-struct ReadVersionCoalescerKey {
-    disk: usize,
-    incl_free_versions: bool,
-    read_data: bool,
-    healing: bool,
-}
-
-impl ReadVersionCoalescerKey {
-    fn new(disk: &DiskStore, opts: &ReadOptions) -> Self {
-        Self {
-            disk: Arc::as_ptr(disk) as usize,
-            incl_free_versions: opts.incl_free_versions,
-            read_data: opts.read_data,
-            healing: opts.healing,
-        }
-    }
-}
-
-#[derive(Default)]
-struct ReadVersionCoalescer {
-    lanes: HashMap<ReadVersionCoalescerKey, Vec<CoalescedReadVersionRequest>>,
-}
-
-fn read_version_coalescer() -> &'static Mutex<ReadVersionCoalescer> {
-    static COALESCER: OnceLock<Mutex<ReadVersionCoalescer>> = OnceLock::new();
-    COALESCER.get_or_init(|| Mutex::new(ReadVersionCoalescer::default()))
-}
-
-fn record_read_version_coalescer_event(event: &'static str, item_count: usize) {
-    counter!(
-        METRIC_GET_METADATA_READ_VERSION_COALESCER_TOTAL,
-        "event" => event,
-        "item_count" => item_count.to_string()
-    )
-    .increment(1);
-}
-
-async fn read_version_via_coalescer(
-    disk: DiskStore,
-    org_bucket: &str,
-    bucket: &str,
-    object: &str,
-    version_id: &str,
-    opts: &ReadOptions,
-    allow_coalescing: bool,
-) -> disk::error::Result<FileInfo> {
-    if !allow_coalescing || !read_version_coalescing_enabled() {
-        return disk.read_version(org_bucket, bucket, object, version_id, opts).await;
-    }
-    if !matches!(disk.as_ref(), Disk::Remote(_)) {
-        record_read_version_coalescer_event("bypass_non_remote", 1);
-        return disk.read_version(org_bucket, bucket, object, version_id, opts).await;
-    }
-
-    let (tx, rx) = oneshot::channel();
-    let item = BatchReadVersionItem {
-        org_volume: org_bucket.to_string(),
-        volume: bucket.to_string(),
-        path: object.to_string(),
-        version_id: version_id.to_string(),
-    };
-    let lane_key = ReadVersionCoalescerKey::new(&disk, opts);
-    let pending = {
-        let mut coalescer = read_version_coalescer().lock().await;
-        let lane = coalescer.lanes.entry(lane_key).or_default();
-        let schedule_delayed_flush = lane.is_empty();
-        lane.push(CoalescedReadVersionRequest { item, tx });
-        if lane.len() >= BATCH_READ_VERSION_MAX_ITEMS {
-            coalescer.lanes.remove(&lane_key)
-        } else if schedule_delayed_flush {
-            let disk = disk.clone();
-            let task_opts = *opts;
-            tokio::spawn(async move {
-                tokio::time::sleep(read_version_coalescing_delay()).await;
-                flush_read_version_coalescer_lane(lane_key, disk, task_opts).await;
-            });
-            None
-        } else {
-            None
-        }
-    };
-
-    if let Some(pending) = pending {
-        flush_read_version_coalescer_pending(lane_key, disk, *opts, pending).await;
-    }
-
-    rx.await
-        .unwrap_or_else(|_| Err(DiskError::other("coalesced read_version response channel closed")))
-}
-
-async fn flush_read_version_coalescer_lane(lane_key: ReadVersionCoalescerKey, disk: DiskStore, opts: ReadOptions) {
-    let pending = {
-        let mut coalescer = read_version_coalescer().lock().await;
-        coalescer.lanes.remove(&lane_key).unwrap_or_default()
-    };
-    flush_read_version_coalescer_pending(lane_key, disk, opts, pending).await;
-}
-
-async fn flush_read_version_coalescer_pending(
-    lane_key: ReadVersionCoalescerKey,
-    disk: DiskStore,
-    opts: ReadOptions,
-    pending: Vec<CoalescedReadVersionRequest>,
-) {
-    if pending.is_empty() {
-        return;
-    }
-
-    #[cfg(test)]
-    {
-        let mut observed_paths = HashSet::new();
-        for request in &pending {
-            if observed_paths.insert(request.item.path.as_str()) {
-                disk_call_counters::record(&request.item.path, disk_call_counters::KIND_BATCH_READ_VERSION, lane_key.disk);
-            }
-        }
-    }
-
-    let mut senders = Vec::with_capacity(pending.len());
-    let mut items = Vec::with_capacity(pending.len());
-    for request in pending {
-        senders.push(request.tx);
-        items.push(request.item);
-    }
-
-    let expected_items = items.clone();
-    record_read_version_coalescer_event("attempted_batch", items.len());
-    let result =
-        match tokio::time::timeout(get_drive_metadata_timeout(), disk.batch_read_version(BatchReadVersionReq { items, opts }))
-            .await
-        {
-            Ok(result) => result,
-            Err(_) => Err(DiskError::Timeout),
-        };
-    match result {
-        Ok(responses) => {
-            let results = map_batch_read_version_responses(&expected_items, responses);
-            for (tx, result) in senders.into_iter().zip(results) {
-                let _ = tx.send(result);
-            }
-        }
-        Err(err) => {
-            let message = err.to_string();
-            for tx in senders {
-                let _ = tx.send(Err(DiskError::other(message.clone())));
-            }
-        }
-    }
-}
-
-fn map_batch_read_version_responses(
-    expected_items: &[BatchReadVersionItem],
-    responses: Vec<BatchReadVersionResp>,
-) -> Vec<crate::disk::error::Result<FileInfo>> {
-    let mut results = (0..expected_items.len())
-        .map(|_| Err(DiskError::other("coalesced read_version response missing")))
-        .collect::<Vec<_>>();
-    let mut seen = vec![false; expected_items.len()];
-    for response in responses {
-        let Some(expected) = expected_items.get(response.index) else {
-            continue;
-        };
-        let Some(slot) = results.get_mut(response.index) else {
-            continue;
-        };
-        if seen[response.index] {
-            *slot = Err(DiskError::other("coalesced read_version response duplicate index"));
-            continue;
-        }
-        seen[response.index] = true;
-        if response.path != expected.path || response.version_id != expected.version_id {
-            *slot = Err(DiskError::other("coalesced read_version response identity mismatch"));
-        } else {
-            *slot = if response.success {
-                Ok(response.file_info)
-            } else {
-                Err(batch_read_version_response_error(response.error_code, response.error))
-            };
-        }
-    }
-    results
-}
-
-fn batch_read_version_response_error(error_code: u32, error: String) -> DiskError {
-    match DiskError::from_u32(error_code) {
-        Some(DiskError::Io(_)) | None => DiskError::other(error),
-        Some(error) => error,
-    }
-}
-
 pub(in crate::set_disk) fn bounded_metadata_fanout_order(
    bucket: &str,
    object: &str,
@@ -370,15 +133,11 @@ pub(in crate::set_disk) fn bounded_metadata_fanout_order(
    order
 }
 use tokio::io::{AsyncRead, ReadBuf};
-use tokio::sync::{Mutex, RwLock, oneshot};
+use tokio::sync::RwLock;
 use tokio::task::JoinSet;

 pub(in crate::set_disk) const EVENT_SET_DISK_READ: &str = "set_disk_read";
 pub(in crate::set_disk) const ENV_RUSTFS_GET_DATA_BLOCKS_FIRST_READER_SETUP: &str = "RUSTFS_GET_DATA_BLOCKS_FIRST_READER_SETUP";
-const ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE: &str = "RUSTFS_GET_METADATA_READ_VERSION_COALESCE";
-const ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE_DELAY_MICROS: &str = "RUSTFS_GET_METADATA_READ_VERSION_COALESCE_DELAY_MICROS";
-const DEFAULT_GET_METADATA_READ_VERSION_COALESCE_DELAY_MICROS: u64 = 200;
-const METRIC_GET_METADATA_READ_VERSION_COALESCER_TOTAL: &str = "rustfs_get_metadata_read_version_coalescer_total";
 pub(in crate::set_disk) const ENV_RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE: &str = "RUSTFS_PUT_RENAME_EARLY_ACK_ENABLE";
 /// Default reader-setup strategy for the GET read path (rustfs/backlog#1215,
 /// #1159, #923).
@@ -2597,7 +2356,6 @@ impl SetDisks {
            false,
            true,
            0,
-            false,
        )
        .await?;
        Ok((ress, errors))
@@ -2628,36 +2386,6 @@ impl SetDisks {
            true,
            caller_allows_early_stop,
            default_parity_count,
-            false,
-        )
-        .await
-    }
-
-    #[allow(clippy::too_many_arguments)]
-    pub(in crate::set_disk) async fn read_all_fileinfo_observed_for_get_object(
-        disks: &[Option<DiskStore>],
-        org_bucket: &str,
-        bucket: &str,
-        object: &str,
-        version_id: &str,
-        read_data: bool,
-        incl_free_versions: bool,
-        caller_allows_early_stop: bool,
-        default_parity_count: usize,
-    ) -> disk::error::Result<(Vec<FileInfo>, Vec<Option<DiskError>>, MetadataFanoutDiagnostics)> {
-        Self::read_all_fileinfo_inner(
-            disks,
-            org_bucket,
-            bucket,
-            object,
-            version_id,
-            read_data,
-            false,
-            incl_free_versions,
-            true,
-            caller_allows_early_stop,
-            default_parity_count,
-            true,
        )
        .await
    }
@@ -2680,7 +2408,6 @@ impl SetDisks {
        // subset would fail write quorum (backlog#872 regression).
        caller_allows_early_stop: bool,
        default_parity_count: usize,
-        allow_coalescing: bool,
    ) -> disk::error::Result<(Vec<FileInfo>, Vec<Option<DiskError>>, MetadataFanoutDiagnostics)> {
        let early_stop_enabled =
            caller_allows_early_stop && observe && (is_get_metadata_early_stop_enabled() || is_version_early_stop_enabled());
@@ -2697,7 +2424,6 @@ impl SetDisks {
                healing,
                incl_free_versions,
                default_parity_count,
-                allow_coalescing,
            )
            .await;
        }
@@ -2720,7 +2446,6 @@ impl SetDisks {
            healing,
            incl_free_versions,
            observe,
-            allow_coalescing,
        )
        .await
    }
@@ -2736,7 +2461,6 @@ impl SetDisks {
        healing: bool,
        incl_free_versions: bool,
        observe: bool,
-        allow_coalescing: bool,
    ) -> disk::error::Result<(Vec<FileInfo>, Vec<Option<DiskError>>, MetadataFanoutDiagnostics)> {
        let fanout_start = observe.then(Instant::now);
        let mut ress = Vec::with_capacity(disks.len());
@@ -2768,7 +2492,7 @@ impl SetDisks {
                    if let Some(delay) = slowtail_fault.as_ref().and_then(|fault| fault.delay_for_disk(disk_index)) {
                        tokio::time::sleep(delay).await;
                    }
-                    read_version_via_coalescer(disk, &org_bucket, &bucket, &object, &version_id, &task_opts, allow_coalescing)
+                    disk.read_version(&org_bucket, &bucket, &object, &version_id, &task_opts)
                        .await
                } else {
                    Err(DiskError::DiskNotFound)
@@ -2835,7 +2559,6 @@ impl SetDisks {
        healing: bool,
        incl_free_versions: bool,
        default_parity_count: usize,
-        allow_coalescing: bool,
    ) -> disk::error::Result<(Vec<FileInfo>, Vec<Option<DiskError>>, MetadataFanoutDiagnostics)> {
        let fanout_start = Instant::now();
        let mut ress = vec![FileInfo::default(); disks.len()];
@@ -2884,7 +2607,7 @@ impl SetDisks {
                        if let Some(delay) = slowtail_fault.as_ref().and_then(|fault| fault.delay_for_disk(index)) {
                            tokio::time::sleep(delay).await;
                        }
-                        read_version_via_coalescer(disk, &org_bucket, &bucket, &object, &version_id, &task_opts, allow_coalescing)
+                        disk.read_version(&org_bucket, &bucket, &object, &version_id, &task_opts)
                            .await
                    } else {
                        Err(DiskError::DiskNotFound)
@@ -6014,7 +5737,6 @@ pub(crate) mod disk_call_counters {

    /// Kind label for the per-disk `read_version` metadata RPC.
    pub const KIND_READ_VERSION: &str = "read_version";
-    pub const KIND_BATCH_READ_VERSION: &str = "batch_read_version";

    /// Registry key: (object, kind, disk_index).
    type CountKey = (String, String, usize);
@@ -6738,286 +6460,6 @@ mod tests {
        drop(dirs);
    }

-    #[tokio::test(flavor = "multi_thread", worker_threads = 4)]
-    async fn metadata_read_version_coalescer_bypasses_local_disks() {
-        const DISKS: usize = 4;
-        let bucket = "coalesced-read-version-local-bypass-bucket";
-        let object_a = "coalesced-local-object-a";
-        let object_b = "coalesced-local-object-b";
-        let (dirs, disks) = call_counter_local_disks(bucket, DISKS).await;
-        install_metadata_fanout_fileinfo(&disks, bucket, object_a, None).await;
-        install_metadata_fanout_fileinfo(&disks, bucket, object_b, None).await;
-
-        temp_env::async_with_vars(
-            [
-                (ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE, Some("auto")),
-                (ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE_DELAY_MICROS, Some("5000")),
-            ],
-            async {
-                let calls = disk_call_counters::observe(object_a);
-                let disks_a = disks.clone();
-                let disks_b = disks.clone();
-                let read_a = tokio::spawn(async move {
-                    SetDisks::read_all_fileinfo_observed_for_get_object(
-                        &disks_a, "", bucket, object_a, "", false, false, false, 2,
-                    )
-                    .await
-                    .map(|(file_infos, errors, _)| (file_infos, errors))
-                });
-                tokio::task::yield_now().await;
-                let read_b = tokio::spawn(async move {
-                    SetDisks::read_all_fileinfo_observed_for_get_object(
-                        &disks_b, "", bucket, object_b, "", false, false, false, 2,
-                    )
-                    .await
-                    .map(|(file_infos, errors, _)| (file_infos, errors))
-                });
-
-                let (metadata_a, errs_a) = read_a
-                    .await
-                    .expect("first read task should not panic")
-                    .expect("first coalesced read should resolve");
-                let (metadata_b, errs_b) = read_b
-                    .await
-                    .expect("second read task should not panic")
-                    .expect("second coalesced read should resolve");
-
-                assert_eq!(metadata_a.iter().filter(|fi| fi.name == object_a).count(), DISKS);
-                assert_eq!(metadata_b.iter().filter(|fi| fi.name == object_b).count(), DISKS);
-                assert!(errs_a.iter().all(Option::is_none));
-                assert!(errs_b.iter().all(Option::is_none));
-                assert_eq!(
-                    calls.total(disk_call_counters::KIND_READ_VERSION),
-                    DISKS as u64,
-                    "local disks still execute the ordinary per-disk read_version path"
-                );
-                assert_eq!(
-                    calls.total(disk_call_counters::KIND_BATCH_READ_VERSION),
-                    0,
-                    "GET coalescing targets internode RPC count only and must not batch local disk reads"
-                );
-            },
-        )
-        .await;
-
-        drop(dirs);
-    }
-
-    #[tokio::test]
-    async fn metadata_read_version_coalescer_requires_get_object_intent() {
-        const DISKS: usize = 4;
-        let bucket = "coalesced-read-version-default-bypass-bucket";
-        let object = "default-bypass-object";
-        let (dirs, disks) = call_counter_local_disks(bucket, DISKS).await;
-        install_metadata_fanout_fileinfo(&disks, bucket, object, None).await;
-
-        temp_env::async_with_vars([(ENV_RUSTFS_GET_METADATA_READ_VERSION_COALESCE, Some("auto"))], async {
-            let calls = disk_call_counters::observe(object);
-            let (metadata, errs) = SetDisks::read_all_fileinfo(&disks, "", bucket, object, "", false, false, false)
-                .await
-                .expect("default metadata read should resolve");
-
-            assert_eq!(metadata.iter().filter(|fi| fi.name == object).count(), DISKS);
-            assert!(errs.iter().all(Option::is_none));
-            assert_eq!(calls.total(disk_call_counters::KIND_READ_VERSION), DISKS as u64);
-            assert_eq!(
-                calls.total(disk_call_counters::KIND_BATCH_READ_VERSION),
-                0,
-                "non-GET metadata paths must bypass coalescer even when the env gate is enabled"
-            );
-        })
-        .await;
-
-        drop(dirs);
-    }
-
-    #[test]
-    fn batch_read_version_response_mapping_preserves_index_and_errors() {
-        let expected_items = vec![
-            BatchReadVersionItem {
-                org_volume: String::new(),
-                volume: "bucket".to_string(),
-                path: "object-a".to_string(),
-                version_id: "v-a".to_string(),
-            },
-            BatchReadVersionItem {
-                org_volume: String::new(),
-                volume: "bucket".to_string(),
-                path: "object-b".to_string(),
-                version_id: "v-b".to_string(),
-            },
-            BatchReadVersionItem {
-                org_volume: String::new(),
-                volume: "bucket".to_string(),
-                path: "object-c".to_string(),
-                version_id: "v-c".to_string(),
-            },
-        ];
-        let ok_file_info = FileInfo {
-            name: "object-a".to_string(),
-            ..Default::default()
-        };
-        let responses = vec![
-            BatchReadVersionResp {
-                index: 2,
-                path: "object-c".to_string(),
-                version_id: "v-c".to_string(),
-                success: false,
-                file_info: FileInfo::default(),
-                error: "disk read failed".to_string(),
-                error_code: 0,
-            },
-            BatchReadVersionResp {
-                index: 0,
-                path: "object-a".to_string(),
-                version_id: "v-a".to_string(),
-                success: true,
-                file_info: ok_file_info,
-                error: String::new(),
-                error_code: 0,
-            },
-        ];
-
-        let mut results = map_batch_read_version_responses(&expected_items, responses).into_iter();
-        let first = results
-            .next()
-            .expect("slot 0 should exist")
-            .expect("slot 0 should map the success response by index");
-        assert_eq!(first.name, "object-a");
-
-        let missing = results
-            .next()
-            .expect("slot 1 should exist")
-            .expect_err("slot 1 should stay missing");
-        assert!(
-            missing.to_string().contains("response missing"),
-            "unexpected missing response error: {missing}"
-        );
-
-        let failed = results
-            .next()
-            .expect("slot 2 should exist")
-            .expect_err("slot 2 should map the response error");
-        assert!(failed.to_string().contains("disk read failed"), "unexpected per-item error: {failed}");
-        assert!(results.next().is_none());
-    }
-
-    #[test]
-    fn batch_read_version_response_mapping_preserves_typed_not_found_errors() {
-        let expected_items = vec![
-            BatchReadVersionItem {
-                org_volume: String::new(),
-                volume: "bucket".to_string(),
-                path: "object-a".to_string(),
-                version_id: "v-a".to_string(),
-            },
-            BatchReadVersionItem {
-                org_volume: String::new(),
-                volume: "bucket".to_string(),
-                path: "object-b".to_string(),
-                version_id: "v-b".to_string(),
-            },
-        ];
-        let results = map_batch_read_version_responses(
-            &expected_items,
-            vec![
-                BatchReadVersionResp {
-                    index: 0,
-                    path: "object-a".to_string(),
-                    version_id: "v-a".to_string(),
-                    success: false,
-                    file_info: FileInfo::default(),
-                    error: DiskError::FileNotFound.to_string(),
-                    error_code: DiskError::FileNotFound.to_u32(),
-                },
-                BatchReadVersionResp {
-                    index: 1,
-                    path: "object-b".to_string(),
-                    version_id: "v-b".to_string(),
-                    success: false,
-                    file_info: FileInfo::default(),
-                    error: DiskError::FileVersionNotFound.to_string(),
-                    error_code: DiskError::FileVersionNotFound.to_u32(),
-                },
-            ],
-        );
-
-        assert!(matches!(results.first().expect("slot 0 should exist"), Err(DiskError::FileNotFound)));
-        assert!(matches!(
-            results.get(1).expect("slot 1 should exist"),
-            Err(DiskError::FileVersionNotFound)
-        ));
-    }
-
-    #[test]
-    fn batch_read_version_response_mapping_rejects_identity_mismatch_and_duplicate_index() {
-        let expected_items = vec![BatchReadVersionItem {
-            org_volume: String::new(),
-            volume: "bucket".to_string(),
-            path: "object-a".to_string(),
-            version_id: "v-a".to_string(),
-        }];
-        let mismatched = map_batch_read_version_responses(
-            &expected_items,
-            vec![BatchReadVersionResp {
-                index: 0,
-                path: "object-b".to_string(),
-                version_id: "v-a".to_string(),
-                success: true,
-                file_info: FileInfo {
-                    name: "object-b".to_string(),
-                    ..Default::default()
-                },
-                error: String::new(),
-                error_code: 0,
-            }],
-        )
-        .pop()
-        .expect("slot 0 should exist")
-        .expect_err("identity mismatch should fail closed");
-        assert!(
-            mismatched.to_string().contains("identity mismatch"),
-            "unexpected mismatch error: {mismatched}"
-        );
-
-        let duplicate = map_batch_read_version_responses(
-            &expected_items,
-            vec![
-                BatchReadVersionResp {
-                    index: 0,
-                    path: "object-a".to_string(),
-                    version_id: "v-a".to_string(),
-                    success: true,
-                    file_info: FileInfo {
-                        name: "object-a".to_string(),
-                        ..Default::default()
-                    },
-                    error: String::new(),
-                    error_code: 0,
-                },
-                BatchReadVersionResp {
-                    index: 0,
-                    path: "object-a".to_string(),
-                    version_id: "v-a".to_string(),
-                    success: true,
-                    file_info: FileInfo {
-                        name: "object-a".to_string(),
-                        ..Default::default()
-                    },
-                    error: String::new(),
-                    error_code: 0,
-                },
-            ],
-        )
-        .pop()
-        .expect("slot 0 should exist")
-        .expect_err("duplicate response index should fail closed");
-        assert!(
-            duplicate.to_string().contains("duplicate index"),
-            "unexpected duplicate error: {duplicate}"
-        );
-    }
-
    /// Isolation guard: unobserved objects record nothing (so parallel tests do
    /// not inflate one another), and a scope clears its own counts on drop.
    #[tokio::test]
@@ -97,7 +97,7 @@ use crate::storage_api_contracts::{
        CompletePart, ListMultipartsInfo, ListPartsInfo, MultipartInfo, MultipartOperations as _, MultipartUploadResult, PartInfo,
    },
    namespace::NamespaceLocking as _,
-    object::{DeleteAccounting, DeletedObject, HTTPPreconditions, ObjectIO as _, ObjectOperations as _, ObjectToDelete},
+    object::{DeletedObject, HTTPPreconditions, ObjectIO as _, ObjectOperations as _, ObjectToDelete},
    range::HTTPRangeSpec,
 };
 use crate::store::utils::is_reserved_or_invalid_bucket;
@@ -735,12 +735,8 @@ pub(crate) use core::io_primitives::disk_call_counters;
 mod ctx;
 mod metadata;
 mod ops;
-#[cfg(test)]
-pub(crate) use ops::multipart::NewMultipartUploadCommitObservation;
 #[cfg(any(test, feature = "test-util"))]
 pub use ops::multipart::{MultipartCommitBarrier, MultipartCommitPause};
-#[cfg(test)]
-pub(crate) use ops::object::DeleteObjectCommitBarrier;
 #[cfg(feature = "test-util")]
 pub(crate) use ops::object::TransitionCleanupStoreBarrier as SetDiskTransitionCleanupStoreBarrier;
 pub(crate) use ops::object::body_cache_plaintext_len;
@@ -3029,16 +3025,6 @@ pub struct SetDisks {
    storage_class_config_override: Arc<std::sync::RwLock<Option<Arc<storageclass::Config>>>>,
 }

-// DistributedLock sends the raw ObjectKey to its clients; LockRegistry clones
-// each endpoint's canonical Arc, so an exact Arc set identifies the lock domain.
-pub(crate) fn same_distributed_lock_domain(left: &[Arc<dyn LockClient>], right: &[Arc<dyn LockClient>]) -> bool {
-    left.iter()
-        .all(|left_client| right.iter().any(|right_client| Arc::ptr_eq(left_client, right_client)))
-        && right
-            .iter()
-            .all(|right_client| left.iter().any(|left_client| Arc::ptr_eq(left_client, right_client)))
-}
-
 const ERASURE_CACHE_MAX_ENTRIES: usize = 32;

 #[derive(Clone, Copy, Debug, Eq, Hash, PartialEq)]
@@ -3614,15 +3600,6 @@ impl SetDisks {
        &self.ctx
    }

-    /// Whether both sets' namespace-lock implementations cover the same object key.
-    pub(crate) async fn shares_namespace_lock_domain(&self, other: &Self) -> bool {
-        match (self.ctx.is_dist_erasure().await, other.ctx.is_dist_erasure().await) {
-            (false, false) => Arc::ptr_eq(&self.local_lock_manager, &other.local_lock_manager),
-            (true, true) => same_distributed_lock_domain(&self.lockers, &other.lockers),
-            _ => false,
-        }
-    }
-
    /// The lock manager this set actually uses (test-only; Phase 5 Slice 3).
    #[cfg(test)]
    pub(crate) fn local_lock_manager_for_test(&self) -> &Arc<rustfs_lock::GlobalLockManager> {
@@ -4607,11 +4584,11 @@ fn should_preserve_delete_replication_state(opts: &ObjectOptions) -> bool {
 }

 fn should_force_delete_marker_for_missing_version(opts: &ObjectOptions) -> bool {
-    opts.delete_marker || ((opts.versioned || opts.version_suspended) && opts.version_id.is_none() && !opts.data_movement)
+    opts.delete_marker || (opts.versioned && opts.version_id.is_none() && !opts.data_movement)
 }

 fn resolve_delete_version_state(opts: &ObjectOptions, goi: &ObjectInfo, version_found: bool) -> (bool, bool) {
-    let mut mark_delete = goi.version_id.is_some() || ((opts.versioned || opts.version_suspended) && opts.version_id.is_none());
+    let mut mark_delete = goi.version_id.is_some() || (opts.versioned && opts.version_id.is_none());
    let mut delete_marker = opts.versioned;

    if opts.version_id.is_some() {
@@ -32,8 +32,6 @@ use crate::crash_inject::{self, CrashPoint};
 use crate::multipart_listing::paginate_multipart_listing;
 use futures::{StreamExt, stream};
 use std::future::Future;
-#[cfg(test)]
-use std::sync::atomic::AtomicBool;
 #[cfg(any(test, feature = "test-util"))]
 use std::sync::atomic::{AtomicUsize, Ordering};
 use std::time::Duration;
@@ -67,7 +65,6 @@ impl StaleMultipartCleanupGuard {
 #[cfg(any(test, feature = "test-util"))]
 #[derive(Clone, Copy, PartialEq, Eq)]
 pub enum MultipartCommitPause {
-    NewUploadBeforeLockLost,
    PutPartBeforeLockAcquire,
    PutPartBeforeLockLost,
    PutPartAfterRename,
@@ -159,72 +156,6 @@ impl Drop for MultipartCommitBarrier {
    }
 }

-#[cfg(test)]
-struct NewMultipartUploadCommitObservationState {
-    bucket: String,
-    object: String,
-    committed: AtomicBool,
-}
-
-#[cfg(test)]
-pub(crate) struct NewMultipartUploadCommitObservation {
-    state: Arc<NewMultipartUploadCommitObservationState>,
-}
-
-#[cfg(test)]
-static NEW_MULTIPART_UPLOAD_COMMIT_OBSERVATION: std::sync::OnceLock<
-    std::sync::Mutex<Option<Arc<NewMultipartUploadCommitObservationState>>>,
-> = std::sync::OnceLock::new();
-
-#[cfg(test)]
-impl NewMultipartUploadCommitObservation {
-    pub(crate) fn install(bucket: &str, object: &str) -> Self {
-        let state = Arc::new(NewMultipartUploadCommitObservationState {
-            bucket: bucket.to_string(),
-            object: object.to_string(),
-            committed: AtomicBool::new(false),
-        });
-        let mut slot = NEW_MULTIPART_UPLOAD_COMMIT_OBSERVATION
-            .get_or_init(|| std::sync::Mutex::new(None))
-            .lock()
-            .expect("new multipart upload commit observation mutex should not poison");
-        assert!(slot.is_none(), "new multipart upload commit observation must be unique");
-        *slot = Some(Arc::clone(&state));
-        Self { state }
-    }
-
-    pub(crate) fn committed(&self) -> bool {
-        self.state.committed.load(Ordering::Acquire)
-    }
-}
-
-#[cfg(test)]
-impl Drop for NewMultipartUploadCommitObservation {
-    fn drop(&mut self) {
-        let mut slot = NEW_MULTIPART_UPLOAD_COMMIT_OBSERVATION
-            .get_or_init(|| std::sync::Mutex::new(None))
-            .lock()
-            .expect("new multipart upload commit observation mutex should not poison");
-        if slot.as_ref().is_some_and(|state| Arc::ptr_eq(state, &self.state)) {
-            *slot = None;
-        }
-    }
-}
-
-#[cfg(test)]
-fn observe_new_multipart_upload_commit(bucket: &str, object: &str) {
-    let state = NEW_MULTIPART_UPLOAD_COMMIT_OBSERVATION
-        .get_or_init(|| std::sync::Mutex::new(None))
-        .lock()
-        .expect("new multipart upload commit observation mutex should not poison")
-        .as_ref()
-        .filter(|state| state.bucket == bucket && state.object == object)
-        .cloned();
-    if let Some(state) = state {
-        state.committed.store(true, Ordering::Release);
-    }
-}
-
 #[cfg(any(test, feature = "test-util"))]
 async fn pause_multipart_commit(bucket: &str, object: &str, pause: MultipartCommitPause) {
    let barrier = {
@@ -1684,30 +1615,6 @@ impl crate::storage_api_contracts::multipart::MultipartOperations for SetDisks {

        let upload_path = Self::get_multipart_upload_dir(bucket, object, upload_uuid.as_str(), opts.data_movement);

-        #[cfg(any(test, feature = "test-util"))]
-        pause_multipart_commit(bucket, object, MultipartCommitPause::NewUploadBeforeLockLost).await;
-        if _object_lock_guard.as_ref().is_some_and(|guard| guard.is_lock_lost()) {
-            return Err(StorageError::NamespaceLockQuorumUnavailable {
-                mode: "new_multipart_upload_commit",
-                bucket: bucket.to_string(),
-                object: object.to_string(),
-                required: 1,
-                achieved: 0,
-            });
-        }
-        if opts
-            .namespace_lock_fence
-            .as_ref()
-            .is_some_and(NamespaceLockFence::is_lock_lost)
-        {
-            return Err(StorageError::NamespaceLockQuorumUnavailable {
-                mode: "new_multipart_upload_outer_lock",
-                bucket: bucket.to_string(),
-                object: object.to_string(),
-                required: 1,
-                achieved: 0,
-            });
-        }
        ensure_multipart_bucket_lifecycle_lock_held(bucket, object, opts)?;
        Self::write_unique_file_info(
            &shuffle_disks,
@@ -1719,8 +1626,6 @@ impl crate::storage_api_contracts::multipart::MultipartOperations for SetDisks {
        )
        .await
        .map_err(|e| to_object_err(e.into(), vec![bucket, object]))?;
-        #[cfg(test)]
-        observe_new_multipart_upload_commit(bucket, object);

        // evalDisks

@@ -45,7 +45,6 @@ use crate::bucket::replication::{
    DeleteReplicationConfigSnapshot, ReplicationLifecycleBridge, ReplicationStatusType, VersionPurgeStatusType,
    replication_state_to_filemeta, replication_status_from_filemeta, version_purge_status_to_filemeta,
 };
-use crate::data_usage::quota_object_size;
 use crate::diagnostics::get::GetObjectFailureReason;
 use crate::disk::{DataDirDeleteStatus, OldCurrentSize};
 use crate::error::is_err_invalid_upload_id;
@@ -1294,7 +1293,7 @@ impl crate::storage_api_contracts::object::ObjectIO for SetDisks {
            (prepared.snapshot, prepared.object_info)
        } else {
            match self
-                .get_object_fileinfo_for_get_object_reader(
+                .get_object_fileinfo(
                    bucket,
                    object,
                    opts,
@@ -2123,27 +2122,14 @@ impl SetDisks {
            let erasure = Arc::new(erasure_from_file_info(&fi, false)?);

            let put_object_size = known_put_object_storage_size(data.size());
-            let shard_file_size_raw = erasure.shard_file_size(put_object_size);
            let is_inline_buffer =
-                storage_class_config.should_inline(shard_file_size_raw, erasure.data_shards, opts.versioned);
+                storage_class_config.should_inline(erasure.shard_file_size(put_object_size), erasure.data_shards, opts.versioned);

            let collect_stage_timing = rustfs_io_metrics::put_stage_metrics_enabled() || issue3031_diag_enabled();
-            let shard_file_size = shard_file_size_raw;
+            let shard_file_size = erasure.shard_file_size(put_object_size);
            let shard_size = erasure.shard_size();
            let write_path = classify_put_write_path(is_inline_buffer, put_object_size, fi.erasure.block_size);
            let direct_inline_commit = matches!(write_path, SmallWritePath::Inline);
-            {
-                use std::io::Write;
-                let msg = format!(
-                    "INLINE_DEBUG: bucket={} obj={} size={} shard_fs={} ds={} bs={} inline={} direct={} path={} iblock={} ver={}\n",
-                    bucket, object, put_object_size, shard_file_size_raw, erasure.data_shards, fi.erasure.block_size,
-                    is_inline_buffer, direct_inline_commit, write_path.metric_label(), storage_class_config.inline_block(), opts.versioned
-                );
-                if let Ok(mut f) = std::fs::OpenOptions::new().create(true).append(true).open("/tmp/rustfs_inline_debug.log") {
-                    let _ = f.write_all(msg.as_bytes());
-                }
-                let _ = std::io::stderr().write_all(msg.as_bytes());
-            }
            rustfs_io_metrics::record_put_object_path(write_path.metric_label());
            let writer_setup_stage_start = collect_stage_timing.then(Instant::now);
            let (mut writers, errors) = if direct_inline_commit {
@@ -2497,7 +2483,6 @@ impl SetDisks {
                        })
                        .await?,
                    );
-                    notify_put_object_commit_namespace_acquired(bucket, object);
                }
                #[cfg(not(any(test, feature = "test-util")))]
                {
@@ -4645,7 +4630,6 @@ struct PutObjectCommitBarrierState {
    arrived: tokio::sync::Notify,
    release: tokio::sync::Notify,
    namespace_pending: tokio::sync::Notify,
-    namespace_acquired: std::sync::atomic::AtomicBool,
 }

 #[cfg(any(test, feature = "test-util"))]
@@ -4667,7 +4651,6 @@ impl PutObjectCommitBarrier {
            arrived: tokio::sync::Notify::new(),
            release: tokio::sync::Notify::new(),
            namespace_pending: tokio::sync::Notify::new(),
-            namespace_acquired: std::sync::atomic::AtomicBool::new(false),
        });
        let mut slot = PUT_OBJECT_COMMIT_BARRIER
            .get_or_init(|| std::sync::Mutex::new(Vec::new()))
@@ -4702,10 +4685,6 @@ impl PutObjectCommitBarrier {
            .await
            .expect("put object should wait for the namespace lock after leaving the commit barrier");
    }
-
-    pub fn namespace_acquired(&self) -> bool {
-        self.state.namespace_acquired.load(std::sync::atomic::Ordering::Acquire)
-    }
 }

 #[cfg(any(test, feature = "test-util"))]
@@ -4762,22 +4741,6 @@ fn notify_put_object_commit_namespace_pending(bucket: &str, object: &str) {
    }
 }

-#[cfg(any(test, feature = "test-util"))]
-fn notify_put_object_commit_namespace_acquired(bucket: &str, object: &str) {
-    let barrier = PUT_OBJECT_COMMIT_BARRIER
-        .get_or_init(|| std::sync::Mutex::new(Vec::new()))
-        .lock()
-        .expect("put object commit barrier mutex should not poison")
-        .iter()
-        .find(|barrier| {
-            barrier.bucket == bucket && barrier.object == object && barrier.pause == PutObjectCommitPause::BeforeNamespace
-        })
-        .cloned();
-    if let Some(barrier) = barrier {
-        barrier.namespace_acquired.store(true, std::sync::atomic::Ordering::Release);
-    }
-}
-
 #[cfg(test)]
 struct DeleteObjectCommitBarrierState {
    bucket: String,
@@ -4787,7 +4750,7 @@ struct DeleteObjectCommitBarrierState {
 }

 #[cfg(test)]
-pub(crate) struct DeleteObjectCommitBarrier {
+struct DeleteObjectCommitBarrier {
    state: Arc<DeleteObjectCommitBarrierState>,
 }

@@ -4797,7 +4760,7 @@ static DELETE_OBJECT_COMMIT_BARRIER: std::sync::OnceLock<std::sync::Mutex<Option

 #[cfg(test)]
 impl DeleteObjectCommitBarrier {
-    pub(crate) fn install(bucket: &str, object: &str) -> Self {
+    fn install(bucket: &str, object: &str) -> Self {
        let state = Arc::new(DeleteObjectCommitBarrierState {
            bucket: bucket.to_string(),
            object: object.to_string(),
@@ -4813,13 +4776,13 @@ impl DeleteObjectCommitBarrier {
        Self { state }
    }

-    pub(crate) async fn wait_until_paused(&self) {
+    async fn wait_until_paused(&self) {
        tokio::time::timeout(Duration::from_secs(30), self.state.arrived.notified())
            .await
            .expect("delete object should reach the deterministic commit barrier");
    }

-    pub(crate) fn release(&self) {
+    fn release(&self) {
        self.state.release.notify_one();
    }
 }
@@ -5692,18 +5655,7 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {
        objects: Vec<ObjectToDelete>,
        opts: ObjectOptions,
    ) -> (Vec<DeletedObject>, Vec<Option<Error>>) {
-        let (deleted, errors, _) = self.delete_objects_with_accounting(bucket, objects, opts).await;
-        (deleted, errors)
-    }
-
-    async fn delete_objects_with_accounting(
-        &self,
-        bucket: &str,
-        objects: Vec<ObjectToDelete>,
-        opts: ObjectOptions,
-    ) -> (Vec<DeletedObject>, Vec<Option<Error>>, Vec<Option<DeleteAccounting>>) {
        let mut del_objects = vec![DeletedObject::default(); objects.len()];
-        let mut accounting = vec![None; objects.len()];
        let delete_config_snapshot = opts
            .delete_replication_config_snapshot
            .clone()
@@ -5793,7 +5745,7 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {
                                *item = Some(Error::other(message.clone()));
                            }
                        }
-                        return (del_objects, del_errs, accounting);
+                        return (del_objects, del_errs);
                    }
                },
            }
@@ -5840,22 +5792,6 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {
            let source_missing = gerr
                .as_ref()
                .is_some_and(|err| is_err_object_not_found(err) || is_err_version_not_found(err));
-            // Resolve accounting from the generation selected under this
-            // object's write lock. A request-layer pre-stat is only an
-            // optimization and cannot identify a concurrent overwrite.
-            let (accounting_size, accounting_version_id, removed_current_object) = if source_missing
-                || dobj.synthetic_version_id
-                || set_disk_delete_creates_delete_marker(&check_opts)
-                || goi.delete_marker
-            {
-                (None, None, false)
-            } else {
-                (
-                    quota_object_size(&goi).ok(),
-                    goi.version_id.filter(|version_id| !version_id.is_nil()),
-                    (dobj.version_id.is_none() || is_explicit_null_version(dobj.version_id)) && !dobj.synthetic_version_id,
-                )
-            };
            // Normalize both sides before comparing. `goi.version_id` is the
            // client-facing identity, where `from_file_info` synthesizes
            // `Some(Uuid::nil())` for a null version on a versioned or
@@ -5941,7 +5877,6 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {
            if dobj.version_id.is_none() && (version_suspended || versioned) {
                vr.mod_time = Some(OffsetDateTime::now_utc());
                vr.deleted = true;
-                vr.mark_deleted = true;
                if versioned {
                    vr.version_id = Some(Uuid::new_v4());
                }
@@ -5985,12 +5920,7 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {
                    },
                    replication_state: vr.replication_state_internal.clone(),
                    ..Default::default()
-                };
-                accounting[i] = Some(DeleteAccounting {
-                    size: accounting_size,
-                    version_id: accounting_version_id,
-                    removed_current_object,
-                });
+                }
            }

            // Only add to vers_map if we hold the lock
@@ -6036,7 +5966,7 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {
                    });
                }
            }
-            return (del_objects, del_errs, accounting);
+            return (del_objects, del_errs);
        }

        let mut persisted_journal_entries = Vec::with_capacity(journal_entries.len());
@@ -6274,16 +6204,7 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {
            }
        }

-        // An accounting identity is actionable only when the delete result is
-        // successful. Never let a failed commit (including a partial quorum
-        // failure) reach the request-layer fast delta path.
-        for (index, err) in del_errs.iter().enumerate() {
-            if err.is_some() {
-                accounting[index] = None;
-            }
-        }
-
-        (del_objects, del_errs, accounting)
+        (del_objects, del_errs)
    }

    #[tracing::instrument(skip(self))]
@@ -6612,12 +6533,6 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {

        let mut obj_info = ObjectInfo::from_file_info(&dfi, bucket, object, opts.versioned || opts.version_suspended);
        obj_info.size = goi.size;
-        // Keep the committed source metadata on the internal delete result so
-        // the request layer can derive canonical accounting for this exact
-        // generation. Delete responses do not expose these fields.
-        obj_info.actual_size = goi.actual_size;
-        obj_info.user_defined = Arc::clone(&goi.user_defined);
-        obj_info.parts = Arc::clone(&goi.parts);
        obj_info.user_tags = Arc::clone(&goi.user_tags);
        self.invalidate_get_object_metadata_cache(bucket, object).await;
        Ok(obj_info)
@@ -7909,113 +7824,6 @@ mod replication_quota_safety_tests {
        assert_eq!(stored.get_actual_size().expect("stored logical size should parse"), 1);
    }

-    #[tokio::test]
-    async fn delete_returns_canonical_compressed_accounting_size() {
-        let (_temp_dirs, disks, set_disks) = hermetic_set_disks(4).await;
-        let bucket = "compressed-delete-accounting";
-        for disk in &disks {
-            disk.make_volume(bucket).await.expect("bucket volume should be created");
-        }
-
-        let mut user_defined = HashMap::new();
-        insert_str(
-            &mut user_defined,
-            rustfs_utils::http::SUFFIX_COMPRESSION,
-            "klauspost/compress/s2".to_string(),
-        );
-        insert_str(&mut user_defined, SUFFIX_ACTUAL_SIZE, "1000".to_string());
-        let mut reader = PutObjReader::new(
-            HashReader::from_stream(Cursor::new(vec![0x5a; 400]), 400, 1000, None, None, false)
-                .expect("compressed fixture reader should be valid"),
-        );
-        set_disks
-            .put_object(
-                bucket,
-                "object",
-                &mut reader,
-                &ObjectOptions {
-                    user_defined,
-                    ..Default::default()
-                },
-            )
-            .await
-            .expect("compressed object should be written");
-
-        let (deleted, errors, accounting) = set_disks
-            .delete_objects_with_accounting(
-                bucket,
-                vec![ObjectToDelete {
-                    object_name: "object".to_string(),
-                    ..Default::default()
-                }],
-                ObjectOptions {
-                    object_lock_config_snapshot: Some(Arc::new(ObjectLockConfigSnapshot::new(
-                        ObjectLockConfigState::ConfirmedAbsent,
-                    ))),
-                    ..Default::default()
-                },
-            )
-            .await;
-
-        assert!(errors[0].is_none(), "compressed delete should succeed: {:?}", errors[0]);
-        assert!(deleted[0].found, "the committed object must be reported as found");
-        assert_eq!(accounting[0].as_ref().and_then(|value| value.size), Some(1000));
-        assert!(accounting[0].as_ref().is_some_and(|value| value.version_id.is_none()));
-        assert!(accounting[0].as_ref().is_some_and(|value| value.removed_current_object));
-    }
-
-    #[tokio::test]
-    async fn suspended_delete_marker_does_not_return_body_accounting() {
-        let (_temp_dirs, disks, set_disks) = hermetic_set_disks(4).await;
-        let bucket = "suspended-delete-accounting";
-        for disk in &disks {
-            disk.make_volume(bucket).await.expect("bucket volume should be created");
-        }
-
-        let mut user_defined = HashMap::new();
-        insert_str(
-            &mut user_defined,
-            rustfs_utils::http::SUFFIX_COMPRESSION,
-            "klauspost/compress/s2".to_string(),
-        );
-        insert_str(&mut user_defined, SUFFIX_ACTUAL_SIZE, "1000".to_string());
-        let mut reader = PutObjReader::new(
-            HashReader::from_stream(Cursor::new(vec![0x5a; 400]), 400, 1000, None, None, false)
-                .expect("compressed fixture reader should be valid"),
-        );
-        let suspended_opts = ObjectOptions {
-            version_suspended: true,
-            delete_replication_config_snapshot: Some(Arc::new(DeleteReplicationConfigSnapshot::from_configs_for_test(
-                s3s::dto::VersioningConfiguration {
-                    status: Some(s3s::dto::BucketVersioningStatus::from_static(s3s::dto::BucketVersioningStatus::SUSPENDED)),
-                    ..Default::default()
-                },
-                None,
-            ))),
-            user_defined,
-            object_lock_config_snapshot: Some(Arc::new(ObjectLockConfigSnapshot::new(ObjectLockConfigState::ConfirmedAbsent))),
-            ..Default::default()
-        };
-        set_disks
-            .put_object(bucket, "object", &mut reader, &suspended_opts)
-            .await
-            .expect("compressed object should be written");
-
-        let (deleted, errors, accounting) = set_disks
-            .delete_objects_with_accounting(
-                bucket,
-                vec![ObjectToDelete {
-                    object_name: "object".to_string(),
-                    ..Default::default()
-                }],
-                suspended_opts,
-            )
-            .await;
-        assert!(errors[0].is_none(), "suspended delete should create a marker: {:?}", errors[0]);
-        assert!(deleted[0].delete_marker);
-        assert!(accounting[0].is_none(), "a delete marker must not carry body accounting");
-    }
-
    #[tokio::test]
    async fn direct_put_cannot_persist_a_tiny_logical_size() {
        let (_temp_dirs, disks, set_disks) = hermetic_set_disks(4).await;
@@ -11828,7 +11636,6 @@ mod transition_upload_integrity_tests {
                crate::data_movement::SourceCleanupBucketFence {
                    expected_incarnation_id: None,
                    lifecycle_guard: Some(&bucket_guard),
-                    ..Default::default()
                },
                "test_data_movement",
            )
@@ -259,33 +259,10 @@ impl SetDisks {
        read_data: bool,
        caller_allows_early_stop: bool,
    ) -> Result<GetObjectFileInfo> {
-        self.get_object_fileinfo_gated_inner(bucket, object, opts, read_data, caller_allows_early_stop, false)
+        self.get_object_fileinfo_gated(bucket, object, opts, read_data, caller_allows_early_stop)
            .await
    }

-    #[tracing::instrument(level = "debug", skip(self))]
-    #[hotpath::measure(impl_type = "SetDisks")]
-    pub(super) async fn get_object_fileinfo_for_get_object_reader(
-        &self,
-        bucket: &str,
-        object: &str,
-        opts: &ObjectOptions,
-        read_data: bool,
-        caller_allows_early_stop: bool,
-    ) -> Result<GetObjectFileInfo> {
-        let allow_read_version_coalescing = !crate::bucket::utils::is_meta_bucketname(bucket)
-            && crate::runtime::global::get_metadata_read_version_coalescing_service_ready();
-        self.get_object_fileinfo_gated_inner(
-            bucket,
-            object,
-            opts,
-            read_data,
-            caller_allows_early_stop,
-            allow_read_version_coalescing,
-        )
-        .await
-    }
-
    /// Like `get_object_fileinfo`, but `allow_early_stop=false` forces the full
    /// quorum fanout. Read-before-write callers (object tagging) must use this:
    /// the returned online-disk set is the write target, and the early-stop
@@ -298,20 +275,6 @@ impl SetDisks {
        opts: &ObjectOptions,
        read_data: bool,
        allow_early_stop: bool,
-    ) -> Result<GetObjectFileInfo> {
-        self.get_object_fileinfo_gated_inner(bucket, object, opts, read_data, allow_early_stop, false)
-            .await
-    }
-
-    #[allow(clippy::too_many_arguments)]
-    async fn get_object_fileinfo_gated_inner(
-        &self,
-        bucket: &str,
-        object: &str,
-        opts: &ObjectOptions,
-        read_data: bool,
-        allow_early_stop: bool,
-        allow_read_version_coalescing: bool,
    ) -> Result<GetObjectFileInfo> {
        let vid = opts.version_id.clone().unwrap_or_default();
        let stage_metrics_enabled = rustfs_io_metrics::get_stage_metrics_enabled();
@@ -374,34 +337,19 @@ impl SetDisks {
        // read_all_fileinfo_observed (see read_all_fileinfo_early_stop in
        // core/io_primitives.rs); unsafe requests and callers that opt out
        // (allow_early_stop=false) fall back to full-wait.
-        let (mut parts_metadata, errs, metadata_fanout_diagnostics) = if allow_read_version_coalescing {
-            Self::read_all_fileinfo_observed_for_get_object(
-                &disks,
-                "",
-                bucket,
-                object,
-                vid.as_str(),
-                read_data,
-                opts.incl_free_versions,
-                allow_early_stop,
-                self.default_parity_count,
-            )
-            .await?
-        } else {
-            Self::read_all_fileinfo_observed(
-                &disks,
-                "",
-                bucket,
-                object,
-                vid.as_str(),
-                read_data,
-                false,
-                opts.incl_free_versions,
-                allow_early_stop,
-                self.default_parity_count,
-            )
-            .await?
-        };
+        let (mut parts_metadata, errs, metadata_fanout_diagnostics) = Self::read_all_fileinfo_observed(
+            &disks,
+            "",
+            bucket,
+            object,
+            vid.as_str(),
+            read_data,
+            false,
+            opts.incl_free_versions,
+            allow_early_stop,
+            self.default_parity_count,
+        )
+        .await?;
        let metadata_metrics_path = if crate::bucket::utils::is_meta_bucketname(bucket) {
            GET_OBJECT_PATH_INTERNAL_META
        } else {
@@ -62,8 +62,8 @@ pub(crate) mod object {
    use super::{Debug, Error, FileInfo, GetObjectReader, ObjectInfo, ObjectOptions, PutObjReader};
    use crate::storage_api_contracts::range::HTTPRangeSpec;
    pub(crate) use rustfs_storage_api::{
-        DeleteAccounting, DeletedObject, HTTPPreconditions, ObjectIO, ObjectLockDeleteOptions, ObjectLockRetentionOptions,
-        ObjectOperations, ObjectPreconditionError, ObjectPreconditionPart, ObjectPreconditionState, ObjectToDelete,
+        DeletedObject, HTTPPreconditions, ObjectIO, ObjectLockDeleteOptions, ObjectLockRetentionOptions, ObjectOperations,
+        ObjectPreconditionError, ObjectPreconditionPart, ObjectPreconditionState, ObjectToDelete,
    };

    pub(crate) trait EcstoreObjectIO:
@@ -297,16 +297,10 @@ impl ECStore {
 #[cfg(test)]
 mod tests {
    use super::*;
-    use crate::bucket::metadata_sys;
    use crate::core::pools::{PoolDecommissionInfo, PoolStatus};
-    use crate::disk::{DeleteOptions, DiskOption, format::FormatV3, new_disk};
-    use crate::layout::endpoints::{EndpointServerPools, Endpoints, PoolEndpoints};
-    use crate::runtime::instance::InstanceContext;
-    use crate::storage_api_contracts::bucket::{BucketOperations, MakeBucketOptions};
-    use crate::storage_api_contracts::object::{ObjectIO as _, ObjectOperations};
+    use crate::disk::{DiskOption, format::FormatV3, new_disk};
+    use crate::layout::endpoints::{Endpoints, PoolEndpoints};
    use crate::store::init_format::{load_format_erasure, save_format_file};
-    use crate::store::init_local_disks_with_instance_ctx;
-    use tokio_util::sync::CancellationToken;

    async fn minimal_heal_pool(pool_idx: usize) -> Arc<Sets> {
        let format = FormatV3::new(1, 1);
@@ -353,51 +347,6 @@ mod tests {
        }
    }

-    async fn multi_pool_heal_store() -> (tempfile::TempDir, Arc<ECStore>, CancellationToken) {
-        let temp_dir = tempfile::tempdir().expect("multi-pool heal test directory should be created");
-        let mut pool_endpoints = Vec::new();
-        for pool_index in 0..2 {
-            let mut endpoints = Vec::new();
-            for disk_index in 0..4 {
-                let disk_path = temp_dir.path().join(format!("pool{pool_index}-disk{disk_index}"));
-                tokio::fs::create_dir_all(&disk_path)
-                    .await
-                    .expect("multi-pool heal test disk should be created");
-                let mut endpoint = Endpoint::try_from(disk_path.to_str().expect("disk path should be utf8"))
-                    .expect("test endpoint should parse");
-                endpoint.set_pool_index(pool_index);
-                endpoint.set_set_index(0);
-                endpoint.set_disk_index(disk_index);
-                endpoints.push(endpoint);
-            }
-            pool_endpoints.push(PoolEndpoints {
-                legacy: false,
-                set_count: 1,
-                drives_per_set: 4,
-                endpoints: Endpoints::from(endpoints),
-                cmd_line: format!("heal-owner-pool-{pool_index}"),
-                platform: "test".to_string(),
-            });
-        }
-
-        let endpoint_pools = EndpointServerPools::from(pool_endpoints);
-        let instance_ctx = Arc::new(InstanceContext::new());
-        init_local_disks_with_instance_ctx(&instance_ctx, endpoint_pools.clone())
-            .await
-            .expect("multi-pool local disks should initialize");
-        let shutdown = CancellationToken::new();
-        let store = ECStore::new_with_instance_ctx(
-            "127.0.0.1:0".parse().expect("test address should parse"),
-            endpoint_pools,
-            shutdown.clone(),
-            instance_ctx,
-        )
-        .await
-        .expect("multi-pool test store should initialize");
-        metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
-        (temp_dir, store, shutdown)
-    }
-
    #[tokio::test]
    async fn heal_object_pool_scope_selects_only_requested_pool() {
        let store = minimal_heal_store().await;
@@ -557,229 +506,6 @@ mod tests {
        }
    }

-    #[tokio::test]
-    #[serial_test::serial]
-    async fn unscoped_heal_object_suspended_owner_semantics() {
-        let (_temp_dir, store, shutdown) = multi_pool_heal_store().await;
-        let bucket = format!("heal-owner-{}", Uuid::new_v4().simple());
-        let active_object = "active-owner";
-        let suspended_only_object = "suspended-only";
-        let duplicate_object = "duplicate-owner";
-        let marker_object = "marker-owner";
-        let quorum_object = "quorum-owner";
-        store
-            .make_bucket(&bucket, &MakeBucketOptions::default())
-            .await
-            .expect("bucket should be created in all pools");
-
-        let mut active_reader = PutObjReader::from_vec(b"active owner".to_vec());
-        store.pools[0]
-            .put_object(&bucket, active_object, &mut active_reader, &ObjectOptions::default())
-            .await
-            .expect("active owner object should be written");
-        let active_disks = store.pools[0].disk_set[0].disks.read().await.clone();
-        let missing_active_disk = active_disks[0].clone().expect("active disk should be online");
-        missing_active_disk
-            .delete(
-                &bucket,
-                active_object,
-                DeleteOptions {
-                    recursive: true,
-                    immediate: true,
-                    ..Default::default()
-                },
-            )
-            .await
-            .expect("active owner shard should be removed for repair");
-        assert!(
-            missing_active_disk.read_xl(&bucket, active_object, false).await.is_err(),
-            "the active owner fixture must start with one missing metadata copy"
-        );
-
-        let mut suspended_reader = PutObjReader::from_vec(b"suspended owner".to_vec());
-        store.pools[1]
-            .put_object(&bucket, suspended_only_object, &mut suspended_reader, &ObjectOptions::default())
-            .await
-            .expect("suspended owner object should be written");
-        for (pool_index, mod_time) in [1_i64, 2_i64].into_iter().enumerate() {
-            let mut duplicate_reader = PutObjReader::from_vec(format!("duplicate-pool-{pool_index}").into_bytes());
-            store.pools[pool_index]
-                .put_object(
-                    &bucket,
-                    duplicate_object,
-                    &mut duplicate_reader,
-                    &ObjectOptions {
-                        mod_time: Some(OffsetDateTime::UNIX_EPOCH + time::Duration::seconds(mod_time)),
-                        ..Default::default()
-                    },
-                )
-                .await
-                .expect("duplicate owner object should be written");
-        }
-        let duplicate_missing_disk = store.pools[0].disk_set[0].disks.read().await[0]
-            .clone()
-            .expect("duplicate active owner disk should be online");
-        duplicate_missing_disk
-            .delete(
-                &bucket,
-                duplicate_object,
-                DeleteOptions {
-                    recursive: true,
-                    immediate: true,
-                    ..Default::default()
-                },
-            )
-            .await
-            .expect("duplicate active owner shard should be removed for repair");
-        let history_version = Uuid::new_v4();
-        let mut history_reader = PutObjReader::from_vec(b"marker history".to_vec());
-        store.pools[0]
-            .put_object(
-                &bucket,
-                marker_object,
-                &mut history_reader,
-                &ObjectOptions {
-                    versioned: true,
-                    version_id: Some(history_version.to_string()),
-                    mod_time: Some(OffsetDateTime::UNIX_EPOCH + time::Duration::seconds(1)),
-                    ..Default::default()
-                },
-            )
-            .await
-            .expect("versioned marker history should be written");
-        store.pools[0]
-            .delete_object(
-                &bucket,
-                marker_object,
-                ObjectOptions {
-                    versioned: true,
-                    mod_time: Some(OffsetDateTime::UNIX_EPOCH + time::Duration::seconds(2)),
-                    ..Default::default()
-                },
-            )
-            .await
-            .expect("delete marker should be written");
-        let mut quorum_reader = PutObjReader::from_vec(b"quorum boundary".to_vec());
-        store.pools[0]
-            .put_object(&bucket, quorum_object, &mut quorum_reader, &ObjectOptions::default())
-            .await
-            .expect("quorum boundary object should be written");
-        {
-            let mut pool_meta = store.pool_meta.write().await;
-            let mut next = PoolMeta::new(&store.pools, &pool_meta);
-            next.pools[1].decommission = Some(PoolDecommissionInfo {
-                start_time: Some(OffsetDateTime::UNIX_EPOCH),
-                ..Default::default()
-            });
-            *pool_meta = next;
-        }
-
-        let (_, duplicate_owner) = store
-            .get_latest_object_info_with_idx(&bucket, duplicate_object, &ObjectOptions::default())
-            .await
-            .expect("duplicate owner should resolve");
-        assert_eq!(duplicate_owner, 1, "latest duplicate must win when all pools are eligible");
-        let (_, active_duplicate_owner) = store
-            .get_latest_object_info_with_idx(
-                &bucket,
-                duplicate_object,
-                &ObjectOptions {
-                    skip_decommissioned: true,
-                    ..Default::default()
-                },
-            )
-            .await
-            .expect("active duplicate owner should resolve");
-        assert_eq!(
-            active_duplicate_owner, 0,
-            "suspended duplicate must be excluded from active owner selection"
-        );
-        let (duplicate_result, duplicate_err) = store
-            .handle_heal_object(&bucket, duplicate_object, "", &HealOpts::default())
-            .await
-            .expect("duplicate owner heal should complete through the production path");
-        assert_eq!(duplicate_result.object, duplicate_object);
-        assert!(duplicate_err.is_none(), "active duplicate should be repaired: {duplicate_err:?}");
-        assert!(
-            duplicate_missing_disk.read_xl(&bucket, duplicate_object, false).await.is_ok(),
-            "production heal must repair the active duplicate owner rather than the suspended owner"
-        );
-        let (marker_info, marker_owner) = store
-            .get_latest_object_info_with_idx(
-                &bucket,
-                marker_object,
-                &ObjectOptions {
-                    skip_decommissioned: true,
-                    versioned: true,
-                    ..Default::default()
-                },
-            )
-            .await
-            .expect("latest delete marker should resolve");
-        assert_eq!(marker_owner, 0);
-        assert!(marker_info.delete_marker, "latest version must preserve delete-marker semantics");
-
-        let (active_result, active_err) = store
-            .handle_heal_object(&bucket, active_object, "", &HealOpts::default())
-            .await
-            .expect("unscoped active-owner heal should complete");
-        assert_eq!(active_result.object, active_object);
-        assert!(active_err.is_none(), "active owner must be selected even with a suspended pool");
-        assert!(
-            missing_active_disk.read_xl(&bucket, active_object, false).await.is_ok(),
-            "active owner heal must write the missing disk metadata: result={active_result:?}, err={active_err:?}"
-        );
-        assert!(
-            store.pools[1]
-                .get_object_info(&bucket, active_object, &ObjectOptions::default())
-                .await
-                .is_err(),
-            "the suspended pool must not be written for an active-owner object"
-        );
-
-        let (suspended_result, suspended_err) = store
-            .handle_heal_object(&bucket, suspended_only_object, "", &HealOpts::default())
-            .await
-            .expect("unscoped suspended-only heal should return a terminal result");
-        assert!(suspended_result.object.is_empty());
-        assert!(matches!(suspended_err, Some(Error::FileNotFound)));
-        assert!(
-            store.pools[1]
-                .get_object_info(&bucket, suspended_only_object, &ObjectOptions::default())
-                .await
-                .is_ok(),
-            "suspended-only data must remain untouched when unscoped heal reports absent"
-        );
-
-        let (_, explicit_err) = store
-            .handle_heal_object(
-                &bucket,
-                suspended_only_object,
-                "",
-                &HealOpts {
-                    pool: Some(1),
-                    ..Default::default()
-                },
-            )
-            .await
-            .expect("explicit suspended-owner heal should return a mapped error");
-        assert!(matches!(explicit_err, Some(Error::SlowDown)));
-
-        let original_quorum_disks = store.pools[0].disk_set[0].disks.read().await.clone();
-        let surviving_quorum_disk = original_quorum_disks[3].clone();
-        *store.pools[0].disk_set[0].disks.write().await = vec![None, None, None, surviving_quorum_disk];
-        let (_, quorum_err) = store
-            .handle_heal_object(&bucket, quorum_object, "", &HealOpts::default())
-            .await
-            .expect("quorum boundary heal should return a mapped result");
-        *store.pools[0].disk_set[0].disks.write().await = original_quorum_disks;
-        assert!(
-            matches!(quorum_err, Some(Error::ErasureReadQuorum)),
-            "quorum-boundary heal must preserve quorum error, got {quorum_err:?}"
-        );
-        shutdown.cancel();
-    }
-
    #[tokio::test]
    async fn handle_heal_format_continues_after_a_pool_error() {
        let canonical_format = FormatV3::new(1, 3);
@@ -33,7 +33,7 @@ use crate::bucket::utils::check_put_object_part_args;
 use crate::bucket::utils::{check_valid_bucket_name, check_valid_bucket_name_strict, is_meta_bucketname};
 use crate::cluster::rpc::{RemoteClient, S3PeerSys};
 use crate::config::storageclass;
-use crate::core::pools::{DecommissionCanceler, PoolMeta};
+use crate::core::pools::PoolMeta;
 use crate::disk::endpoint::{Endpoint, EndpointType};
 use crate::disk::{DiskAPI, DiskInfo, DiskInfoOptions};
 use crate::error::{Error, Result};
@@ -151,7 +151,7 @@ pub(crate) mod init_format;
 pub(crate) mod list_objects;
 mod multipart;
 mod object;
-pub(crate) use object::{ObjectLockDiagGuard, SourceCleanupMutationFence};
+pub(crate) use object::ObjectLockDiagGuard;
 pub use object::{
    PrepareSelectObjectSnapshotError, PreparedGetObjectReader, SelectObjectSnapshot, SelectObjectSnapshotReadError,
    SnapshotConsistencyError,
@@ -176,7 +176,7 @@ pub struct ECStore {
    // pub local_disks: Vec<DiskStore>,
    pub pool_meta: RwLock<PoolMeta>,
    pub rebalance_meta: RwLock<Option<RebalanceMeta>>,
-    pub decommission_cancelers: RwLock<Vec<Option<DecommissionCanceler>>>,
+    pub decommission_cancelers: RwLock<Vec<Option<CancellationToken>>>,
    /// Serializes rebalance/decommission start transitions.
    ///
    /// Lock order: acquire `start_gate` before `pool_meta`, `rebalance_meta`,
@@ -400,7 +400,7 @@ impl ECStore {
        object: &str,
        opts: &ObjectOptions,
    ) -> Result<MultipartUploadResult> {
-        self.handle_new_multipart_upload_with_pool_idx(bucket, object, opts, None)
+        self.handle_new_multipart_upload_with_pool_idx(bucket, object, opts)
            .await
            .map(|(res, _, _)| res)
    }
@@ -410,22 +410,20 @@ impl ECStore {
        bucket: &str,
        object: &str,
        opts: &ObjectOptions,
-        mutation_fence: Option<&ObjectLockDiagGuard>,
    ) -> Result<(MultipartUploadResult, usize, Option<Uuid>)> {
        check_new_multipart_args(bucket, object)?;
-        let (mut opts, _bucket_lifecycle_guard) = self.guard_multipart_bucket_incarnation(bucket, opts).await?;
+        let (opts, _bucket_lifecycle_guard) = self.guard_multipart_bucket_incarnation(bucket, opts).await?;
+        let opts = &opts;

        if self.single_pool() {
-            self.apply_decommission_target_mutation_fence(0, object, &mut opts, mutation_fence)
-                .await;
            return self.pools[0]
-                .new_multipart_upload(bucket, object, &opts)
+                .new_multipart_upload(bucket, object, opts)
                .await
                .map(|res| (res, 0, opts.expected_bucket_incarnation_id));
        }

        if opts.data_movement && opts.version_id.is_some() {
-            let idx = self.select_data_movement_pool_idx(bucket, object, -1, &opts, false).await?;
+            let idx = self.select_data_movement_pool_idx(bucket, object, -1, opts, false).await?;
            if idx == opts.src_pool_idx {
                return Err(StorageError::DataMovementOverwriteErr(
                    bucket.to_owned(),
@@ -433,9 +431,7 @@ impl ECStore {
                    opts.version_id.clone().unwrap_or_default(),
                ));
            }
-            self.apply_decommission_target_mutation_fence(idx, object, &mut opts, mutation_fence)
-                .await;
-            let res = self.pools[idx].new_multipart_upload(bucket, object, &opts).await?;
+            let res = self.pools[idx].new_multipart_upload(bucket, object, opts).await?;
            return Ok((res, idx, opts.expected_bucket_incarnation_id));
        }

@@ -458,9 +454,7 @@ impl ECStore {
            .await?;

            if !res.uploads.is_empty() {
-                self.apply_decommission_target_mutation_fence(idx, object, &mut opts, mutation_fence)
-                    .await;
-                let res = self.pools[idx].new_multipart_upload(bucket, object, &opts).await?;
+                let res = self.pools[idx].new_multipart_upload(bucket, object, opts).await?;
                return Ok((res, idx, opts.expected_bucket_incarnation_id));
            }
        }
@@ -473,9 +467,7 @@ impl ECStore {
            ));
        }

-        self.apply_decommission_target_mutation_fence(idx, object, &mut opts, mutation_fence)
-            .await;
-        let res = self.pools[idx].new_multipart_upload(bucket, object, &opts).await?;
+        let res = self.pools[idx].new_multipart_upload(bucket, object, opts).await?;
        Ok((res, idx, opts.expected_bucket_incarnation_id))
    }

@@ -712,14 +704,13 @@ impl ECStore {

    pub(crate) async fn complete_multipart_upload_for_data_movement(
        self: Arc<Self>,
-        target: (usize, Option<&ObjectLockDiagGuard>),
+        target_pool_idx: usize,
        bucket: &str,
        object: &str,
        upload_id: &str,
        uploaded_parts: Vec<CompletePart>,
        opts: &ObjectOptions,
    ) -> Result<ObjectInfo> {
-        let (target_pool_idx, mutation_fence) = target;
        check_complete_multipart_args(bucket, object, upload_id)?;
        if !opts.data_movement {
            return Err(Error::other("targeted multipart completion requires data_movement options"));
@@ -748,8 +739,6 @@ impl ECStore {
            snapshot.add_lock_fences(&mut opts);
            opts.object_lock_config_snapshot = Some(snapshot);
        }
-        self.apply_decommission_target_mutation_fence(target_pool_idx, object, &mut opts, mutation_fence)
-            .await;
        #[cfg(test)]
        pause_data_movement_multipart_before_selected_completion(bucket).await;
        let pool = self
@@ -859,7 +859,6 @@ fn lifecycle_delete_all_test_failure(phase: crate::object_api::LifecycleDeleteAl
 #[cfg(test)]
 mod tests {
    use super::*;
-    use crate::bucket::replication::{ReplicationStatusType, VersionPurgeStatusType};
    use crate::config::storageclass::{CLASS_RRS, CLASS_STANDARD, lookup_config_for_pools_without_env};
    use crate::disk::error::DiskError;
    use crate::layout::endpoint::Endpoint;
@@ -1424,14 +1423,6 @@ mod tests {
        }
    }

-    fn object_info_with_identity(unix_ts: i64, delete_marker: bool, version_id: Uuid, etag: Option<String>) -> ObjectInfo {
-        ObjectInfo {
-            version_id: Some(version_id),
-            etag,
-            ..object_info_with_mod_time(unix_ts, delete_marker)
-        }
-    }
-
    #[test]
    fn resolve_latest_object_info_candidates_returns_latest_delete_marker() {
        let candidates = vec![
@@ -1455,7 +1446,7 @@ mod tests {
    }

    #[test]
-    fn resolve_latest_object_info_candidates_prefers_higher_pool_idx_on_equal_mod_time_for_equivalent_candidates() {
+    fn resolve_latest_object_info_candidates_prefers_higher_pool_idx_on_equal_mod_time() {
        let candidates = vec![
            LatestObjectInfoCandidate {
                info: Some(object_info_with_mod_time(10, false)),
@@ -1475,382 +1466,6 @@ mod tests {
        assert_eq!(idx, 1);
    }

-    #[test]
-    fn resolve_latest_object_info_candidates_keeps_index_fallback_for_fully_equivalent_identities() {
-        let candidates = vec![
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()))),
-                idx: 2,
-                err: None,
-            },
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()))),
-                idx: 7,
-                err: None,
-            },
-        ];
-
-        let (info, idx) = resolve_latest_object_info_candidates(candidates, "bucket", "object", &ObjectOptions::default())
-            .expect("equivalent replicas must resolve deterministically");
-
-        assert_eq!(idx, 7);
-        assert_eq!(info.version_id, Some(Uuid::from_u128(1)));
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_rejects_equal_time_version_id_conflict() {
-        let candidates = vec![
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()))),
-                idx: 0,
-                err: None,
-            },
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, false, Uuid::from_u128(2), Some("etag-a".to_string()))),
-                idx: 1,
-                err: None,
-            },
-        ];
-
-        let err = resolve_latest_object_info_candidates(candidates, "bucket", "object", &ObjectOptions::default())
-            .expect_err("divergent version ids must not silently resolve to the higher pool index");
-
-        assert_eq!(err, Error::ErasureReadQuorum);
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_rejects_equal_time_etag_conflict() {
-        let candidates = vec![
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-old".to_string()))),
-                idx: 0,
-                err: None,
-            },
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-new".to_string()))),
-                idx: 1,
-                err: None,
-            },
-        ];
-
-        let err = resolve_latest_object_info_candidates(candidates, "bucket", "object", &ObjectOptions::default())
-            .expect_err("divergent etags must not silently resolve to the higher pool index");
-
-        assert_eq!(err, Error::ErasureReadQuorum);
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_rejects_equal_time_delete_marker_conflict() {
-        let candidates = vec![
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, false, Uuid::from_u128(1), None)),
-                idx: 0,
-                err: None,
-            },
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, true, Uuid::from_u128(1), Some("etag-a".to_string()))),
-                idx: 1,
-                err: None,
-            },
-        ];
-
-        let err = resolve_latest_object_info_candidates(candidates, "bucket", "object", &ObjectOptions::default())
-            .expect_err("a delete marker tied with a live version must not be masked by the pool index");
-
-        assert_eq!(err, Error::ErasureReadQuorum);
-    }
-
-    fn assert_equal_time_identity_conflict(left: ObjectInfo, right: ObjectInfo) {
-        let err = resolve_latest_object_info_candidates(
-            vec![
-                LatestObjectInfoCandidate {
-                    info: Some(left),
-                    idx: 0,
-                    err: None,
-                },
-                LatestObjectInfoCandidate {
-                    info: Some(right),
-                    idx: 1,
-                    err: None,
-                },
-            ],
-            "bucket",
-            "object",
-            &ObjectOptions::default(),
-        )
-        .expect_err("equal-time identity divergence must fail closed");
-
-        assert_eq!(err, Error::ErasureReadQuorum);
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_rejects_equal_time_payload_identity_conflicts() {
-        let base = object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()));
-
-        let mut data_dir = base.clone();
-        data_dir.data_dir = Some(Uuid::from_u128(2));
-        assert_equal_time_identity_conflict(base.clone(), data_dir);
-
-        let mut size = base.clone();
-        size.size = 1;
-        assert_equal_time_identity_conflict(base.clone(), size);
-
-        let mut actual_size = base.clone();
-        actual_size.actual_size = 1;
-        assert_equal_time_identity_conflict(base.clone(), actual_size);
-
-        let mut checksum = base.clone();
-        checksum.checksum = Some(bytes::Bytes::from_static(b"checksum"));
-        assert_equal_time_identity_conflict(base.clone(), checksum);
-
-        let mut parts = base.clone();
-        parts.parts = std::sync::Arc::new(vec![rustfs_filemeta::ObjectPartInfo {
-            etag: "part-etag".to_string(),
-            number: 1,
-            size: 1,
-            ..Default::default()
-        }]);
-        assert_equal_time_identity_conflict(base.clone(), parts);
-
-        let mut transition = base;
-        transition.transitioned_object.tier = "tier-a".to_string();
-        assert_equal_time_identity_conflict(
-            object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string())),
-            transition,
-        );
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_accepts_internal_metadata_aliases() {
-        let base = object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()));
-        let mut rustfs_alias = base.clone();
-        rustfs_alias.user_defined = std::sync::Arc::new(std::collections::HashMap::from([(
-            "x-rustfs-internal-compression".to_string(),
-            "zstd".to_string(),
-        )]));
-        let mut minio_alias = base.clone();
-        minio_alias.user_defined = std::sync::Arc::new(std::collections::HashMap::from([(
-            "X-MINIO-INTERNAL-COMPRESSION".to_string(),
-            "zstd".to_string(),
-        )]));
-
-        let (_, idx) = resolve_latest_object_info_candidates(
-            vec![
-                LatestObjectInfoCandidate {
-                    info: Some(rustfs_alias),
-                    idx: 0,
-                    err: None,
-                },
-                LatestObjectInfoCandidate {
-                    info: Some(minio_alias),
-                    idx: 1,
-                    err: None,
-                },
-            ],
-            "bucket",
-            "object",
-            &ObjectOptions::default(),
-        )
-        .expect("same-value internal aliases should resolve");
-        assert_eq!(idx, 1);
-
-        let mut dual_alias = base.clone();
-        dual_alias.user_defined = std::sync::Arc::new(std::collections::HashMap::from([
-            ("x-rustfs-internal-compression".to_string(), "zstd".to_string()),
-            ("x-minio-internal-compression".to_string(), "zstd".to_string()),
-        ]));
-        let mut single_alias = base;
-        single_alias.user_defined = std::sync::Arc::new(std::collections::HashMap::from([(
-            "x-rustfs-internal-compression".to_string(),
-            "zstd".to_string(),
-        )]));
-
-        let (_, idx) = resolve_latest_object_info_candidates(
-            vec![
-                LatestObjectInfoCandidate {
-                    info: Some(dual_alias),
-                    idx: 0,
-                    err: None,
-                },
-                LatestObjectInfoCandidate {
-                    info: Some(single_alias),
-                    idx: 1,
-                    err: None,
-                },
-            ],
-            "bucket",
-            "object",
-            &ObjectOptions::default(),
-        )
-        .expect("dual-key and single-key internal metadata should resolve");
-        assert_eq!(idx, 1);
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_rejects_different_internal_metadata_alias_values() {
-        let base = object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()));
-        let mut rustfs_alias = base.clone();
-        rustfs_alias.user_defined = std::sync::Arc::new(std::collections::HashMap::from([(
-            "x-rustfs-internal-compression".to_string(),
-            "zstd".to_string(),
-        )]));
-        let mut minio_alias = base;
-        minio_alias.user_defined = std::sync::Arc::new(std::collections::HashMap::from([(
-            "x-minio-internal-compression".to_string(),
-            "snappy".to_string(),
-        )]));
-
-        assert_equal_time_identity_conflict(rustfs_alias, minio_alias);
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_preserves_dynamic_internal_metadata_identity_case() {
-        for suffix_prefix in ["replication-reset-", "replication-delete-marker-version-"] {
-            let base = object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()));
-            let mut rustfs_alias = base.clone();
-            rustfs_alias.user_defined = std::sync::Arc::new(std::collections::HashMap::from([(
-                format!(
-                    "X-RUSTFS-INTERNAL-{}{suffix}",
-                    suffix_prefix.to_uppercase(),
-                    suffix = "arn:aws:s3:::Bucket"
-                ),
-                "value".to_string(),
-            )]));
-            let mut minio_alias = base.clone();
-            minio_alias.user_defined = std::sync::Arc::new(std::collections::HashMap::from([(
-                format!("x-minio-internal-{suffix_prefix}arn:aws:s3:::Bucket"),
-                "value".to_string(),
-            )]));
-
-            let (_, idx) = resolve_latest_object_info_candidates(
-                vec![
-                    LatestObjectInfoCandidate {
-                        info: Some(rustfs_alias.clone()),
-                        idx: 0,
-                        err: None,
-                    },
-                    LatestObjectInfoCandidate {
-                        info: Some(minio_alias),
-                        idx: 1,
-                        err: None,
-                    },
-                ],
-                "bucket",
-                "object",
-                &ObjectOptions::default(),
-            )
-            .expect("dynamic internal aliases with the same target should resolve");
-            assert_eq!(idx, 1);
-
-            let mut different_target_case = base;
-            different_target_case.user_defined = std::sync::Arc::new(std::collections::HashMap::from([(
-                format!("x-minio-internal-{suffix_prefix}arn:aws:s3:::bucket"),
-                "value".to_string(),
-            )]));
-
-            assert_equal_time_identity_conflict(rustfs_alias, different_target_case);
-        }
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_rejects_conflicting_internal_metadata_aliases_in_one_candidate() {
-        let base = object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()));
-        let mut first = base.clone();
-        first.user_defined = std::sync::Arc::new(std::collections::HashMap::from([
-            ("x-rustfs-internal-compression".to_string(), "zstd".to_string()),
-            ("x-minio-internal-compression".to_string(), "snappy".to_string()),
-        ]));
-        let mut second = base;
-        second.user_defined = first.user_defined.clone();
-
-        assert_equal_time_identity_conflict(first, second);
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_rejects_replication_identity_conflict() {
-        let base = object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()));
-
-        let mut replication = base.clone();
-        replication.replication_status_internal = Some("PENDING".to_string());
-        replication.replication_status = ReplicationStatusType::Pending;
-        assert_equal_time_identity_conflict(base.clone(), replication);
-
-        let mut purge = base.clone();
-        purge.version_purge_status_internal = Some("PENDING".to_string());
-        purge.version_purge_status = VersionPurgeStatusType::Pending;
-        assert_equal_time_identity_conflict(base.clone(), purge);
-
-        let mut decision = base;
-        decision.replication_decision = "replicate".to_string();
-        assert_equal_time_identity_conflict(
-            object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string())),
-            decision,
-        );
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_rejects_none_vs_unix_epoch_mod_time() {
-        let mut without_mod_time = object_info_with_identity(0, false, Uuid::from_u128(1), Some("etag-a".to_string()));
-        without_mod_time.mod_time = None;
-        let with_unix_epoch = object_info_with_identity(0, false, Uuid::from_u128(1), Some("etag-a".to_string()));
-
-        assert_equal_time_identity_conflict(without_mod_time, with_unix_epoch);
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_ignores_older_identity_conflicts() {
-        let latest = object_info_with_identity(20, false, Uuid::from_u128(1), Some("etag-latest".to_string()));
-        let mut older = object_info_with_identity(10, true, Uuid::from_u128(2), Some("etag-old".to_string()));
-        older.data_dir = Some(Uuid::from_u128(2));
-
-        let (info, idx) = resolve_latest_object_info_candidates(
-            vec![
-                LatestObjectInfoCandidate {
-                    info: Some(latest),
-                    idx: 0,
-                    err: None,
-                },
-                LatestObjectInfoCandidate {
-                    info: Some(older),
-                    idx: 9,
-                    err: None,
-                },
-            ],
-            "bucket",
-            "object",
-            &ObjectOptions::default(),
-        )
-        .expect("older identity divergence must not affect the latest candidate");
-
-        assert_eq!(idx, 0);
-        assert_eq!(
-            info.mod_time,
-            Some(OffsetDateTime::from_unix_timestamp(20).expect("operation should succeed"))
-        );
-    }
-
-    #[test]
-    fn resolve_latest_object_info_candidates_ignores_not_found_pools_when_resolving() {
-        let candidates = vec![
-            LatestObjectInfoCandidate {
-                info: Some(object_info_with_identity(10, false, Uuid::from_u128(1), Some("etag-a".to_string()))),
-                idx: 0,
-                err: None,
-            },
-            LatestObjectInfoCandidate {
-                info: None,
-                idx: 1,
-                err: Some(Error::ObjectNotFound("bucket".to_string(), "object".to_string())),
-            },
-        ];
-
-        let (info, idx) = resolve_latest_object_info_candidates(candidates, "bucket", "object", &ObjectOptions::default())
-            .expect("not-found pools must not block resolution of found candidates");
-
-        assert_eq!(idx, 0);
-        assert_eq!(info.version_id, Some(Uuid::from_u128(1)));
-    }
-
    #[test]
    fn resolve_latest_object_info_candidates_returns_non_not_found_error() {
        let err = resolve_latest_object_info_candidates(
@@ -12,14 +12,10 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.

-use std::collections::HashMap;
+use std::cmp::Ordering;

 use crate::error::{Error, Result, StorageError, is_err_object_not_found, is_err_version_not_found};
 use crate::object_api::{ObjectInfo, ObjectOptions};
-use rustfs_utils::http::metadata_compat::{
-    SUFFIX_REPLICATION_DELETE_MARKER_VERSION_ARN_PREFIX, SUFFIX_REPLICATION_RESET_ARN_PREFIX,
-    strip_internal_prefix_preserving_case,
-};
 use rustfs_utils::path::decode_dir_object;
 use time::OffsetDateTime;

@@ -77,7 +73,7 @@ pub(super) fn resolve_rebalance_delete_from_all_pools_result(
    object: &str,
 ) -> Result<ObjectInfo> {
    result.map_err(|err| {
-        if matches!(&err, Error::PreconditionFailed | Error::PrefixAccessDenied(_, _)) {
+        if err == Error::PreconditionFailed {
            err
        } else {
            Error::other(format!("failed to delete rebalance source object {bucket}/{object}: {err}"))
@@ -90,7 +86,7 @@ fn is_ignorable_rebalance_delete_error(err: &Error) -> bool {
 }

 fn rebalance_delete_pool_error(pool_idx: usize, bucket: &str, object: &str, err: Error) -> Error {
-    if matches!(&err, Error::PreconditionFailed | Error::PrefixAccessDenied(_, _)) {
+    if err == Error::PreconditionFailed {
        err
    } else {
        Error::other(format!("pool {pool_idx} delete failed for {bucket}/{object}: {err}"))
@@ -141,158 +137,37 @@ pub(super) fn rebalance_disk_set_lookup_error(pool_idx: usize, set_idx: usize, p
    ))
 }

-fn latest_candidate_mod_time(candidate: &LatestObjectInfoCandidate) -> Option<OffsetDateTime> {
-    candidate
-        .info
-        .as_ref()
-        .map(|info| info.mod_time.unwrap_or(OffsetDateTime::UNIX_EPOCH))
-}
-
-fn same_transition_identity(left: &ObjectInfo, right: &ObjectInfo) -> bool {
-    left.transition_version_state == right.transition_version_state
-        && left.transitioned_object.name == right.transitioned_object.name
-        && left.transitioned_object.version_id == right.transitioned_object.version_id
-        && left.transitioned_object.tier == right.transitioned_object.tier
-        && left.transitioned_object.free_version == right.transitioned_object.free_version
-        && left.transitioned_object.status == right.transitioned_object.status
-}
-
-#[derive(PartialEq, Eq)]
-struct LatestUserDefinedIdentity {
-    internal: HashMap<String, String>,
-    other: HashMap<String, String>,
-}
-
-fn normalize_internal_identity_suffix(key: &str) -> Option<String> {
-    let suffix = strip_internal_prefix_preserving_case(key)?;
-
-    for dynamic_prefix in [
-        SUFFIX_REPLICATION_RESET_ARN_PREFIX,
-        SUFFIX_REPLICATION_DELETE_MARKER_VERSION_ARN_PREFIX,
-    ] {
-        let prefix_len = dynamic_prefix.len();
-        if let (Some(prefix), Some(remainder)) = (suffix.get(..prefix_len), suffix.get(prefix_len..))
-            && prefix.eq_ignore_ascii_case(dynamic_prefix)
-        {
-            return Some(format!("{dynamic_prefix}{remainder}"));
-        }
-    }
-
-    Some(suffix.to_lowercase())
-}
-
-fn normalize_user_defined_identity(user_defined: &HashMap<String, String>) -> Option<LatestUserDefinedIdentity> {
-    let mut identity = LatestUserDefinedIdentity {
-        internal: HashMap::with_capacity(user_defined.len()),
-        other: HashMap::with_capacity(user_defined.len()),
-    };
-
-    for (key, value) in user_defined {
-        if let Some(suffix) = normalize_internal_identity_suffix(key) {
-            if identity
-                .internal
-                .insert(suffix, value.clone())
-                .is_some_and(|previous| previous != *value)
-            {
-                return None;
-            }
-        } else {
-            identity.other.insert(key.clone(), value.clone());
-        }
-    }
-
-    Some(identity)
-}
-
-fn same_user_defined_identity(left: &ObjectInfo, right: &ObjectInfo) -> bool {
-    match (
-        normalize_user_defined_identity(&left.user_defined),
-        normalize_user_defined_identity(&right.user_defined),
-    ) {
-        (Some(left), Some(right)) => left == right,
-        _ => false,
-    }
-}
-
-/// Pool-specific erasure geometry is intentionally excluded: `get_object_info`
-/// returns each pool's own `data_blocks`/`parity_blocks`, so those values can
-/// differ for the same object version while the selected winner still carries
-/// the chosen pool's layout. `put_object_reader` is also intentionally
-/// excluded because it is a transient request handle that `ObjectInfo::clone`
-/// drops. Every other ObjectInfo field is part of the production-visible
-/// identity and must agree before the pool index can provide a deterministic
-/// tie-break.
-fn same_latest_object_info_identity(left: &ObjectInfo, right: &ObjectInfo) -> bool {
-    left.bucket == right.bucket
-        && left.name == right.name
-        && left.storage_class == right.storage_class
-        && left.mod_time == right.mod_time
-        && left.size == right.size
-        && left.actual_size == right.actual_size
-        && left.is_dir == right.is_dir
-        && same_user_defined_identity(left, right)
-        && left.user_tags == right.user_tags
-        && left.version_id == right.version_id
-        && left.data_dir == right.data_dir
-        && left.delete_marker == right.delete_marker
-        && same_transition_identity(left, right)
-        && left.restore_ongoing == right.restore_ongoing
-        && left.restore_expires == right.restore_expires
-        && left.parts == right.parts
-        && left.is_latest == right.is_latest
-        && left.content_type == right.content_type
-        && left.content_encoding == right.content_encoding
-        && left.expires == right.expires
-        && left.num_versions == right.num_versions
-        && left.successor_mod_time == right.successor_mod_time
-        && left.etag == right.etag
-        && left.inlined == right.inlined
-        && left.metadata_only == right.metadata_only
-        && left.version_only == right.version_only
-        && left.replication_status_internal == right.replication_status_internal
-        && left.replication_status == right.replication_status
-        && left.version_purge_status_internal == right.version_purge_status_internal
-        && left.version_purge_status == right.version_purge_status
-        && left.replication_decision == right.replication_decision
-        && left.checksum == right.checksum
-}
-
 pub(super) fn resolve_latest_object_info_candidates(
-    candidates: Vec<LatestObjectInfoCandidate>,
+    mut candidates: Vec<LatestObjectInfoCandidate>,
    bucket: &str,
    object: &str,
    opts: &ObjectOptions,
 ) -> Result<(ObjectInfo, usize)> {
-    let latest_mod_time = candidates.iter().filter_map(latest_candidate_mod_time).max();
-
-    if let Some(latest_mod_time) = latest_mod_time {
-        let mut latest_candidates = candidates
-            .into_iter()
-            .filter(|candidate| latest_candidate_mod_time(candidate) == Some(latest_mod_time))
-            .collect::<Vec<_>>();
-
-        latest_candidates.sort_by(|left, right| right.idx.cmp(&left.idx));
-
-        let Some(winner) = latest_candidates.first() else {
-            return Err(Error::ErasureReadQuorum);
-        };
-        let Some(winner_info) = winner.info.as_ref() else {
-            return Err(Error::ErasureReadQuorum);
+    candidates.sort_by(|a, b| {
+        let a_mod = if let Some(info) = &a.info {
+            info.mod_time.unwrap_or(OffsetDateTime::UNIX_EPOCH)
+        } else {
+            OffsetDateTime::UNIX_EPOCH
        };

-        if latest_candidates.iter().skip(1).any(|candidate| {
-            candidate
-                .info
-                .as_ref()
-                .is_none_or(|info| !same_latest_object_info_identity(winner_info, info))
-        }) {
-            return Err(Error::ErasureReadQuorum);
+        let b_mod = if let Some(info) = &b.info {
+            info.mod_time.unwrap_or(OffsetDateTime::UNIX_EPOCH)
+        } else {
+            OffsetDateTime::UNIX_EPOCH
+        };
+
+        if a_mod == b_mod {
+            return if a.idx < b.idx { Ordering::Greater } else { Ordering::Less };
        }

-        return Ok((winner_info.clone(), winner.idx));
-    }
+        b_mod.cmp(&a_mod)
+    });

    for candidate in candidates {
+        if let Some(info) = candidate.info {
+            return Ok((info, candidate.idx));
+        }
+
        if let Some(err) = candidate.err
            && !is_err_object_not_found(&err)
            && !is_err_version_not_found(&err)
@@ -316,18 +191,6 @@ mod tests {
        assert_eq!(err, Error::PreconditionFailed);
    }

-    #[test]
-    fn rebalance_delete_result_preserves_prefix_access_denied() {
-        let err = resolve_rebalance_delete_from_all_pools_result(
-            Err(Error::PrefixAccessDenied("bucket".to_owned(), "object".to_owned())),
-            "bucket",
-            "object",
-        )
-        .expect_err("prefix access denial should remain structured");
-
-        assert_eq!(err, Error::PrefixAccessDenied("bucket".to_owned(), "object".to_owned()));
-    }
-
    #[test]
    fn rebalance_delete_pool_result_preserves_precondition_failed() {
        let err = resolve_rebalance_delete_from_all_pools_results(
@@ -342,19 +205,4 @@ mod tests {

        assert_eq!(err, Error::PreconditionFailed);
    }
-
-    #[test]
-    fn rebalance_delete_pool_result_preserves_prefix_access_denied() {
-        let err = resolve_rebalance_delete_from_all_pools_results(
-            vec![RebalanceDeletePoolResult {
-                pool_idx: 0,
-                result: Err(Error::PrefixAccessDenied("bucket".to_owned(), "object".to_owned())),
-            }],
-            "bucket",
-            "object",
-        )
-        .expect_err("prefix access denial should remain structured");
-
-        assert_eq!(err, Error::PrefixAccessDenied("bucket".to_owned(), "object".to_owned()));
-    }
 }
@@ -54,13 +54,6 @@ pub const INTERNODE_STAGE_READ_VERSION_RESPONSE_JSON_ENCODE: &str = "read_versio
 pub const INTERNODE_STAGE_READ_VERSION_RESPONSE_MSGPACK_ENCODE: &str = "read_version_response_msgpack_encode";
 pub const INTERNODE_STAGE_READ_VERSION_RPC_ROUNDTRIP: &str = "read_version_rpc_roundtrip";
 pub const INTERNODE_STAGE_READ_VERSION_RESPONSE_DECODE: &str = "read_version_response_decode";
-pub const INTERNODE_STAGE_BATCH_READ_VERSION_REQUEST_ENCODE: &str = "batch_read_version_request_encode";
-pub const INTERNODE_STAGE_BATCH_READ_VERSION_REQUEST_DECODE: &str = "batch_read_version_request_decode";
-pub const INTERNODE_STAGE_BATCH_READ_VERSION_DISK_READ: &str = "batch_read_version_disk_read";
-pub const INTERNODE_STAGE_BATCH_READ_VERSION_RESPONSE_JSON_ENCODE: &str = "batch_read_version_response_json_encode";
-pub const INTERNODE_STAGE_BATCH_READ_VERSION_RESPONSE_MSGPACK_ENCODE: &str = "batch_read_version_response_msgpack_encode";
-pub const INTERNODE_STAGE_BATCH_READ_VERSION_RPC_ROUNDTRIP: &str = "batch_read_version_rpc_roundtrip";
-pub const INTERNODE_STAGE_BATCH_READ_VERSION_RESPONSE_DECODE: &str = "batch_read_version_response_decode";

 const OPERATION_LABEL: &str = "operation";
 const BACKEND_LABEL: &str = "backend";
@@ -689,14 +689,6 @@ pub struct ScannerMetrics {
    pub cycle_max_objects: u64,
    #[serde(rename = "cycle_max_directories", default)]
    pub cycle_max_directories: u64,
-    #[serde(rename = "cycle_timeout_total", default)]
-    pub cycle_timeout_total: u64,
-    #[serde(rename = "cycle_recovery_required_total", default)]
-    pub cycle_recovery_required_total: u64,
-    #[serde(rename = "cycle_last_progress_age", default)]
-    pub cycle_last_progress_age: u64,
-    #[serde(rename = "leader_lease_without_progress", default)]
-    pub leader_lease_without_progress: bool,
    #[serde(rename = "bitrot_cycle_enabled", default)]
    pub bitrot_cycle_enabled: bool,
    #[serde(rename = "bitrot_cycle_seconds", default)]
@@ -772,8 +764,6 @@ impl ScannerMetrics {
            self.cycle_max_duration_seconds = other.cycle_max_duration_seconds;
            self.cycle_max_objects = other.cycle_max_objects;
            self.cycle_max_directories = other.cycle_max_directories;
-            self.cycle_last_progress_age = other.cycle_last_progress_age;
-            self.leader_lease_without_progress = other.leader_lease_without_progress;
            self.bitrot_cycle_enabled = other.bitrot_cycle_enabled;
            self.bitrot_cycle_seconds = other.bitrot_cycle_seconds;
        }
@@ -867,12 +857,6 @@ impl ScannerMetrics {
            .saturating_add(other.last_cycle_replication_checks);
        self.last_cycle_usage_saves = self.last_cycle_usage_saves.saturating_add(other.last_cycle_usage_saves);
        self.failed_cycles = self.failed_cycles.saturating_add(other.failed_cycles);
-        self.cycle_timeout_total = self.cycle_timeout_total.saturating_add(other.cycle_timeout_total);
-        self.cycle_recovery_required_total = self
-            .cycle_recovery_required_total
-            .saturating_add(other.cycle_recovery_required_total);
-        self.cycle_last_progress_age = self.cycle_last_progress_age.max(other.cycle_last_progress_age);
-        self.leader_lease_without_progress |= other.leader_lease_without_progress;
        self.superseded_cycles = self.superseded_cycles.saturating_add(other.superseded_cycles);
        self.partial_cycles_unknown = self.partial_cycles_unknown.saturating_add(other.partial_cycles_unknown);
        self.partial_cycles_runtime = self.partial_cycles_runtime.saturating_add(other.partial_cycles_runtime);
@@ -195,8 +195,7 @@ pub struct BucketPolicyArgs<'a> {
 #[derive(Serialize, Deserialize, Clone, Default, Debug)]
 #[serde(deny_unknown_fields)]
 pub struct BucketPolicy {
-    // RUSTFS_COMPAT_TODO(rustfs-6339): accept bucket policies persisted with the legacy "ID" key. Remove after migration tooling rewrites every retained legacy bucket policy.
-    #[serde(default, rename = "Id", alias = "ID", skip_serializing_if = "ID::is_empty")]
+    #[serde(default, rename = "Id", skip_serializing_if = "ID::is_empty")]
    pub id: ID,
    #[serde(rename = "Version")]
    pub version: String,
@@ -2787,7 +2786,7 @@ mod test {
        let parsed: serde_json::Value = serde_json::from_str(&json).expect("Should parse");

        // Verify empty fields are omitted
-        assert!(parsed.get("Id").is_none(), "Empty ID should be omitted");
+        assert!(!parsed.as_object().unwrap().contains_key("ID"), "Empty ID should be omitted");

        let statement = &parsed["Statement"][0];
        assert!(!statement.as_object().unwrap().contains_key("Sid"), "Empty Sid should be omitted");
@@ -2810,43 +2809,6 @@ mod test {
        assert_eq!(statement["Principal"]["AWS"], "*");
    }

-    #[test]
-    fn test_bucket_policy_deserializes_legacy_id() {
-        let legacy_policy = br#"{"ID":"","Version":"2012-10-17","Statement":[{"Sid":"","Effect":"Allow","Principal":{"AWS":["*"]},"Action":["s3:GetObject"],"NotAction":[],"Resource":["arn:aws:s3:::bucket/*"],"NotResource":[],"Condition":{}}]}"#;
-
-        let policy: BucketPolicy =
-            serde_json::from_slice(legacy_policy).expect("bucket policy with legacy ID should deserialize");
-        assert!(policy.id.is_empty());
-        policy.is_valid().expect("legacy bucket policy should remain valid");
-
-        let policy: BucketPolicy = serde_json::from_str(r#"{"ID":"legacy-policy","Version":"2012-10-17","Statement":[]}"#)
-            .expect("non-empty legacy ID should deserialize");
-        assert_eq!(policy.id.0, "legacy-policy");
-
-        let serialized = serde_json::to_value(&policy).expect("bucket policy should serialize");
-        assert_eq!(serialized["Id"], "legacy-policy");
-        assert!(serialized.get("ID").is_none(), "legacy ID spelling should not be serialized");
-    }
-
-    #[test]
-    fn test_bucket_policy_legacy_id_alias_remains_strict() {
-        let unknown_field = r#"{"Version":"2012-10-17","Statement":[],"Unexpected":true}"#;
-        let error =
-            serde_json::from_str::<BucketPolicy>(unknown_field).expect_err("unrelated unknown fields should remain rejected");
-        assert!(
-            error.to_string().contains("unknown field `Unexpected`"),
-            "unexpected deserialization error: {error}"
-        );
-
-        let duplicate_id = r#"{"Id":"current-policy","ID":"legacy-policy","Version":"2012-10-17","Statement":[]}"#;
-        let error = serde_json::from_str::<BucketPolicy>(duplicate_id)
-            .expect_err("canonical and legacy ID fields should not be accepted together");
-        assert!(
-            error.to_string().contains("duplicate field `Id`"),
-            "unexpected deserialization error: {error}"
-        );
-    }
-
    #[test]
    fn test_existing_object_tag_condition_helpers() {
        let identity_policy = Policy::parse_config(
@@ -60,9 +60,7 @@ pub const REPLICATION_READ_ONLY_HISTORICAL_FIELDS: &[&str] = &[
    "Destination.ReplicationTime",
 ];

-// v2: disableProxy moved from unsupported to writable (per-target read-proxy
-// opt-out is accepted by set-remote-target and the `proxy` update op).
-pub const REMOTE_TARGET_CAPABILITY_CONTRACT_VERSION: u32 = 2;
+pub const REMOTE_TARGET_CAPABILITY_CONTRACT_VERSION: u32 = 1;

 pub const REMOTE_TARGET_WRITABLE_FIELDS: &[&str] = &[
    "sourcebucket",
@@ -85,12 +83,9 @@ pub const REMOTE_TARGET_WRITABLE_FIELDS: &[&str] = &[
    // madmin default of 60s); the per-target health-check interval is not
    // yet applied — the heartbeat keeps its global env-configured interval.
    "healthCheckDuration",
-    // Per-target read-proxy opt-out, consumed by the proxy-target selector
-    // (contract v2; previously only importable via MinIO bucket-targets.json).
-    "disableProxy",
 ];

-pub const REMOTE_TARGET_UNSUPPORTED_FIELDS: &[&str] = &["edge", "edgeSyncBeforeExpiry"];
+pub const REMOTE_TARGET_UNSUPPORTED_FIELDS: &[&str] = &["disableProxy", "edge", "edgeSyncBeforeExpiry"];

 #[derive(Debug, Clone, Serialize, Deserialize, Default)]
 pub struct ObjectOpts {
@@ -125,34 +125,6 @@ pub(crate) async fn read_config_with_revision<S: ScannerObjectIO>(
    }
 }

-/// Read only the object revision without materializing its body.
-pub(crate) async fn read_config_revision<S: ScannerObjectIO>(store: Arc<S>, path: &str) -> StorageResult<DataUsageCacheRevision> {
-    match store
-        .get_object_reader(
-            RUSTFS_META_BUCKET,
-            path,
-            None,
-            HeaderMap::new(),
-            &ObjectOptions {
-                no_lock: true,
-                ..Default::default()
-            },
-        )
-        .await
-    {
-        Ok(reader) => reader
-            .object_info
-            .etag
-            .filter(|etag| !etag.is_empty())
-            .map(DataUsageCacheRevision::Etag)
-            .ok_or_else(|| StorageError::other(format!("scanner config object {path} has no ETag"))),
-        Err(Error::FileNotFound | Error::VolumeNotFound | Error::ObjectNotFound(_, _) | Error::BucketNotFound(_)) => {
-            Ok(DataUsageCacheRevision::Missing)
-        }
-        Err(err) => Err(err),
-    }
-}
-
 #[derive(Clone, Debug)]
 pub(crate) struct DataUsageCacheRevisions {
    main: DataUsageCacheRevision,
@@ -174,11 +146,6 @@ pub static LEGACY_DATA_USAGE_OBJ_NAME_PATH: LazyLock<String> =
 pub static DATA_USAGE_BLOOM_NAME_PATH: LazyLock<String> =
    LazyLock::new(|| format!("{BUCKET_META_PREFIX}{SLASH_SEPARATOR}{DATA_USAGE_BLOOM_NAME}"));

-/// Durable companion object for a cycle-state object which cannot be decoded.
-/// The primary object is deliberately never replaced or deleted by recovery.
-pub static DATA_USAGE_BLOOM_RECOVERY_PATH: LazyLock<String> =
-    LazyLock::new(|| format!("{}.recovery-required.json", DATA_USAGE_BLOOM_NAME_PATH.as_str()));
-
 pub static BACKGROUND_HEAL_INFO_PATH: LazyLock<String> =
    LazyLock::new(|| format!("{BUCKET_META_PREFIX}{SLASH_SEPARATOR}.background-heal.json"));

@@ -74,7 +74,7 @@ impl DataUsageCache {
        let loaded = Self::load_cache(store.clone(), name).await?;
        let backup = match loaded.backup_revision {
            Some(revision) => Some(revision),
-            None => match read_config_revision(store, &backup_path).await {
+            None => match Self::revision_for_path(store, &backup_path).await {
                Ok(revision) => Some(revision),
                Err(err) => {
                    counter!(METRIC_CACHE_BACKUP_REVISION_FAILURE_TOTAL).increment(1);
@@ -336,6 +336,33 @@ impl DataUsageCache {
        }
    }

+    async fn revision_for_path<S: ScannerObjectIO>(store: Arc<S>, path: &str) -> StorageResult<DataUsageCacheRevision> {
+        match store
+            .get_object_reader(
+                RUSTFS_META_BUCKET,
+                path,
+                None,
+                HeaderMap::new(),
+                &ObjectOptions {
+                    no_lock: true,
+                    ..Default::default()
+                },
+            )
+            .await
+        {
+            Ok(reader) => reader
+                .object_info
+                .etag
+                .filter(|etag| !etag.is_empty())
+                .map(DataUsageCacheRevision::Etag)
+                .ok_or_else(|| StorageError::other(format!("scanner cache object {path} has no ETag"))),
+            Err(Error::FileNotFound | Error::VolumeNotFound | Error::ObjectNotFound(_, _) | Error::BucketNotFound(_)) => {
+                Ok(DataUsageCacheRevision::Missing)
+            }
+            Err(err) => Err(err),
+        }
+    }
+
    pub(super) fn cache_save_timeout() -> Duration {
        crate::runtime_config::scanner_cache_save_timeout()
    }
@@ -16,7 +16,7 @@ use super::persistence::DataUsageCacheLoadAttempt;
 use super::*;
 use crate::storage_api::scanner_io::{HTTPRangeSpec, ObjectIO};
 use crate::{ScannerGetObjectReader, ScannerPutObjReader};
-use rustfs_data_usage::{ReplicationAllStats, ReplicationTargetUsage};
+use rustfs_data_usage::{ReplicationAllStats, ReplicationStats};
 use serde_json::Value;
 use std::io::Cursor;
 use std::pin::Pin;
@@ -1636,7 +1636,7 @@ fn size_recursive_prunes_empty_and_preserves_threshold_replication_stats() {
            replication_stats: Some(ReplicationAllStats {
                targets: HashMap::from([(
                    "arn:test:threshold".to_string(),
-                    ReplicationTargetUsage {
+                    ReplicationStats {
                        after_threshold_count: 1,
                        ..Default::default()
                    },
@@ -75,10 +75,7 @@ pub use remote_scanner::{
 };
 pub use runtime_config::{apply_scanner_runtime_config, scanner_runtime_config_status, validate_scanner_runtime_config};
 pub use rustfs_common::last_minute;
-pub use scanner::{
-    ScannerCycleRecoveryMarker, ScannerCycleRecoveryStatus, ScannerCycleScheduleStatus, init_data_scanner,
-    reset_scanner_cycle_recovery, scanner_cycle_recovery_status, scanner_cycle_schedule_status, scanner_topology_digest,
-};
+pub use scanner::{ScannerCycleScheduleStatus, init_data_scanner, scanner_cycle_schedule_status, scanner_topology_digest};
 pub use scanner_io::{
    ScannerDirtyUsageAckError, ScannerDirtyUsageState, acknowledge_dirty_usage_generation, clear_dirty_usage_bucket,
    record_dirty_usage_bucket, record_scanner_maintenance_change, scanner_activity_epoch, scanner_dirty_usage_state,
@@ -125,10 +125,7 @@ impl Default for ScannerRuntimeConfig {
            cycle_interval_source: ScannerRuntimeConfigSource::Default,
            bitrot_cycle: Some(Duration::from_secs(DEFAULT_HEAL_BITROT_CYCLE_SECS)),
            bitrot_cycle_source: ScannerRuntimeConfigSource::Default,
-            cycle_budget: ScannerCycleBudgetConfig {
-                max_duration: Some(Duration::from_secs(DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS)),
-                ..Default::default()
-            },
+            cycle_budget: ScannerCycleBudgetConfig::default(),
            cycle_max_duration_source: ScannerRuntimeConfigSource::Default,
            cycle_max_objects_source: ScannerRuntimeConfigSource::Default,
            cycle_max_directories_source: ScannerRuntimeConfigSource::Default,
@@ -377,10 +374,7 @@ fn validate_persisted_scanner_runtime_config(config: &ServerConfig) -> Result<()
    }
    validate_optional_config_u64(scanner_kvs, SCANNER_START_DELAY, "")?;
    validate_optional_config_u64(scanner_kvs, SCANNER_CYCLE, "")?;
-    if let Some(value) = config_value(scanner_kvs, SCANNER_CYCLE_MAX_DURATION, DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS) {
-        let secs = parse_config_u64(SCANNER_CYCLE_MAX_DURATION, value)?;
-        cycle_duration_from_secs(SCANNER_CYCLE_MAX_DURATION, secs)?;
-    }
+    validate_optional_config_u64(scanner_kvs, SCANNER_CYCLE_MAX_DURATION, DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS)?;
    validate_optional_config_u64(scanner_kvs, SCANNER_CYCLE_MAX_OBJECTS, DEFAULT_SCANNER_CYCLE_MAX_OBJECTS)?;
    validate_optional_config_u64(scanner_kvs, SCANNER_CYCLE_MAX_DIRECTORIES, DEFAULT_SCANNER_CYCLE_MAX_DIRECTORIES)?;
    if let Some(value) = config_value(heal_kvs, HEAL_BITROT_CYCLE, DEFAULT_HEAL_BITROT_CYCLE_SECS) {
@@ -442,46 +436,19 @@ fn lookup_max_wait(
    Ok((speed.max_sleep(), speed_source))
 }

-fn lookup_cycle_duration(kvs: Option<&KVS>) -> Result<(Option<Duration>, ScannerRuntimeConfigSource), ScannerRuntimeConfigError> {
-    match rustfs_utils::get_env_parse_outcome::<u64>(ENV_SCANNER_CYCLE_MAX_DURATION_SECS) {
-        rustfs_utils::EnvParseOutcome::Parsed(secs) => {
-            return cycle_duration_from_secs(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, secs)
-                .map(|duration| (duration, ScannerRuntimeConfigSource::Env));
-        }
-        rustfs_utils::EnvParseOutcome::Invalid => {
-            // Do not include the raw environment value in the typed error:
-            // deployments occasionally put sensitive material in inherited
-            // environment snapshots. The key still identifies the control.
-            return Err(invalid_value(
-                ENV_SCANNER_CYCLE_MAX_DURATION_SECS,
-                "<invalid>",
-                "expected unsigned integer seconds",
-            ));
-        }
-        rustfs_utils::EnvParseOutcome::Absent => {}
+fn lookup_optional_seconds(
+    kvs: Option<&KVS>,
+    key: &'static str,
+    env_key: &'static str,
+    default: u64,
+) -> Result<(Option<Duration>, ScannerRuntimeConfigSource), ScannerRuntimeConfigError> {
+    if let Some(secs) = rustfs_utils::get_env_opt_u64(env_key) {
+        return Ok((Some(Duration::from_secs(secs)), ScannerRuntimeConfigSource::Env));
    }
-
-    if let Some(value) = config_value(kvs, SCANNER_CYCLE_MAX_DURATION, DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS) {
-        let secs = parse_config_u64(SCANNER_CYCLE_MAX_DURATION, value)?;
-        return cycle_duration_from_secs(SCANNER_CYCLE_MAX_DURATION, secs)
-            .map(|duration| (duration, ScannerRuntimeConfigSource::Config));
+    if let Some(value) = config_value(kvs, key, default) {
+        return parse_config_u64(key, value).map(|secs| (Some(Duration::from_secs(secs)), ScannerRuntimeConfigSource::Config));
    }
-
-    Ok((
-        Some(Duration::from_secs(DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS)),
-        ScannerRuntimeConfigSource::Default,
-    ))
-}
-
-fn cycle_duration_from_secs(key: &'static str, secs: u64) -> Result<Option<Duration>, ScannerRuntimeConfigError> {
-    if secs == 0 {
-        return Ok(None);
-    }
-    let duration = Duration::from_secs(secs);
-    if std::time::Instant::now().checked_add(duration).is_none() {
-        return Err(invalid_value(key, "<overflow>", "duration exceeds the timer range"));
-    }
-    Ok(Some(duration))
+    Ok((None, ScannerRuntimeConfigSource::Default))
 }

 fn lookup_start_delay(kvs: Option<&KVS>) -> Result<(Option<Duration>, ScannerRuntimeConfigSource), ScannerRuntimeConfigError> {
@@ -586,7 +553,12 @@ pub(crate) fn lookup_scanner_runtime_config(
        (speed.cycle_interval(), speed_source)
    };

-    let (cycle_max_duration, cycle_max_duration_source) = lookup_cycle_duration(scanner_kvs)?;
+    let (cycle_max_duration, cycle_max_duration_source) = lookup_optional_seconds(
+        scanner_kvs,
+        SCANNER_CYCLE_MAX_DURATION,
+        ENV_SCANNER_CYCLE_MAX_DURATION_SECS,
+        DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS,
+    )?;
    let (cycle_max_objects, cycle_max_objects_source) = lookup_count_budget(
        scanner_kvs,
        SCANNER_CYCLE_MAX_OBJECTS,
@@ -891,10 +863,10 @@ mod tests {
    use rustfs_config::server_config::{Config as ServerConfig, KVS};
    use rustfs_config::{
        DEFAULT_DELIMITER, DEFAULT_HEAL_BITROT_CYCLE_SECS, ENV_SCANNER_BITROT_CYCLE_SECS, ENV_SCANNER_CACHE_SAVE_TIMEOUT_SECS,
-        ENV_SCANNER_CYCLE, ENV_SCANNER_CYCLE_MAX_DURATION_SECS, ENV_SCANNER_CYCLE_MAX_OBJECTS, ENV_SCANNER_DELAY,
-        ENV_SCANNER_MAX_WAIT_SECS, ENV_SCANNER_SPEED, HEAL_BITROT_CYCLE, HEAL_SUB_SYS, SCANNER_BITROT_CYCLE,
-        SCANNER_CACHE_SAVE_TIMEOUT, SCANNER_CYCLE, SCANNER_CYCLE_MAX_DIRECTORIES, SCANNER_CYCLE_MAX_DURATION,
-        SCANNER_CYCLE_MAX_OBJECTS, SCANNER_DELAY, SCANNER_IDLE_MODE, SCANNER_SPEED, SCANNER_SUB_SYS, ScannerSpeed,
+        ENV_SCANNER_CYCLE, ENV_SCANNER_CYCLE_MAX_OBJECTS, ENV_SCANNER_DELAY, ENV_SCANNER_MAX_WAIT_SECS, ENV_SCANNER_SPEED,
+        HEAL_BITROT_CYCLE, HEAL_SUB_SYS, SCANNER_BITROT_CYCLE, SCANNER_CACHE_SAVE_TIMEOUT, SCANNER_CYCLE,
+        SCANNER_CYCLE_MAX_DIRECTORIES, SCANNER_CYCLE_MAX_DURATION, SCANNER_CYCLE_MAX_OBJECTS, SCANNER_DELAY, SCANNER_IDLE_MODE,
+        SCANNER_SPEED, SCANNER_SUB_SYS, ScannerSpeed,
    };
    use std::collections::HashMap;
    use std::time::Duration;
@@ -969,50 +941,6 @@ mod tests {
        });
    }

-    #[test]
-    fn scanner_unset_budget_uses_safe_default_but_explicit_zero_is_unbounded() {
-        let config = server_config_with_scanner(&[]);
-        with_var_unset(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, || {
-            let resolved = lookup_scanner_runtime_config(Some(&config)).expect("scanner runtime config");
-            assert_eq!(resolved.cycle_budget.max_duration, Some(Duration::from_secs(1800)));
-            assert_eq!(resolved.cycle_max_duration_source, ScannerRuntimeConfigSource::Default);
-        });
-
-        let config = server_config_with_scanner(&[(SCANNER_CYCLE_MAX_DURATION, "0")]);
-        with_var_unset(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, || {
-            let resolved = lookup_scanner_runtime_config(Some(&config)).expect("scanner runtime config");
-            assert_eq!(resolved.cycle_budget.max_duration, None);
-            assert_eq!(resolved.cycle_max_duration_source, ScannerRuntimeConfigSource::Config);
-        });
-    }
-
-    #[test]
-    fn cycle_budget_invalid_or_overflow_config_is_rejected() {
-        with_var(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, Some("invalid"), || {
-            let error = lookup_scanner_runtime_config(None).expect_err("invalid duration env must be rejected");
-            assert!(error.to_string().contains(ENV_SCANNER_CYCLE_MAX_DURATION_SECS));
-            assert!(error.to_string().contains("<invalid>"));
-            assert!(!error.to_string().contains(": invalid ("));
-        });
-        with_var(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, Some("18446744073709551616"), || {
-            assert!(lookup_scanner_runtime_config(None).is_err());
-        });
-        with_var(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, Some("18446744073709551615"), || {
-            assert!(lookup_scanner_runtime_config(None).is_err());
-        });
-        let config = server_config_with_scanner(&[(SCANNER_CYCLE_MAX_DURATION, "not-a-duration")]);
-        assert!(lookup_scanner_runtime_config(Some(&config)).is_err());
-    }
-
-    #[test]
-    fn scanner_runtime_config_validation_rejects_overflow_persisted_duration() {
-        let config = server_config_with_scanner(&[(SCANNER_CYCLE_MAX_DURATION, "18446744073709551615")]);
-
-        let error = validate_scanner_runtime_config(&config)
-            .expect_err("persisted duration that exceeds the timer range must be rejected");
-        assert!(error.to_string().contains(SCANNER_CYCLE_MAX_DURATION));
-    }
-
    #[test]
    fn scanner_runtime_config_normalizes_persisted_default_speed() {
        let config = server_config_with_scanner(&[(SCANNER_SPEED, "default")]);
@@ -20,7 +20,7 @@ use std::sync::{Arc, LazyLock, RwLock};

 use crate::data_usage_define::{
    BACKGROUND_HEAL_INFO_PATH, DATA_USAGE_BLOOM_NAME_PATH, DATA_USAGE_OBJ_NAME_PATH, DATA_USAGE_OBSERVED_OBJ_NAME_PATH,
-    DataUsageCache, DataUsageCacheRevision, LEGACY_DATA_USAGE_OBJ_NAME_PATH, read_config_revision, read_config_with_revision,
+    DataUsageCache, DataUsageCacheRevision, LEGACY_DATA_USAGE_OBJ_NAME_PATH, read_config_with_revision,
 };
 use crate::runtime_config::{
    ScannerRuntimeConfig, ScannerRuntimeConfigSource, refresh_scanner_runtime_config_from_global, scanner_bitrot_cycle,
@@ -52,10 +52,11 @@ use rustfs_config::{
 };
 use rustfs_config::{ENV_SCANNER_CYCLE, ENV_SCANNER_SPEED, ENV_SCANNER_START_DELAY_SECS};
 use rustfs_data_usage::observed_data_usage_is_newer;
-use rustfs_lock::NamespaceLockGuard;
 use serde::{Deserialize, Serialize};
 use sha2::{Digest as _, Sha256};
-use tokio::sync::{Notify, mpsc};
+#[cfg(test)]
+use tokio::sync::Notify;
+use tokio::sync::mpsc;
 use tokio::time::{Duration, Instant};
 use tokio_util::sync::CancellationToken;
 use tokio_util::task::AbortOnDropHandle;
@@ -103,13 +104,6 @@ const CLEAN_IDLE_BACKOFF_FACTOR: u32 = 2;
 /// unavailable peer cannot drive a tight retry loop.
 const SCANNER_RETRY_BASE_INTERVAL: Duration = Duration::from_secs(5);
 const SCANNER_RETRY_MAX_INTERVAL: Duration = Duration::from_secs(30 * 60);
-/// A transient backend outage remains self-healing after the short retry
-/// budget is exhausted, but the probe is intentionally sparse until storage
-/// recovers or an operator reset wakes the scanner.
-const SCANNER_CYCLE_RECOVERY_PAUSED_INTERVAL: Duration = Duration::from_secs(5 * 60);
-/// Permanent recovery states still get a sparse status probe so a reset that
-/// races the wait registration cannot leave the scanner asleep forever.
-const SCANNER_CYCLE_RECOVERY_BLOCKED_PROBE_INTERVAL: Duration = Duration::from_secs(5 * 60);
 const SCANNER_LEADER_LOCK_POLL_INTERVAL: Duration = Duration::from_secs(1);
 #[cfg(not(test))]
 const SCANNER_LOCK_LOSS_SHUTDOWN_TIMEOUT: Duration = Duration::from_secs(30);
@@ -131,12 +125,6 @@ type ScannerCycleStatePersistTestHook = (u64, Arc<Notify>);
 static SCANNER_CYCLE_STATE_PERSIST_TEST_HOOK: LazyLock<StdMutex<Option<ScannerCycleStatePersistTestHook>>> =
    LazyLock::new(|| StdMutex::new(None));

-static SCANNER_CYCLE_RECOVERY_WAKE: LazyLock<Notify> = LazyLock::new(Notify::new);
-
-pub(super) fn notify_scanner_cycle_recovery_wake() {
-    SCANNER_CYCLE_RECOVERY_WAKE.notify_one();
-}
-
 #[cfg(test)]
 struct ScannerCycleStatePersistTestHookGuard;

@@ -588,21 +576,19 @@ pub async fn init_data_scanner(ctx: CancellationToken, storeapi: Arc<ECStore>) {
            tokio::time::sleep(sleep_time).await;
        }

-        let mut transient_backoff = ScannerRetryBackoff::default();
-        let mut recovery_retry_count = 0_u32;
        loop {
            if ctx_clone.is_cancelled() {
                break;
            }

-            let run_result = run_data_scanner_with_maintenance_state(
+            if let Err(e) = run_data_scanner_with_maintenance_state(
                ctx_clone.clone(),
                storeapi_clone.clone(),
                startup_features,
                startup_maintenance_generation,
            )
-            .await;
-            if let Err(e) = &run_result {
+            .await
+            {
                error!(
                    target: "rustfs::scanner",
                    event = EVENT_SCANNER_CYCLE_STATE,
@@ -613,52 +599,11 @@ pub async fn init_data_scanner(ctx: CancellationToken, storeapi: Arc<ECStore>) {
                    "Scanner runtime iteration failed"
                );
            }
-            let recovery_status = scanner_cycle_recovery_status();
-            if recovery_status.retryable {
-                recovery_retry_count = recovery_retry_count.saturating_add(1);
-                let _ = record_scanner_cycle_recovery_retry(recovery_retry_count);
-            } else {
-                recovery_retry_count = 0;
-            }
-
-            let recovery_status = scanner_cycle_recovery_status();
-            if recovery_status.state == "paused" {
-                transient_backoff.record_retryable_cycle(false);
-                tokio::select! {
-                    _ = ctx_clone.cancelled() => break,
-                    _ = SCANNER_CYCLE_RECOVERY_WAKE.notified() => {},
-                    _ = tokio::time::sleep(SCANNER_CYCLE_RECOVERY_PAUSED_INTERVAL) => {},
-                }
-                recovery_retry_count = 0;
-                continue;
-            }
-            if !recovery_status.retryable
-                && matches!(recovery_status.state.as_str(), "blocked" | "recovery-required" | "cleanup-pending")
-            {
-                transient_backoff.record_retryable_cycle(false);
-                tokio::select! {
-                    _ = ctx_clone.cancelled() => break,
-                    _ = SCANNER_CYCLE_RECOVERY_WAKE.notified() => {},
-                    _ = tokio::time::sleep(SCANNER_CYCLE_RECOVERY_BLOCKED_PROBE_INTERVAL) => {},
-                }
-                continue;
-            }
-
-            let retry_delay = if recovery_status.retryable || run_result.is_err() {
-                transient_backoff.record_retryable_cycle(true);
-                transient_backoff
-                    .retry_interval(scanner_cycle_interval())
-                    .unwrap_or(SCANNER_RETRY_BASE_INTERVAL)
-            } else {
-                transient_backoff.record_retryable_cycle(false);
-                randomized_cycle_delay()
-            };
            // Backoff before retrying after lock contention or scanner-level failures.
            // Keep this cancellation-aware so shutdown is not delayed by backoff sleep.
            tokio::select! {
                _ = ctx_clone.cancelled() => break,
-                _ = SCANNER_CYCLE_RECOVERY_WAKE.notified() => {},
-                _ = tokio::time::sleep(retry_delay) => {}
+                _ = tokio::time::sleep(randomized_cycle_delay()) => {}
            }
        }
    });
@@ -1038,116 +983,20 @@ fn data_usage_persist_timeout() -> Duration {
    DataUsageCache::persistence_timeout()
 }

-#[cfg(not(test))]
-const SCANNER_CYCLE_EPOCH_FENCE_TIMEOUT: Duration = Duration::from_secs(30);
-#[cfg(test)]
-const SCANNER_CYCLE_EPOCH_FENCE_TIMEOUT: Duration = Duration::from_millis(50);
-
-async fn fence_scanner_epoch_after_cycle_timeout<Store, LockLost>(
-    ctx: &CancellationToken,
-    storeapi: Arc<Store>,
-    cycle_info: &mut CurrentCycle,
-    cycle_revision: &mut DataUsageCacheRevision,
-    leader_epoch: &mut u64,
-    lock_lost: LockLost,
-) -> bool
-where
-    Store: ScannerObjectIO,
-    LockLost: Future<Output = ()>,
-{
-    let fence_ctx = ctx.child_token();
-    let claim = claim_scanner_leadership(&fence_ctx, storeapi, cycle_info, cycle_revision, leader_epoch);
-    tokio::pin!(claim);
-    tokio::pin!(lock_lost);
-    tokio::select! {
-        biased;
-        _ = &mut lock_lost => {
-            fence_ctx.cancel();
-            false
-        }
-        result = tokio::time::timeout(SCANNER_CYCLE_EPOCH_FENCE_TIMEOUT, &mut claim) => {
-            result.unwrap_or(false) && !fence_ctx.is_cancelled()
-        }
-    }
-}
-
-struct ScannerCycleDeadlineState<'a> {
-    cycle_info: &'a mut CurrentCycle,
-    cycle_revision: &'a mut DataUsageCacheRevision,
-    leader_epoch: &'a mut u64,
-    cycle_budget: &'a ScannerCycleBudget,
-}
-
-fn cycle_timeout_requires_recovery(worker_stopped: bool, cycle_state_persisted: bool, generation_fenced: bool) -> bool {
-    !worker_stopped || !cycle_state_persisted || !generation_fenced
-}
-
-async fn handle_scanner_cycle_deadline<Store>(
-    ctx: &CancellationToken,
-    storeapi: Arc<Store>,
-    state: ScannerCycleDeadlineState<'_>,
-    worker_stopped: bool,
-    guard: &mut NamespaceLockGuard,
-) where
-    Store: ScannerObjectIO,
-{
-    let fenced = fence_scanner_epoch_after_cycle_timeout(
-        ctx,
-        storeapi,
-        state.cycle_info,
-        state.cycle_revision,
-        state.leader_epoch,
-        guard.lock_lost_notified(),
-    )
-    .await;
-    let cycle_state_persisted = state.cycle_budget.cycle_state_persisted();
-    let recovery_required = cycle_timeout_requires_recovery(worker_stopped, cycle_state_persisted, fenced);
-    warn!(
-        target: "rustfs::scanner",
-        event = EVENT_SCANNER_CYCLE_STATE,
-        component = LOG_COMPONENT_SCANNER,
-        subsystem = LOG_SUBSYSTEM_RUNTIME,
-        state = "cycle_timeout",
-        worker_stopped,
-        cycle_state_persisted,
-        generation_fenced = fenced,
-        recovery_required,
-        "Scanner cycle deadline expired; durable cursor/generation fencing completed when possible"
-    );
-    global_metrics().record_scanner_cycle_timeout(recovery_required, state.cycle_budget.progress_age());
-    // Stop renewing before releasing the lease. A new leader can then claim the
-    // higher persisted generation instead of inheriting the expired worker.
-    guard.release();
-    global_metrics().set_cycle(None).await;
-}
-
 async fn mark_scan_cycle_idle(cycle_info: &mut CurrentCycle, cycle_metrics_guard: &mut ScannerCycleMetricsGuard) {
    cycle_info.current = 0;
    global_metrics().clear_current_scan_mode();
    cycle_metrics_guard.finish(cycle_info.clone()).await;
 }

-#[cfg(test)]
+#[instrument(skip_all)]
+#[hotpath::measure]
 async fn run_data_scanner_cycle(
    ctx: &CancellationToken,
    storeapi: &Arc<ECStore>,
    cycle_info: &mut CurrentCycle,
    cycle_revision: &mut DataUsageCacheRevision,
    leader_epoch: u64,
-) -> ScannerCycleOutcome {
-    let cycle_budget = ScannerCycleBudget::new(ctx, scanner_cycle_budget_config());
-    run_data_scanner_cycle_with_budget(ctx, storeapi, cycle_info, cycle_revision, leader_epoch, cycle_budget).await
-}
-
-#[instrument(skip_all)]
-#[hotpath::measure]
-async fn run_data_scanner_cycle_with_budget(
-    ctx: &CancellationToken,
-    storeapi: &Arc<ECStore>,
-    cycle_info: &mut CurrentCycle,
-    cycle_revision: &mut DataUsageCacheRevision,
-    leader_epoch: u64,
-    cycle_budget: Arc<ScannerCycleBudget>,
 ) -> ScannerCycleOutcome {
    let _activity_guard = ScannerActivityGuard::new();
    if let Err(err) = refresh_scanner_runtime_config_from_global() {
@@ -1163,11 +1012,7 @@ async fn run_data_scanner_cycle_with_budget(
    }
    let configured_cycle_interval = scanner_cycle_interval();
    let configured_bitrot_cycle = scanner_bitrot_cycle();
-    let cycle_budget_config = ScannerCycleBudgetConfig {
-        max_duration: cycle_budget.max_duration(),
-        max_objects: cycle_budget.max_objects(),
-        max_directories: cycle_budget.max_directories(),
-    };
+    let cycle_budget_config = scanner_cycle_budget_config();
    let usage_persist_timeout = data_usage_persist_timeout();
    global_metrics().record_scanner_cycle_config(
        configured_cycle_interval,
@@ -1238,6 +1083,7 @@ async fn run_data_scanner_cycle_with_budget(
    let (sender, receiver) = mpsc::channel::<DataUsageInfo>(1);

    let done_cycle = Metrics::time(Metric::ScanCycle);
+    let cycle_budget = ScannerCycleBudget::new(ctx, cycle_budget_config);
    let scan_result = storeapi
        .clone()
        .nsscanner_with_status(
@@ -1377,7 +1223,7 @@ async fn run_data_scanner_cycle_with_budget(
                "Scanner cycle is recovering to a newer durable cache generation"
            );
            emit_scan_cycle_partial_with_source(cycle_start.elapsed(), ScanCyclePartialReason::Unknown, None);
-            let persisted = persist_required_scanner_cycle_floor(
+            return if persist_required_scanner_cycle_floor(
                ctx,
                storeapi.clone(),
                cycle_info,
@@ -1386,9 +1232,8 @@ async fn run_data_scanner_cycle_with_budget(
                required_cycle,
                &mut cycle_metrics_guard,
            )
-            .await;
-            return if persisted {
-                cycle_budget.mark_cycle_state_persisted();
+            .await
+            {
                ScannerCycleOutcome::Partial
            } else {
                ScannerCycleOutcome::Failed
@@ -1446,7 +1291,7 @@ async fn run_data_scanner_cycle_with_budget(
            scan_cycle_partial_reason(budget_reason),
            scan_cycle_partial_source(budget_reason),
        );
-        let persisted = finalize_partial_scan_cycle(
+        return if finalize_partial_scan_cycle(
            ctx,
            storeapi.clone(),
            cycle_info,
@@ -1454,9 +1299,8 @@ async fn run_data_scanner_cycle_with_budget(
            leader_epoch,
            &mut cycle_metrics_guard,
        )
-        .await;
-        return if persisted {
-            cycle_budget.mark_cycle_state_persisted();
+        .await
+        {
            ScannerCycleOutcome::Partial
        } else {
            ScannerCycleOutcome::Failed
@@ -1531,7 +1375,7 @@ async fn run_data_scanner_cycle_with_budget(
                );
            }
            emit_scan_cycle_partial_with_source(cycle_start.elapsed(), ScanCyclePartialReason::Unknown, None);
-            let persisted = finalize_partial_scan_cycle(
+            return if finalize_partial_scan_cycle(
                ctx,
                storeapi.clone(),
                cycle_info,
@@ -1539,9 +1383,8 @@ async fn run_data_scanner_cycle_with_budget(
                leader_epoch,
                &mut cycle_metrics_guard,
            )
-            .await;
-            return if persisted {
-                cycle_budget.mark_cycle_state_persisted();
+            .await
+            {
                ScannerCycleOutcome::Partial
            } else {
                ScannerCycleOutcome::Failed
@@ -1582,7 +1425,6 @@ async fn run_data_scanner_cycle_with_budget(
            )
            .await
            {
-                cycle_budget.mark_cycle_state_persisted();
                emit_scan_cycle_superseded(cycle_start.elapsed());
                return ScannerCycleOutcome::Superseded;
            }
@@ -1615,7 +1457,6 @@ async fn run_data_scanner_cycle_with_budget(
        emit_scan_cycle_complete(false, cycle_start.elapsed());
        return ScannerCycleOutcome::Failed;
    }
-    cycle_budget.mark_cycle_state_persisted();

    done_cycle();
    emit_scan_cycle_complete(true, cycle_start.elapsed());
@@ -1680,7 +1521,7 @@ async fn run_data_scanner_with_maintenance_state(
 ) -> Result<(), ScannerError> {
    reset_scanner_cycle_schedule();
    // Acquire leader lock (write lock) to ensure only one scanner runs
-    let mut guard = match storeapi.new_ns_lock(RUSTFS_META_BUCKET, "leader.lock").await {
+    let guard = match storeapi.new_ns_lock(RUSTFS_META_BUCKET, "leader.lock").await {
        Ok(ns_lock) => match ns_lock.get_write_lock_quiet(get_lock_acquire_timeout()).await {
            Ok(guard) => {
                record_scanner_leader_lock_state("acquired");
@@ -1765,22 +1606,40 @@ async fn run_data_scanner_with_maintenance_state(
        observe_scanner_activity(&storeapi, distributed, &mut scanner_activity_seen).await;
    }

-    let (mut cycle_info, mut leader_epoch, mut cycle_revision) =
-        match load_scanner_cycle_state_for_startup(storeapi.clone()).await {
-            ScannerCycleStateStartup::Ready {
-                cycle,
-                leader_epoch,
-                revision,
-            } => (cycle, leader_epoch, revision),
-            ScannerCycleStateStartup::Blocked => {
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleStateStartup::Transient(err) => {
-                global_metrics().set_cycle(None).await;
-                return Err(err);
-            }
-        };
+    let (buf, mut cycle_revision) = match read_config_with_revision(storeapi.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str()).await {
+        Ok((buf, revision)) => (buf.unwrap_or_default(), revision),
+        Err(err) => {
+            error!(
+                target: "rustfs::scanner",
+                event = EVENT_SCANNER_PERSIST_STATE,
+                component = LOG_COMPONENT_SCANNER,
+                subsystem = LOG_SUBSYSTEM_RUNTIME,
+                path = %&*DATA_USAGE_BLOOM_NAME_PATH,
+                state = "revision_load_failed",
+                error = %err,
+                "Scanner cycle state revision load failed"
+            );
+            global_metrics().set_cycle(None).await;
+            return Ok(());
+        }
+    };
+    let (mut cycle_info, mut leader_epoch) = match decode_scanner_cycle_state_for_startup(&buf) {
+        Ok(state) => state,
+        Err(err) => {
+            error!(
+                target: "rustfs::scanner",
+                event = EVENT_SCANNER_PERSIST_STATE,
+                component = LOG_COMPONENT_SCANNER,
+                subsystem = LOG_SUBSYSTEM_RUNTIME,
+                path = %&*DATA_USAGE_BLOOM_NAME_PATH,
+                state = "cycle_decode_failed",
+                error = %err,
+                "Scanner stopped because persisted cycle state is invalid"
+            );
+            global_metrics().set_cycle(None).await;
+            return Ok(());
+        }
+    };
    let usage_floor = match persisted_usage_floor(storeapi.clone()).await {
        Ok(floor) => floor,
        Err(err) => {
@@ -1845,49 +1704,13 @@ async fn run_data_scanner_with_maintenance_state(
            return Ok(());
        }
        let cycle_ctx = ctx.child_token();
-        let cycle_budget = ScannerCycleBudget::new_with_runtime_progress_tracking(&cycle_ctx, scanner_cycle_budget_config());
-        let initial_outcome = match await_scanner_cycle_with_budget_fence(
+        let initial_outcome = await_scanner_cycle_with_lock_fence(
            &cycle_ctx,
-            &cycle_budget,
-            run_data_scanner_cycle_with_budget(
-                &cycle_ctx,
-                &storeapi,
-                &mut cycle_info,
-                &mut cycle_revision,
-                leader_epoch,
-                cycle_budget.clone(),
-            ),
+            run_data_scanner_cycle(&cycle_ctx, &storeapi, &mut cycle_info, &mut cycle_revision, leader_epoch),
            guard.lock_lost_notified(),
        )
        .await
-        {
-            ScannerCycleWaitOutcome::Completed(outcome) => outcome,
-            ScannerCycleWaitOutcome::LockLost => {
-                record_scanner_leader_lock_lost("Scanner leader lock lost during the initial cycle").await;
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleWaitOutcome::Cancelled => {
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleWaitOutcome::Deadline { worker_stopped } => {
-                handle_scanner_cycle_deadline(
-                    &ctx,
-                    storeapi.clone(),
-                    ScannerCycleDeadlineState {
-                        cycle_info: &mut cycle_info,
-                        cycle_revision: &mut cycle_revision,
-                        leader_epoch: &mut leader_epoch,
-                        cycle_budget: &cycle_budget,
-                    },
-                    worker_stopped,
-                    &mut guard,
-                )
-                .await;
-                return Ok(());
-            }
-        };
+        .unwrap_or(ScannerCycleOutcome::Failed);
        superseded_backoff.record_retryable_cycle(initial_outcome == ScannerCycleOutcome::Superseded);
        deferred_backoff.record_retryable_cycle(matches!(initial_outcome, ScannerCycleOutcome::Deferred(_)));
        dirty_usage_generation_seen = dirty_generation_before_cycle;
@@ -2093,49 +1916,13 @@ async fn run_data_scanner_with_maintenance_state(
        }
        let dirty_generation_before_cycle = dirty_usage_generation();
        let cycle_ctx = ctx.child_token();
-        let cycle_budget = ScannerCycleBudget::new_with_runtime_progress_tracking(&cycle_ctx, scanner_cycle_budget_config());
-        let outcome = match await_scanner_cycle_with_budget_fence(
+        let outcome = await_scanner_cycle_with_lock_fence(
            &cycle_ctx,
-            &cycle_budget,
-            run_data_scanner_cycle_with_budget(
-                &cycle_ctx,
-                &storeapi,
-                &mut cycle_info,
-                &mut cycle_revision,
-                leader_epoch,
-                cycle_budget.clone(),
-            ),
+            run_data_scanner_cycle(&cycle_ctx, &storeapi, &mut cycle_info, &mut cycle_revision, leader_epoch),
            guard.lock_lost_notified(),
        )
        .await
-        {
-            ScannerCycleWaitOutcome::Completed(outcome) => outcome,
-            ScannerCycleWaitOutcome::LockLost => {
-                record_scanner_leader_lock_lost("Scanner leader lock lost during a scanner cycle").await;
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleWaitOutcome::Cancelled => {
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleWaitOutcome::Deadline { worker_stopped } => {
-                handle_scanner_cycle_deadline(
-                    &ctx,
-                    storeapi.clone(),
-                    ScannerCycleDeadlineState {
-                        cycle_info: &mut cycle_info,
-                        cycle_revision: &mut cycle_revision,
-                        leader_epoch: &mut leader_epoch,
-                        cycle_budget: &cycle_budget,
-                    },
-                    worker_stopped,
-                    &mut guard,
-                )
-                .await;
-                return Ok(());
-            }
-        };
+        .unwrap_or(ScannerCycleOutcome::Failed);
        superseded_backoff.record_retryable_cycle(outcome == ScannerCycleOutcome::Superseded);
        deferred_backoff.record_retryable_cycle(matches!(outcome, ScannerCycleOutcome::Deferred(_)));
        dirty_usage_generation_seen = dirty_generation_before_cycle;
@@ -2432,12 +2219,7 @@ pub(crate) use activity::{
 pub(crate) use activity::{ScannerCycleOutcome, scanner_cycle_outcome_with_pending_maintenance};
 #[cfg(test)]
 pub(crate) use cycle_state::encode_scanner_cycle_fence_for_test;
-pub use cycle_state::{
-    ScannerCycleRecoveryMarker, ScannerCycleRecoveryStatus, reset_scanner_cycle_recovery, scanner_cycle_recovery_status,
-};
-pub(crate) use cycle_state::{
-    current_scanner_leader_epoch, decode_persisted_scanner_cycle_fence, load_scanner_cycle_state_for_startup,
-};
+pub(crate) use cycle_state::{current_scanner_leader_epoch, decode_persisted_scanner_cycle_fence};
 pub use heal_info::{BackgroundHealInfo, read_background_heal_info, save_background_heal_info};
 pub use usage_store::store_data_usage_in_backend;

@@ -196,7 +196,7 @@ pub(super) async fn claim_scanner_leadership(
        if ctx.is_cancelled() {
            return false;
        }
-        let Some(claimed_epoch) = persisted_epoch.checked_add(1).filter(|epoch| *epoch < u64::MAX) else {
+        let Some(claimed_epoch) = persisted_epoch.checked_add(1) else {
            error!(
                target: "rustfs::scanner",
                event = EVENT_SCANNER_PERSIST_STATE,
@@ -14,16 +14,17 @@

 use std::sync::{
    Arc,
-    atomic::{AtomicBool, AtomicU8, AtomicU64, Ordering},
+    atomic::{AtomicU8, AtomicU64, Ordering},
 };
-use tokio::time::{Duration, Instant};
+use std::time::Instant;
+
+use tokio::time::Duration;
 use tokio_util::sync::CancellationToken;

 const BUDGET_REASON_NONE: u8 = 0;
 const BUDGET_REASON_RUNTIME: u8 = 1;
 const BUDGET_REASON_OBJECTS: u8 = 2;
 const BUDGET_REASON_DIRECTORIES: u8 = 3;
-const PROGRESS_CLOCK_SAMPLE_INTERVAL: u64 = 128;

 #[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
 pub(crate) struct ScannerCycleBudgetConfig {
@@ -62,51 +63,29 @@ pub struct ScannerCycleBudget {
    token: CancellationToken,
    reason: Arc<AtomicU8>,
    started_at: Instant,
-    deadline: Option<Instant>,
    max_duration: Option<Duration>,
    max_objects: Option<u64>,
    max_directories: Option<u64>,
    track_progress: bool,
-    track_unbounded_counts: bool,
    objects_scanned: AtomicU64,
    directories_started: AtomicU64,
    entries_visited: AtomicU64,
-    last_progress_millis: AtomicU64,
-    cycle_state_persisted: AtomicBool,
 }

 impl ScannerCycleBudget {
-    #[cfg(test)]
    pub(crate) fn new(parent: &CancellationToken, config: ScannerCycleBudgetConfig) -> Arc<Self> {
-        Self::new_inner(parent, config, false, false)
+        Self::new_inner(parent, config, false)
    }

    pub(crate) fn new_with_progress_tracking(parent: &CancellationToken, config: ScannerCycleBudgetConfig) -> Arc<Self> {
-        Self::new_inner(parent, config, true, true)
+        Self::new_inner(parent, config, true)
    }

-    pub(crate) fn new_with_runtime_progress_tracking(parent: &CancellationToken, config: ScannerCycleBudgetConfig) -> Arc<Self> {
-        let track_progress = config.max_duration.is_some();
-        Self::new_inner(parent, config, track_progress, false)
-    }
-
-    fn new_inner(
-        parent: &CancellationToken,
-        config: ScannerCycleBudgetConfig,
-        track_progress: bool,
-        track_unbounded_counts: bool,
-    ) -> Arc<Self> {
+    fn new_inner(parent: &CancellationToken, config: ScannerCycleBudgetConfig, track_progress: bool) -> Arc<Self> {
        let token = parent.child_token();
        let reason = Arc::new(AtomicU8::new(BUDGET_REASON_NONE));
-        let started_at = Instant::now();
-        let deadline = config.max_duration.map(|duration| match started_at.checked_add(duration) {
-            Some(deadline) => deadline,
-            // Runtime config rejects this range, but keep programmatic callers
-            // fail-closed instead of panicking or silently disabling the wall clock.
-            None => started_at,
-        });

-        if let Some(deadline) = deadline {
+        if let Some(duration) = config.max_duration {
            let parent = parent.clone();
            let token_wait = token.clone();
            let token_cancel = token.clone();
@@ -115,7 +94,7 @@ impl ScannerCycleBudget {
                tokio::select! {
                    _ = parent.cancelled() => {}
                    _ = token_wait.cancelled() => {}
-                    _ = tokio::time::sleep_until(deadline) => {
+                    _ = tokio::time::sleep(duration) => {
                        Self::cancel_for_reason(&reason, &token_cancel, ScannerCycleBudgetReason::Runtime);
                    }
                }
@@ -125,18 +104,14 @@ impl ScannerCycleBudget {
        Arc::new(Self {
            token,
            reason,
-            started_at,
-            deadline,
+            started_at: Instant::now(),
            max_duration: config.max_duration,
            max_objects: config.max_objects,
            max_directories: config.max_directories,
            track_progress,
-            track_unbounded_counts,
            objects_scanned: AtomicU64::new(0),
            directories_started: AtomicU64::new(0),
            entries_visited: AtomicU64::new(0),
-            last_progress_millis: AtomicU64::new(0),
-            cycle_state_persisted: AtomicBool::new(false),
        })
    }

@@ -156,14 +131,6 @@ impl ScannerCycleBudget {
        self.max_duration
    }

-    pub(crate) fn deadline(&self) -> Option<Instant> {
-        self.deadline
-    }
-
-    pub(crate) fn cancel_for_runtime(&self) {
-        self.cancel_for(ScannerCycleBudgetReason::Runtime);
-    }
-
    pub(crate) fn max_objects(&self) -> Option<u64> {
        self.max_objects
    }
@@ -206,43 +173,15 @@ impl ScannerCycleBudget {
        self.entries_visited.load(Ordering::Relaxed)
    }

-    pub(crate) fn mark_cycle_state_persisted(&self) {
-        self.cycle_state_persisted.store(true, Ordering::Release);
-    }
-
-    pub(crate) fn cycle_state_persisted(&self) -> bool {
-        self.cycle_state_persisted.load(Ordering::Acquire)
-    }
-
-    pub(crate) fn progress_age(&self) -> Duration {
-        let elapsed_millis = u64::try_from(self.started_at.elapsed().as_millis()).unwrap_or(u64::MAX);
-        let last_progress = self.last_progress_millis.load(Ordering::Relaxed);
-        Duration::from_millis(elapsed_millis.saturating_sub(last_progress))
-    }
-
-    fn record_progress_sample(&self, event: u64) {
-        // Clock reads are sampled at batch/count boundaries; the scanner's
-        // per-object path does not add a second progress atomic.
-        if event == 0 || (event != 1 && !event.is_multiple_of(PROGRESS_CLOCK_SAMPLE_INTERVAL)) {
-            return;
-        }
-        let elapsed_millis = u64::try_from(self.started_at.elapsed().as_millis()).unwrap_or(u64::MAX);
-        self.last_progress_millis.store(elapsed_millis, Ordering::Relaxed);
-    }
-
    pub(crate) fn record_entries_visited(&self, entries_visited: u64) {
        if self.track_progress {
-            let entries = saturating_fetch_add(&self.entries_visited, entries_visited);
-            self.record_progress_sample(entries);
+            saturating_fetch_add(&self.entries_visited, entries_visited);
        }
    }

    pub(crate) fn record_remote_progress(&self, objects_scanned: u64, directories_started: u64) {
        if self.track_progress || self.max_objects.is_some() {
            let objects = saturating_fetch_add(&self.objects_scanned, objects_scanned);
-            if self.track_progress {
-                self.record_progress_sample(objects);
-            }
            if self.max_objects.is_some_and(|max_objects| objects >= max_objects) {
                self.cancel_for(ScannerCycleBudgetReason::Objects);
            }
@@ -250,12 +189,9 @@ impl ScannerCycleBudget {

        if self.track_progress || self.max_directories.is_some() {
            let directories = saturating_fetch_add(&self.directories_started, directories_started);
-            if self.track_progress {
-                self.record_progress_sample(directories);
-            }
            if self
                .max_directories
-                .is_some_and(|max_directories| directory_budget_exhausted(directories, max_directories))
+                .is_some_and(|max_directories| directories > max_directories)
            {
                self.cancel_for(ScannerCycleBudgetReason::Directories);
            }
@@ -271,17 +207,14 @@ impl ScannerCycleBudget {
    }

    pub(crate) fn try_start_directory(&self) -> bool {
-        if self.max_directories.is_none() && !self.track_unbounded_counts {
+        if !self.track_progress && self.max_directories.is_none() {
            return true;
        }

        let directories = saturating_fetch_add(&self.directories_started, 1);
-        if self.track_progress {
-            self.record_progress_sample(directories);
-        }
        if self
            .max_directories
-            .is_some_and(|max_directories| directory_budget_exhausted(directories, max_directories))
+            .is_some_and(|max_directories| directories > max_directories)
        {
            self.cancel_for(ScannerCycleBudgetReason::Directories);
            return false;
@@ -291,14 +224,11 @@ impl ScannerCycleBudget {
    }

    pub(crate) fn record_object_scanned(&self) {
-        if self.max_objects.is_none() && !self.track_unbounded_counts {
+        if !self.track_progress && self.max_objects.is_none() {
            return;
        }

        let objects = saturating_fetch_add(&self.objects_scanned, 1);
-        if self.track_progress {
-            self.record_progress_sample(objects);
-        }
        if self.max_objects.is_some_and(|max_objects| objects >= max_objects) {
            self.cancel_for(ScannerCycleBudgetReason::Objects);
        }
@@ -329,13 +259,6 @@ fn saturating_fetch_add(value: &AtomicU64, delta: u64) -> u64 {
    }
 }

-fn directory_budget_exhausted(directories: u64, max_directories: u64) -> bool {
-    // Saturation hides a remote max+1 update when the configured limit is the
-    // largest representable counter. Treat that boundary as exhausted rather
-    // than allowing work to continue indefinitely.
-    directories > max_directories || (directories == u64::MAX && max_directories == u64::MAX)
-}
-
 impl Drop for ScannerCycleBudget {
    fn drop(&mut self) {
        self.token.cancel();
@@ -478,35 +401,6 @@ mod tests {
        assert_eq!(directory_budget.reason(), Some(ScannerCycleBudgetReason::Directories));
    }

-    #[test]
-    fn directory_budget_fails_closed_when_progress_saturates() {
-        let parent = CancellationToken::new();
-        let budget = ScannerCycleBudget::new(
-            &parent,
-            ScannerCycleBudgetConfig {
-                max_directories: Some(u64::MAX),
-                ..Default::default()
-            },
-        );
-
-        budget.record_remote_progress(0, u64::MAX);
-
-        assert_eq!(budget.reason(), Some(ScannerCycleBudgetReason::Directories));
-        assert!(budget.token().is_cancelled());
-
-        let local_budget = ScannerCycleBudget::new(
-            &parent,
-            ScannerCycleBudgetConfig {
-                max_directories: Some(u64::MAX),
-                ..Default::default()
-            },
-        );
-        local_budget.record_remote_progress(0, u64::MAX - 1);
-        assert!(!local_budget.budget_elapsed());
-        assert!(!local_budget.try_start_directory());
-        assert_eq!(local_budget.reason(), Some(ScannerCycleBudgetReason::Directories));
-    }
-
    #[test]
    fn explicit_progress_tracking_counts_unbounded_remote_work_without_cancelling() {
        let parent = CancellationToken::new();
@@ -567,29 +461,4 @@ mod tests {
        assert!(object_limited.requires_serial_progress_accounting());
        assert!(directory_limited.requires_serial_progress_accounting());
    }
-
-    #[tokio::test(start_paused = true)]
-    async fn progress_age_uses_virtual_time_and_sampled_progress() {
-        let parent = CancellationToken::new();
-        let budget = ScannerCycleBudget::new_with_runtime_progress_tracking(
-            &parent,
-            ScannerCycleBudgetConfig {
-                max_duration: Some(Duration::from_secs(60)),
-                ..Default::default()
-            },
-        );
-
-        tokio::time::advance(Duration::from_secs(5)).await;
-        assert_eq!(budget.progress_age(), Duration::from_secs(5));
-        budget.record_entries_visited(1);
-        assert_eq!(budget.progress_age(), Duration::ZERO);
-
-        tokio::time::advance(Duration::from_secs(2)).await;
-        for _ in 0..126 {
-            budget.record_entries_visited(1);
-        }
-        assert_eq!(budget.progress_age(), Duration::from_secs(2));
-        budget.record_entries_visited(1);
-        assert_eq!(budget.progress_age(), Duration::ZERO);
-    }
 }
@@ -48,7 +48,6 @@ use time::OffsetDateTime;
 use tokio::sync::{Mutex, Notify, Semaphore, mpsc};
 use tokio::time::Duration;
 use tokio_util::sync::CancellationToken;
-use tokio_util::task::AbortOnDropHandle;
 use tracing::{debug, error, warn};

 use crate::ScannerObjectInfo as ObjectInfo;
@@ -314,7 +314,7 @@ impl ScannerIOCache for SetDisks {
        let ctx_clone = ctx.clone();
        let completed_bucket_count = Arc::new(AtomicUsize::new(0));
        let completed_bucket_count_clone = completed_bucket_count.clone();
-        let collect_bucket_results_fut = AbortOnDropHandle::new(tokio::spawn(async move {
+        let collect_bucket_results_fut = tokio::spawn(async move {
            let mut cancelled = false;

            loop {
@@ -333,7 +333,7 @@ impl ScannerIOCache for SetDisks {
                    }
                }
            }
-        }));
+        });

        let mut futs = Vec::new();

@@ -365,7 +365,7 @@ impl ScannerIOCache for SetDisks {
                NamespaceScannerWorkerMode::RemoteV4(server_epoch) => Some(server_epoch),
                NamespaceScannerWorkerMode::Coordinator => None,
            };
-            futs.push(AbortOnDropHandle::new(tokio::spawn(async move {
+            futs.push(tokio::spawn(async move {
                let remote_session_id = uuid::Uuid::new_v4();
                let mut remote_session_sequence = 0_u64;
                loop {
@@ -1038,7 +1038,7 @@ impl ScannerIOCache for SetDisks {
                        );
                    }
                }
-            })));
+            }));
        }
        drop(bucket_tx);
        drop(bucket_result_tx);
@@ -242,7 +242,7 @@ impl ScannerIOCycle for ECStore {
                        results[results_index_clone] = result;
                    }
                });
-                wait_futs.push(AbortOnDropHandle::new(receiver_fut));
+                wait_futs.push(receiver_fut);

                let scan_plan = ScannerBucketScanPlan {
                    buckets: set_buckets,
@@ -318,7 +318,7 @@ impl ScannerIOCycle for ECStore {
                        record_set_scan_failure(&mut first_err, e);
                    }
                });
-                wait_futs.push(AbortOnDropHandle::new(scanner_fut));
+                wait_futs.push(scanner_fut);
            }
        }

@@ -13,7 +13,7 @@
 // limitations under the License.

 use super::*;
-use rustfs_data_usage::{ReplicationAllStats, ReplicationTargetUsage};
+use rustfs_data_usage::{ReplicationAllStats, ReplicationStats};

 const TEST_PLAN_DIGEST: DataUsageScanPlanDigest = DataUsageScanPlanDigest([7; 32]);

@@ -271,7 +271,7 @@ fn completed_data_usage_info_flattens_nested_bucket_entries() {
        replication_stats: Some(ReplicationAllStats {
            targets: HashMap::from([(
                "arn:target".to_string(),
-                ReplicationTargetUsage {
+                ReplicationStats {
                    replicated_size: 2048,
                    replicated_count: 2,
                    ..Default::default()
@@ -76,7 +76,6 @@ pub use bucket::{BucketInfo, BucketOperations, BucketOptions, DeleteBucketOption
 pub use capability::{CapabilitySnapshotError, CapabilityState, CapabilityStatus};
 pub use error::{StorageErrorCode, StorageResult};
 pub use multipart::{CompletePart, ListMultipartsInfo, ListPartsInfo, MultipartInfo, MultipartUploadResult, PartInfo};
-pub use object::DeleteAccounting;
 pub use object::ObjectLockDeleteOptions;
 pub use object::{DeletedObject, ObjectToDelete};
 pub use object::{ExpirationOptions, TransitionedObject};
@@ -218,17 +218,6 @@ pub struct DeletedObject {
    pub force_delete_generation: Option<i64>,
 }

-/// Accounting identity returned by the internal commit-time delete path.
-///
-/// This is carried separately from [`DeletedObject`] so adding quota details
-/// does not change the source shape of the public S3 delete result contract.
-#[derive(Debug, Default, Clone, PartialEq, Eq)]
-pub struct DeleteAccounting {
-    pub size: Option<u64>,
-    pub version_id: Option<Uuid>,
-    pub removed_current_object: bool,
-}
-
 impl DeletedObject {
    pub fn version_purge_status(&self) -> VersionPurgeStatusType {
        self.replication_state
@@ -352,19 +341,6 @@ pub trait ObjectOperations: Send + Sync + fmt::Debug {
        objects: Vec<Self::ObjectToDelete>,
        opts: Self::ObjectOptions,
    ) -> (Vec<Self::DeletedObject>, Vec<Option<Self::Error>>);
-    /// Delete objects and optionally return commit-time accounting identities.
-    /// The default preserves the ordinary delete contract for implementations
-    /// that do not expose storage-level accounting details.
-    async fn delete_objects_with_accounting(
-        &self,
-        bucket: &str,
-        objects: Vec<Self::ObjectToDelete>,
-        opts: Self::ObjectOptions,
-    ) -> (Vec<Self::DeletedObject>, Vec<Option<Self::Error>>, Vec<Option<DeleteAccounting>>) {
-        let object_count = objects.len();
-        let (deleted, errors) = self.delete_objects(bucket, objects, opts).await;
-        (deleted, errors, vec![None; object_count])
-    }
    async fn put_object_metadata(
        &self,
        bucket: &str,
@@ -268,7 +268,7 @@ where
        .parse::<T>()
        .map_err(|_| {
            log_once(&format!("env_invalid_value:{used_key}"), || {
-                format!("Invalid {} value for {used_key}. Treating as unset.", type_name::<T>())
+                format!("Invalid {} value for {used_key}: {value}. Treating as unset.", type_name::<T>())
            });
        })
        .ok()
@@ -570,7 +570,7 @@ where
        Ok(parsed) => EnvParseOutcome::Parsed(parsed),
        Err(_) => {
            log_once(&format!("env_invalid_value:{used_key}"), || {
-                format!("Invalid {} value for {used_key}. Treating as unset.", type_name::<T>())
+                format!("Invalid {} value for {used_key}: {value}. Treating as unset.", type_name::<T>())
            });
            EnvParseOutcome::Invalid
        }
@@ -12,7 +12,6 @@ for later deletion.

 ## Open Items

- `rustfs-6339` legacy bucket policy ID casing: earlier RustFS releases persisted the top-level policy identifier as "ID", while current writes use the S3-compatible "Id" spelling. Readers accept both spellings so retained bucket metadata remains usable after upgrade. Remove the legacy alias after migration tooling has rewritten every retained bucket policy using "ID".
 - `table-publication-fence-v1` table publication fencing: nodes that predate table and table-bucket publication fences can mutate live files while a new node is publishing a catalog pointer. New nodes retain exact object guards until the operator confirms that every serving node uses the new fences. Fleet confirmation also requires non-overlapping active warehouse prefixes and lifecycle workers that exclude table buckets. Remove the exact live-file fallback and the fleet-confirmation gate after the minimum supported RustFS release acquires table fences for registered-table mutations and table-bucket fences for unresolved-prefix mutations.
 - `table-catalog-strong-snapshot-v1` durable strong catalog snapshot compatibility: version 1 writes continue during mixed-version rollout until operators confirm that every serving node reads version 2, and version 1 table/view identifier collisions remain available only for cleanup. Remove version 1 writes and collision cleanup after the minimum supported RustFS release reads version 2 and every retained durable strong snapshot is collision-free and has been upgraded to version 2.
 - `table-catalog-migration-fence-v1` durable strong migration fence compatibility: version 1 "PREPARING" fences did not distinguish a known-absent global strong snapshot from an unknown baseline, so retries read them but fail closed if the global snapshot is missing. Version 2 preserves the same JSON shape and records the pre-migration global snapshot ETag in the existing target_snapshot_etag field while the fence is "PREPARING". Remove version 1 reads after every supported direct-upgrade source writes version 2 fences and operators have completed or cancelled every older in-progress backing migration.
@@ -52,7 +52,7 @@ The `/v3/scanner/status` response reports each effective runtime value with a
 | `scanner.max_wait` | `RUSTFS_SCANNER_MAX_WAIT_SECS` | seconds | preset-derived | Caps one scanner sleep. |
 | `scanner.cycle` | `RUSTFS_SCANNER_CYCLE` | seconds | preset-derived | Sets the interval between scanner cycles. |
 | `scanner.start_delay` | `RUSTFS_SCANNER_START_DELAY_SECS` | seconds | unset | Sets startup delay and, for compatibility, the cycle interval when `scanner.cycle` is unset. |
-| `scanner.cycle_max_duration` | `RUSTFS_SCANNER_CYCLE_MAX_DURATION_SECS` | seconds | `1800` | Caps one cycle's runtime. An explicit `0` disables this budget. |
+| `scanner.cycle_max_duration` | `RUSTFS_SCANNER_CYCLE_MAX_DURATION_SECS` | seconds | `0` | Caps one cycle's runtime. `0` disables this budget. |
 | `scanner.cycle_max_objects` | `RUSTFS_SCANNER_CYCLE_MAX_OBJECTS` | objects | `0` | Caps objects processed by one cycle. `0` disables this budget. |
 | `scanner.cycle_max_directories` | `RUSTFS_SCANNER_CYCLE_MAX_DIRECTORIES` | directories | `0` | Caps directories entered by one cycle. `0` disables this budget. |
 | `heal.bitrot_cycle` | `RUSTFS_SCANNER_BITROT_CYCLE_SECS` | seconds | `2592000` | Controls periodic deep bitrot scans. `false`, `off`, `no`, or `disabled` disables periodic deep scans; `0`, `true`, `on`, or `yes` runs deep mode every scanner cycle. |
@@ -70,21 +70,6 @@ sleep multiplier, maximum wait, and cycle interval. Use `scanner.delay`,
 `scanner.max_wait`, and `scanner.cycle` when the preset is close but one axis
 needs a precise override.

-When the cycle duration control is unset, RustFS uses a finite 1800-second
-(30-minute) default, matching the scanner benchmark guidance. An explicit `0`
-preserves the compatibility behavior of an unbounded cycle; object and
-directory budgets likewise remain unbounded when explicitly set to `0`. Invalid
-or overflowing duration environment values are configuration errors rather than
-silent fallback values.
-
-When a finite deadline expires, RustFS cancels cooperative scanner work and
-waits only for the existing bounded shutdown window. A non-yielding I/O future
-is dropped after that window. RustFS then attempts a higher leadership epoch so
-late cycle, usage, cache, and remote writes from the old generation fail closed.
-If the worker cannot stop cooperatively, the cycle state was not confirmed
-durable, or that epoch fence cannot be durably persisted, the scanner reports
-`recovery-required`; it does not claim an uncooperative cursor was saved.
-
 An explicit `scanner.cycle` or `RUSTFS_SCANNER_CYCLE` is a minimum inter-cycle
 cadence: dirty-usage notifications do not bypass that configured interval.
 The default adaptive policy continues to use dirty-usage notifications to wake
@@ -159,10 +144,6 @@ metrics.maintenance_control.primary_control
 metrics.source_work
 metrics.replication_repair
 metrics.scan_checkpoint
-metrics.cycle_timeout_total
-metrics.cycle_last_progress_age
-metrics.leader_lease_without_progress
-metrics.cycle_recovery_required_total
 ```

 ## Reading Pacing Pressure
@@ -158,11 +158,10 @@ added by backlog#1153 infra-4.

 ## Coverage

-Workspace line coverage is measured weekly. Pull requests that touch iam, kms,
-policy, or crypto also run a non-required, report-only comparison against
-`.config/coverage-baselines.toml`. During calibration, a regression is recorded
-in the job summary without failing the job; missing or malformed coverage
-evidence still fails closed (backlog#1153 infra-6).
+Line coverage is measured **weekly, not per-PR**, and is non-blocking: it
+exists for visibility and trend, never as a required check. Per-crate ratchets
+for the security-critical crates (iam / kms / policy / crypto) build on this
+baseline later (backlog#1153 infra-6, report-only first).

 - **CI**: `.github/workflows/coverage.yml` runs every Sunday and on manual
  dispatch: `cargo llvm-cov nextest --workspace --exclude e2e_test` under the
@@ -175,12 +174,6 @@ evidence still fails closed (backlog#1153 infra-6).
  plus the full suite). It prints the same per-crate table via
  `scripts/coverage_per_crate.py` and writes `target/llvm-cov/lcov.info` and
  `coverage.json`.
- **Security-critical ratchet**: relevant pull requests compare iam / kms /
-  policy / crypto line coverage with the versioned baseline. Drops greater than
-  the configured one-percentage-point calibration threshold are marked
-  `REGRESSION (report-only)`. The weekly summary runs the same comparison so
-  calibration continues even when no relevant pull request is open. Baseline
-  changes require a linked coverage run and a reviewed explanation.
 - **Trend comparison**: each run's job summary is the weekly per-crate
  snapshot — open two runs from the Actions history (workflow "coverage") and
  compare their tables. For line-level diffs, download the two runs'
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
overtrue	7d2dffe138	ci(s3tests): omit proxy query strings	2026-08-22 19:15:55 +08:00
overtrue	cbeb191e99	ci(s3tests): add targeted proxy diagnostics	2026-08-22 18:53:25 +08:00
overtrue	ed424e2712	fix(app): wait for peer bucket metadata reload	2026-08-22 18:52:24 +08:00