test(kms): validate Vault circuit recovery

* fix(scanner): isolate corrupt cycle state

* fix(scanner): preserve newer state during recovery reset

* fix(scanner): fence recovery reset state

* fix(scanner): reject terminal recovery epochs

* fix(scanner): reject trailing cycle state bytes

* fix(scanner): reject terminal leadership epochs

* fix(scanner): retain recovery wake notifications

* fix(scanner): recover from oversized markers

.github/workflows/ci.yml

@@ -400,7 +400,7 @@ jobs:
     if: github.event_name != 'pull_request' || github.event.action != 'closed'
     needs: [ quick-checks ]
     runs-on: sm-standard-4
-    timeout-minutes: 45
+    timeout-minutes: 90
     env:
       FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
     steps:
@@ -440,7 +440,7 @@ jobs:
     if: github.event_name != 'pull_request' || github.event.action != 'closed'
     needs: [ quick-checks ]
     runs-on: sm-standard-4
-    timeout-minutes: 60
+    timeout-minutes: 90
     env:
       FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
     steps:
@@ -470,7 +470,7 @@ jobs:
     if: github.event_name != 'pull_request' || github.event.action != 'closed'
     needs: [ quick-checks ]
     runs-on: sm-standard-4
-    timeout-minutes: 60
+    timeout-minutes: 90
     strategy:
       # On a PR, one failing protocol leg is enough to know the PR is not ready,
       # so stop the sibling leg instead of paying another ~40 minutes for it.

.github/workflows/nightly-gnu.yml

@@ -39,11 +39,10 @@ jobs:
     env:
       FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
     steps:
-      - name: Checkout main branch
+      - name: Checkout repository
         uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
         with:
           persist-credentials: false
-          ref: main
 
       - name: Setup Rust environment
         uses: ./.github/actions/setup
@@ -89,11 +88,10 @@ jobs:
       # either casing.
       NO_PROXY: 127.0.0.1,localhost
     steps:
-      - name: Checkout main branch
+      - name: Checkout repository
         uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
         with:
           persist-credentials: false
-          ref: main
 
       - name: Setup Rust environment
         uses: ./.github/actions/setup
@@ -178,11 +176,10 @@ jobs:
       FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
       NO_PROXY: 127.0.0.1,localhost
     steps:
-      - name: Checkout main branch
+      - name: Checkout repository
         uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
         with:
           persist-credentials: false
-          ref: main
 
       - name: Setup Rust environment
         uses: ./.github/actions/setup

Cargo.lock

@@ -9587,7 +9587,6 @@ dependencies = [
  "serde",
  "serde_json",
  "serial_test",
- "sha2 0.11.0",
  "temp-env",
  "tempfile",
  "thiserror 2.0.20",

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

@@ -50,10 +50,10 @@ use rustfs_protos::evict_failed_connection;
 use rustfs_protos::proto_gen::node_service::RenamePartRequest;
 use rustfs_protos::proto_gen::node_service::{
     BatchReadVersionRequest, BatchReadVersionResponse, CheckPartsRequest, DeletePathsRequest, DeleteRequest,
-    DeleteVersionRequest, DeleteVersionsRequest, DeleteVolumeRequest, DiskInfoRequest, ListDirRequest, ListVolumesRequest,
-    MakeVolumeRequest, MakeVolumesRequest, PreparePartTransactionRequest, ReadAllRequest, ReadMetadataRequest,
-    ReadMultipleRequest, ReadMultipleResponse, ReadPartsRequest, ReadVersionRequest, ReadXlRequest, RenameDataRequest,
-    RenameFileRequest, SettlePartTransactionRequest, SnapshotLeaseReleaseRequest, SnapshotLeaseRenewRequest,
+    DeleteVersionRequest, DeleteVersionsRequest, DeleteVersionsResponse, DeleteVolumeRequest, DiskInfoRequest, ListDirRequest,
+    ListVolumesRequest, MakeVolumeRequest, MakeVolumesRequest, PreparePartTransactionRequest, ReadAllRequest,
+    ReadMetadataRequest, ReadMultipleRequest, ReadMultipleResponse, ReadPartsRequest, ReadVersionRequest, ReadXlRequest,
+    RenameDataRequest, RenameFileRequest, SettlePartTransactionRequest, SnapshotLeaseReleaseRequest, SnapshotLeaseRenewRequest,
     SnapshotLeaseRequest, SnapshotLeaseResponse, StatVolumeRequest, UpdateMetadataRequest, VerifyFileRequest, WriteAllRequest,
     WriteMetadataRequest, node_service_client::NodeServiceClient,
 };
@@ -112,6 +112,28 @@ const EVENT_REMOTE_DISK_RPC: &str = "remote_disk_rpc";
 const SNAPSHOT_LEASE_PROTOCOL_VERSION: u32 = 1;
 pub const REMOTE_SNAPSHOT_LEASE_TTL: Duration = Duration::from_secs(60);
 
+fn decode_delete_versions_errors(response: DeleteVersionsResponse, expected_len: usize) -> Vec<Option<Error>> {
+    if !response.item_errors.is_empty() {
+        if response.item_errors.len() != expected_len {
+            return vec![Some(Error::other("malformed delete_versions item errors")); expected_len];
+        }
+        return response
+            .item_errors
+            .into_iter()
+            .map(|error| (error.code != 0).then(|| error.into()))
+            .collect();
+    }
+
+    if response.errors.len() != expected_len {
+        return vec![Some(Error::other("malformed delete_versions errors")); expected_len];
+    }
+    response
+        .errors
+        .into_iter()
+        .map(|error| (!error.is_empty()).then(|| Error::other(error)))
+        .collect()
+}
+
 fn snapshot_lease_token_from_response(response: SnapshotLeaseResponse) -> Result<SnapshotLeaseToken> {
     if !response.success {
         return Err(response.error.unwrap_or_default().into());
@@ -2406,8 +2428,6 @@ impl DiskAPI for RemoteDisk {
             return errors;
         }
 
-        // TODO(backlog): replace string errors with typed `StorageError` variants
-
         let result = self
             .execute_with_timeout(
                 || async {
@@ -2439,17 +2459,7 @@ impl DiskAPI for RemoteDisk {
             }
             return errors;
         }
-        response
-            .errors
-            .iter()
-            .map(|error| {
-                if error.is_empty() {
-                    None
-                } else {
-                    Some(Error::other(error.to_string()))
-                }
-            })
-            .collect()
+        decode_delete_versions_errors(response, versions.len())
     }
 
     #[tracing::instrument(level = "trace", skip_all)]
@@ -3760,6 +3770,63 @@ mod tests {
 
     static INIT: Once = Once::new();
 
+    #[test]
+    fn delete_versions_response_preserves_typed_item_errors() {
+        let errors = decode_delete_versions_errors(
+            DeleteVersionsResponse {
+                success: true,
+                errors: vec!["file not found".to_string(), String::new()],
+                error: None,
+                item_errors: vec![
+                    rustfs_protos::proto_gen::node_service::Error {
+                        code: DiskError::FileNotFound.to_u32(),
+                        error_info: "file not found".to_string(),
+                    },
+                    rustfs_protos::proto_gen::node_service::Error::default(),
+                ],
+            },
+            2,
+        );
+
+        assert!(matches!(errors.as_slice(), [Some(DiskError::FileNotFound), None]));
+    }
+
+    #[test]
+    fn delete_versions_response_accepts_legacy_string_errors() {
+        let errors = decode_delete_versions_errors(
+            DeleteVersionsResponse {
+                success: true,
+                errors: vec!["legacy error".to_string(), String::new()],
+                error: None,
+                item_errors: Vec::new(),
+            },
+            2,
+        );
+
+        assert_eq!(errors.len(), 2);
+        assert_eq!(errors[0].as_ref().map(ToString::to_string).as_deref(), Some("io error legacy error"));
+        assert!(errors[1].is_none());
+    }
+
+    #[test]
+    fn delete_versions_response_rejects_misaligned_item_errors() {
+        let errors = decode_delete_versions_errors(
+            DeleteVersionsResponse {
+                success: true,
+                errors: vec!["file not found".to_string()],
+                error: None,
+                item_errors: vec![rustfs_protos::proto_gen::node_service::Error {
+                    code: DiskError::FileNotFound.to_u32(),
+                    error_info: "file not found".to_string(),
+                }],
+            },
+            2,
+        );
+
+        assert_eq!(errors.len(), 2);
+        assert!(errors.iter().all(Option::is_some));
+    }
+
     #[test]
     fn disk_mutation_digest_marks_rolling_compatibility() {
         let mut request = Request::new(());

crates/ecstore/src/core/pools.rs

@@ -36,7 +36,8 @@ use crate::disk::error::DiskError;
 use crate::disk::{BUCKET_META_PREFIX, RUSTFS_META_BUCKET};
 use crate::error::{Error, Result};
 use crate::error::{
-    StorageError, is_err_bucket_exists, is_err_bucket_not_found, is_err_object_not_found, is_err_version_not_found,
+    StorageError, is_err_bucket_exists, is_err_bucket_not_found, is_err_object_not_found, is_err_operation_canceled,
+    is_err_version_not_found,
 };
 use crate::layout::endpoints::EndpointServerPools;
 use crate::object_api::{GetObjectReader, ObjectOptions};
@@ -89,6 +90,7 @@ const DECOMMISSION_STAGE_SOURCE_CLEANUP: &str = "source_cleanup";
 const DECOMMISSION_STAGE_ENTRY_FINISHED: &str = "entry_finished";
 const DECOMMISSION_PROGRESS_SAVE_INTERVAL: Duration = Duration::seconds(30);
 const DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD: usize = 1000;
+const DECOMMISSION_PROGRESS_SAVE_RETRY_BACKOFF: Duration = Duration::seconds(1);
 const DECOMMISSION_BUCKET_CONCURRENCY_ENV: &str = "RUSTFS_DECOMMISSION_BUCKET_CONCURRENCY";
 const DECOMMISSION_BUCKET_CONCURRENCY_DEFAULT_CAP: usize = 4;
 const DECOMMISSION_TARGET_CAPACITY_OVERHEAD_PERCENT: usize = 30;
@@ -638,22 +640,6 @@ fn track_decommission_current_object(meta: &mut PoolMeta, idx: usize, bucket: &s
     track_decommission_current_object_stage(meta, idx, bucket, object, "")
 }
 
-fn touch_decommission_progress(meta: &mut PoolMeta, idx: usize) -> Result<()> {
-    let pool_count = meta.pools.len();
-    ensure_valid_decommission_pool_index(pool_count, idx)?;
-
-    let Some(pool) = meta.pools.get_mut(idx) else {
-        return Err(invalid_decommission_pool_index_error(pool_count, idx));
-    };
-    let Some(info) = pool.decommission.as_mut() else {
-        return Err(decommission_metadata_not_initialized_error("touch decommission progress"));
-    };
-
-    pool.last_update = OffsetDateTime::now_utc();
-    info.mark_progress_saved();
-    Ok(())
-}
-
 fn resolve_decommission_update_after_result(result: Result<bool>) -> Result<bool> {
     result.map_err(|err| Error::other(format!("decommission metadata update failed: {err}")))
 }
@@ -773,7 +759,76 @@ async fn load_decommission_entry_exact_versions(
 }
 
 fn resolve_decommission_check_after_list_result(list_result: Result<()>, entry_error: Option<Error>) -> Result<()> {
-    if let Some(err) = entry_error { Err(err) } else { list_result }
+    match list_result {
+        Ok(()) => entry_error.map_or(Ok(()), Err),
+        Err(list_err) => resolve_decommission_listing_error(Some(list_err), entry_error).map_or(Ok(()), Err),
+    }
+}
+
+fn resolve_decommission_listing_error(listing_error: Option<Error>, entry_error: Option<Error>) -> Option<Error> {
+    match (listing_error, entry_error) {
+        (Some(listing_error), Some(entry_error)) if is_err_operation_canceled(&listing_error) => Some(entry_error),
+        (Some(listing_error), Some(entry_error)) if is_err_operation_canceled(&entry_error) => Some(listing_error),
+        (Some(listing_error), _) => Some(listing_error),
+        (None, entry_error) => entry_error,
+    }
+}
+
+fn decommission_unresolved_listing_error(
+    bucket: &str,
+    prefix: &str,
+    candidate: Option<&str>,
+    candidate_count: usize,
+    disk_error_count: usize,
+    pool_index: usize,
+    set_index: usize,
+) -> Error {
+    let location = candidate.unwrap_or(prefix);
+    Error::other(format!(
+        "decommission listing could not resolve metadata for {bucket}/{location} on pool {pool_index} set {set_index} ({candidate_count} candidate(s), {disk_error_count} disk error(s))"
+    ))
+}
+
+fn resolve_decommission_partial_listing_entry(
+    entries: MetaCacheEntries,
+    resolver: MetadataResolutionParams,
+    bucket: &str,
+    prefix: &str,
+    disk_error_count: usize,
+    pool_index: usize,
+    set_index: usize,
+) -> Result<MetaCacheEntry> {
+    let candidate_count = entries.as_ref().iter().flatten().count();
+    if let Some(entry) = entries.resolve(resolver) {
+        return Ok(entry);
+    }
+
+    let candidate = entries.as_ref().iter().flatten().map(|entry| entry.name.as_str()).next();
+    Err(decommission_unresolved_listing_error(
+        bucket,
+        prefix,
+        candidate,
+        candidate_count,
+        disk_error_count,
+        pool_index,
+        set_index,
+    ))
+}
+
+async fn record_decommission_entry_error(
+    entry_error: &Arc<tokio::sync::Mutex<Option<Error>>>,
+    rx: &CancellationToken,
+    err: Error,
+) {
+    if rx.is_cancelled() {
+        return;
+    }
+
+    let mut first_err = entry_error.lock().await;
+    if first_err.is_none() && !rx.is_cancelled() {
+        *first_err = Some(err);
+        rx.cancel();
+    }
 }
 
 fn resolve_decommission_pool_meta_reload_result(result: Result<()>, stage: &str) -> Result<()> {
@@ -1483,6 +1538,7 @@ impl TryFrom<PersistedPoolDecommissionInfo> for PoolDecommissionInfo {
             terminal_reload_attempt_at: value.terminal_reload_attempt_at,
             terminal_reload_failures: value.terminal_reload_failures,
             progress_save_item_baseline: value.items_decommissioned.saturating_add(value.items_decommission_failed),
+            progress_save_retry_after: None,
         })
     }
 }
@@ -1514,6 +1570,7 @@ impl TryFrom<LegacyPoolDecommissionInfo> for PoolDecommissionInfo {
             terminal_reload_attempt_at: None,
             terminal_reload_failures: Vec::new(),
             progress_save_item_baseline: value.items_decommissioned.saturating_add(value.items_decommission_failed),
+            progress_save_retry_after: None,
         })
     }
 }
@@ -1627,6 +1684,82 @@ impl PoolMeta {
         }
     }
 
+    fn decommission_progress_checkpoint(
+        &self,
+        idx: usize,
+        duration: Duration,
+        now: OffsetDateTime,
+    ) -> Result<Option<DecommissionProgressCheckpoint>> {
+        let pool_count = self.pools.len();
+        ensure_valid_decommission_pool_index(pool_count, idx)?;
+
+        let Some(pool) = self.pools.get(idx) else {
+            return Err(invalid_decommission_pool_index_error(pool_count, idx));
+        };
+        let Some(info) = pool.decommission.as_ref() else {
+            return Err(decommission_metadata_not_initialized_error("update decommission metadata timestamp"));
+        };
+
+        if info.progress_save_retry_after.is_some_and(|retry_after| now < retry_after) {
+            return Ok(None);
+        }
+
+        let time_threshold_reached = now.unix_timestamp() - pool.last_update.unix_timestamp() >= duration.whole_seconds();
+        let item_threshold_reached = info.items_since_last_progress_save() >= DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD;
+        if !time_threshold_reached && !item_threshold_reached {
+            return Ok(None);
+        }
+
+        Ok(Some(DecommissionProgressCheckpoint {
+            start_time: info.start_time,
+            queued: info.queued,
+            counted_items: info.counted_items(),
+            checkpoint_at: now,
+        }))
+    }
+
+    fn commit_decommission_progress_checkpoint(&mut self, idx: usize, checkpoint: DecommissionProgressCheckpoint) -> bool {
+        let Some(pool) = self.pools.get_mut(idx) else {
+            return false;
+        };
+        let Some(info) = pool.decommission.as_mut() else {
+            return false;
+        };
+
+        if info.start_time != checkpoint.start_time
+            || info.queued != checkpoint.queued
+            || !is_decommission_active(info.complete, info.failed, info.canceled)
+        {
+            return false;
+        }
+
+        info.progress_save_item_baseline = info.progress_save_item_baseline.max(checkpoint.counted_items);
+        info.progress_save_retry_after = None;
+        pool.last_update = pool.last_update.max(checkpoint.checkpoint_at);
+        true
+    }
+
+    fn defer_decommission_progress_checkpoint(
+        &mut self,
+        idx: usize,
+        checkpoint: DecommissionProgressCheckpoint,
+        retry_after: OffsetDateTime,
+    ) {
+        let Some(pool) = self.pools.get_mut(idx) else {
+            return;
+        };
+        let Some(info) = pool.decommission.as_mut() else {
+            return;
+        };
+
+        if info.start_time == checkpoint.start_time
+            && info.queued == checkpoint.queued
+            && is_decommission_active(info.complete, info.failed, info.canceled)
+        {
+            info.progress_save_retry_after = Some(retry_after);
+        }
+    }
+
     fn load_from_config_data(&mut self, data: Vec<u8>) -> Result<()> {
         if data.is_empty() {
             return Ok(());
@@ -1987,30 +2120,9 @@ impl PoolMeta {
     }
 
     pub fn update_after(&mut self, idx: usize, duration: Duration) -> Result<bool> {
-        let pool_count = self.pools.len();
-        ensure_valid_decommission_pool_index(pool_count, idx)?;
-
-        let (last_update, item_threshold_reached) = match self.pools.get(idx) {
-            Some(pool) if let Some(info) = pool.decommission.as_ref() => (
-                pool.last_update,
-                info.items_since_last_progress_save() >= DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD,
-            ),
-            Some(_) => {
-                return Err(decommission_metadata_not_initialized_error("update decommission metadata timestamp"));
-            }
-            None => return Err(invalid_decommission_pool_index_error(pool_count, idx)),
-        };
-        let now = OffsetDateTime::now_utc();
-
-        if now.unix_timestamp() - last_update.unix_timestamp() >= duration.whole_seconds() || item_threshold_reached {
-            let Some(pool) = self.pools.get_mut(idx) else {
-                return Err(invalid_decommission_pool_index_error(pool_count, idx));
-            };
-            pool.last_update = now;
-            return Ok(true);
-        }
-
-        Ok(false)
+        Ok(self
+            .decommission_progress_checkpoint(idx, duration, OffsetDateTime::now_utc())?
+            .is_some())
     }
 
     pub fn validate(&self, pools: Vec<Arc<Sets>>) -> Result<bool> {
@@ -2151,6 +2263,16 @@ pub struct PoolDecommissionInfo {
     pub terminal_reload_failures: Vec<String>,
     #[serde(skip)]
     pub progress_save_item_baseline: usize,
+    #[serde(skip)]
+    pub progress_save_retry_after: Option<OffsetDateTime>,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+struct DecommissionProgressCheckpoint {
+    start_time: Option<OffsetDateTime>,
+    queued: bool,
+    counted_items: usize,
+    checkpoint_at: OffsetDateTime,
 }
 
 impl PoolDecommissionInfo {
@@ -2185,6 +2307,7 @@ impl PoolDecommissionInfo {
 
     fn mark_progress_saved(&mut self) {
         self.progress_save_item_baseline = self.counted_items();
+        self.progress_save_retry_after = None;
     }
 
     pub fn bucket_push(&mut self, bucket: &DecomBucketInfo) {
@@ -2489,6 +2612,40 @@ impl ECStore {
         snapshot.save(self.pools.clone()).await
     }
 
+    async fn save_decommission_progress_checkpoint(&self, idx: usize) -> Result<bool> {
+        // Lock order: save gate, then the short pool metadata read/write sections. Peer
+        // reloads are intentionally performed by the caller after both locks are released.
+        let _save_guard = self.pool_meta_save_gate.lock().await;
+        let (snapshot, checkpoint) = {
+            let pool_meta = self.pool_meta.read().await;
+            let Some(checkpoint) = pool_meta.decommission_progress_checkpoint(
+                idx,
+                DECOMMISSION_PROGRESS_SAVE_INTERVAL,
+                OffsetDateTime::now_utc(),
+            )?
+            else {
+                return Ok(false);
+            };
+
+            let mut snapshot = pool_meta.clone();
+            let Some(pool) = snapshot.pools.get_mut(idx) else {
+                return Err(invalid_decommission_pool_index_error(snapshot.pools.len(), idx));
+            };
+            pool.last_update = checkpoint.checkpoint_at;
+            (snapshot, checkpoint)
+        };
+
+        if let Err(err) = snapshot.save(self.pools.clone()).await {
+            let retry_after = OffsetDateTime::now_utc() + DECOMMISSION_PROGRESS_SAVE_RETRY_BACKOFF;
+            let mut pool_meta = self.pool_meta.write().await;
+            pool_meta.defer_decommission_progress_checkpoint(idx, checkpoint, retry_after);
+            return Err(err);
+        }
+
+        let mut pool_meta = self.pool_meta.write().await;
+        Ok(pool_meta.commit_decommission_progress_checkpoint(idx, checkpoint))
+    }
+
     async fn save_current_pool_meta_for_decommission_start(
         &self,
         indices: &[usize],
@@ -2871,7 +3028,7 @@ impl ECStore {
         Ok(())
     }
 
-    async fn save_decommission_entry_progress_stage(
+    async fn track_decommission_entry_progress_stage(
         &self,
         idx: usize,
         bucket: &str,
@@ -2882,22 +3039,6 @@ impl ECStore {
             let mut pool_meta = self.pool_meta.write().await;
             track_decommission_current_object_stage(&mut pool_meta, idx, bucket, object, stage)
                 .map_err(|err| with_decommission_entry_context(stage, bucket, object, err))?;
-            touch_decommission_progress(&mut pool_meta, idx)
-                .map_err(|err| with_decommission_entry_context(stage, bucket, object, err))?;
-        }
-
-        if let Some(err) = resolve_decommission_progress_save_result(self.save_current_pool_meta().await) {
-            warn!(
-                event = EVENT_DECOMMISSION_ENTRY,
-                component = LOG_COMPONENT_ECSTORE,
-                subsystem = LOG_SUBSYSTEM_POOLS,
-                pool_index = idx,
-                bucket = %bucket,
-                object = %object,
-                stage,
-                error = ?err,
-                "Decommission progress stage save failed"
-            );
         }
 
         Ok(())
@@ -3165,7 +3306,7 @@ impl ECStore {
                 let bucket_name = bucket.clone();
                 let object_name = rd.object_info.name.clone();
 
-                self.save_decommission_entry_progress_stage(
+                self.track_decommission_entry_progress_stage(
                     idx,
                     bucket_name.as_str(),
                     object_name.as_str(),
@@ -3259,7 +3400,7 @@ impl ECStore {
             }
             decommission_cancel_signal_result(rx.is_cancelled())?;
 
-            self.save_decommission_entry_progress_stage(
+            self.track_decommission_entry_progress_stage(
                 idx,
                 bucket.as_str(),
                 entry.name.as_str(),
@@ -3267,7 +3408,7 @@ impl ECStore {
             )
             .await?;
 
-            self.save_decommission_entry_progress_stage(
+            self.track_decommission_entry_progress_stage(
                 idx,
                 bucket.as_str(),
                 entry.name.as_str(),
@@ -3334,34 +3475,42 @@ impl ECStore {
             }
         };
 
-        self.save_decommission_entry_progress_stage(idx, bucket.as_str(), entry.name.as_str(), DECOMMISSION_STAGE_ENTRY_FINISHED)
-            .await?;
+        self.track_decommission_entry_progress_stage(
+            idx,
+            bucket.as_str(),
+            entry.name.as_str(),
+            DECOMMISSION_STAGE_ENTRY_FINISHED,
+        )
+        .await?;
 
         if should_save_progress {
-            let save_result = self.save_current_pool_meta().await;
-            if let Some(err) = resolve_decommission_progress_save_result(save_result) {
-                warn!(
-                    event = EVENT_DECOMMISSION_ENTRY,
-                    component = LOG_COMPONENT_ECSTORE,
-                    subsystem = LOG_SUBSYSTEM_POOLS,
-                    pool_index = idx,
-                    bucket = %bucket,
-                    object = %entry.name,
-                    state = "progress_save_failed",
-                    error = %err,
-                    "Decommission progress save failed; continuing and will retry at the next checkpoint"
-                );
-            } else {
-                let mut pool_meta = self.pool_meta.write().await;
-                pool_meta.mark_decommission_progress_saved();
-                if let Some(notification_sys) = runtime_sources::notification_sys()
-                    && let Err(err) = resolve_decommission_entry_reload_result(
-                        notification_sys.reload_pool_meta().await,
-                        bucket.as_str(),
-                        entry.name.as_str(),
-                    )
-                {
-                    warn!("{err}");
+            match self.save_decommission_progress_checkpoint(idx).await {
+                Ok(true) => {
+                    if let Some(notification_sys) = runtime_sources::notification_sys()
+                        && let Err(err) = resolve_decommission_entry_reload_result(
+                            notification_sys.reload_pool_meta().await,
+                            bucket.as_str(),
+                            entry.name.as_str(),
+                        )
+                    {
+                        warn!("{err}");
+                    }
+                }
+                Ok(false) => {}
+                Err(err) => {
+                    if let Some(err) = resolve_decommission_progress_save_result(Err(err)) {
+                        warn!(
+                            event = EVENT_DECOMMISSION_ENTRY,
+                            component = LOG_COMPONENT_ECSTORE,
+                            subsystem = LOG_SUBSYSTEM_POOLS,
+                            pool_index = idx,
+                            bucket = %bucket,
+                            object = %entry.name,
+                            state = "progress_save_failed",
+                            error = %err,
+                            "Decommission progress save failed; continuing and will retry at the next checkpoint"
+                        );
+                    }
                 }
             }
         }
@@ -3538,6 +3687,7 @@ impl ECStore {
             let rx_clone = rx.clone();
             let bi = bi.clone();
             let set_id = set_idx;
+            let listing_entry_error = entry_error.clone();
             let worker = tokio::spawn(async move {
                 let _listing_permit = listing_permit;
                 run_decommission_listing_with_retry(
@@ -3551,7 +3701,11 @@ impl ECStore {
                         let set = set.clone();
                         let rx = rx_clone.clone();
                         let bucket = bi.clone();
-                        async move { set.list_objects_to_decommission(rx, bucket, callback).await }
+                        let entry_error = listing_entry_error.clone();
+                        async move {
+                            set.list_objects_to_decommission(rx, bucket, callback, entry_error.clone(), idx, set_id)
+                                .await
+                        }
                     },
                 )
                 .await
@@ -3581,11 +3735,7 @@ impl ECStore {
 
         wait_decommission_worker_drain(&workers, worker_limit).await?;
 
-        if let Some(err) = listing_worker_error {
-            return Err(err);
-        }
-
-        if let Some(err) = entry_error.lock().await.clone() {
+        if let Some(err) = resolve_decommission_listing_error(listing_worker_error, entry_error.lock().await.clone()) {
             return Err(err);
         }
 
@@ -4191,7 +4341,7 @@ impl ECStore {
         let buckets = self.get_buckets_to_decommission().await?;
         let pool = self.pools[idx].clone();
 
-        for set in &pool.disk_set {
+        for (set_index, set) in pool.disk_set.iter().enumerate() {
             for bucket_info in &buckets {
                 let mut lifecycle_config = None;
                 let mut object_lock_config = None;
@@ -4286,7 +4436,7 @@ impl ECStore {
                 });
 
                 let list_result = set
-                    .list_objects_to_decommission(callback_rx, bucket_info.clone(), callback)
+                    .list_objects_to_decommission(callback_rx, bucket_info.clone(), callback, entry_error.clone(), idx, set_index)
                     .await;
                 let entry_error = entry_error.lock().await.clone();
                 resolve_decommission_check_after_list_result(list_result, entry_error)?;
@@ -5021,12 +5171,15 @@ mod tests {
 pub type ListCallback = Arc<dyn Fn(MetaCacheEntry) -> BoxFuture<'static, ()> + Send + Sync + 'static>;
 
 impl SetDisks {
-    #[tracing::instrument(skip(self, rx, cb_func))]
+    #[tracing::instrument(skip(self, rx, cb_func, entry_error))]
     async fn list_objects_to_decommission(
         self: &Arc<Self>,
         rx: CancellationToken,
         bucket_info: DecomBucketInfo,
         cb_func: ListCallback,
+        entry_error: Arc<tokio::sync::Mutex<Option<Error>>>,
+        pool_index: usize,
+        set_index: usize,
     ) -> Result<()> {
         let (disks, _) = self.get_online_disks_with_healing(false).await;
         ensure_decommission_listing_disks_available(!disks.is_empty(), &bucket_info.name)?;
@@ -5041,6 +5194,12 @@ impl SetDisks {
         };
 
         let cb1 = cb_func.clone();
+        let unresolved_error = entry_error.clone();
+        let unresolved_rx = rx.clone();
+        let unresolved_bucket = bucket_info.name.clone();
+        let unresolved_prefix = bucket_info.prefix.clone();
+        let unresolved_pool_index = pool_index;
+        let unresolved_set_index = set_index;
 
         list_path_raw(
             rx,
@@ -5053,20 +5212,51 @@ impl SetDisks {
                 skip_walkdir_total_timeout: true,
                 walkdir_stall_timeout: Some(DECOMMISSION_BACKGROUND_WALKDIR_STALL_TIMEOUT),
                 agreed: Some(Box::new(move |entry: MetaCacheEntry| Box::pin(cb1(entry)))),
-                partial: Some(Box::new(move |entries: MetaCacheEntries, _: &[Option<DiskError>]| {
+                partial: Some(Box::new(move |entries: MetaCacheEntries, errs: &[Option<DiskError>]| {
                     let resolver = resolver.clone();
                     let cb_func = cb_func.clone();
-                    match entries.resolve(resolver) {
-                        Some(entry) => {
+                    let bucket = unresolved_bucket.clone();
+                    let prefix = unresolved_prefix.clone();
+                    let unresolved_error = unresolved_error.clone();
+                    let unresolved_rx = unresolved_rx.clone();
+                    let pool_index = unresolved_pool_index;
+                    let set_index = unresolved_set_index;
+                    let disk_error_count = errs.iter().flatten().count();
+                    if unresolved_rx.is_cancelled() {
+                        return Box::pin(async {});
+                    }
+
+                    match resolve_decommission_partial_listing_entry(
+                        entries,
+                        resolver,
+                        &bucket,
+                        &prefix,
+                        disk_error_count,
+                        pool_index,
+                        set_index,
+                    ) {
+                        Ok(entry) => {
                             warn!("decommission_pool: list_objects_to_decommission get {}", &entry.name);
                             Box::pin(async move {
                                 cb_func(entry).await;
                             })
                         }
-                        None => {
-                            warn!("decommission_pool: list_objects_to_decommission get none");
-                            Box::pin(async {})
-                        }
+                        Err(err) => Box::pin(async move {
+                            if unresolved_rx.is_cancelled() {
+                                return;
+                            }
+                            warn!(
+                                event = EVENT_DECOMMISSION_BUCKET,
+                                component = LOG_COMPONENT_ECSTORE,
+                                subsystem = LOG_SUBSYSTEM_POOLS,
+                                bucket = %bucket,
+                                prefix = %prefix,
+                                state = "unresolved_entry",
+                                error = %err,
+                                "Decommission listing failed closed on unresolved metadata"
+                            );
+                            record_decommission_entry_error(&unresolved_error, &unresolved_rx, err).await;
+                        }),
                     }
                 })),
                 ..Default::default()
@@ -5074,6 +5264,10 @@ impl SetDisks {
         )
         .await?;
 
+        if let Some(err) = entry_error.lock().await.clone() {
+            return Err(err);
+        }
+
         Ok(())
     }
 }
@@ -5264,11 +5458,11 @@ pub(crate) fn fallback_free_capacity_dedup(disks: &[rustfs_madmin::Disk]) -> usi
 #[cfg(test)]
 mod pools_tests {
     use super::{
-        DECOMMISSION_PROGRESS_SAVE_INTERVAL, DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD, DecomBucketInfo,
-        DecommissionStartPoolState, DecommissionTerminalState, ListCallback, PoolDecommissionInfo, PoolMeta, PoolSpaceInfo,
-        PoolStatus, apply_decommission_status_space_info, bind_decommission_cancelers, bind_missing_decommission_cancelers,
-        cancel_decommission_canceler, classify_decommission_terminal_state, count_decommission_item,
-        decommission_cancel_signal_result, decommission_item_size, decommission_meta_bucket_options,
+        DECOMMISSION_PROGRESS_SAVE_INTERVAL, DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD, DECOMMISSION_PROGRESS_SAVE_RETRY_BACKOFF,
+        DecomBucketInfo, DecommissionStartPoolState, DecommissionTerminalState, ListCallback, PoolDecommissionInfo, PoolMeta,
+        PoolSpaceInfo, PoolStatus, apply_decommission_status_space_info, bind_decommission_cancelers,
+        bind_missing_decommission_cancelers, cancel_decommission_canceler, classify_decommission_terminal_state,
+        count_decommission_item, decommission_cancel_signal_result, decommission_item_size, decommission_meta_bucket_options,
         decommission_start_pool_state, dedup_indices, default_decommission_bucket_concurrency,
         ensure_decommission_cancel_allowed, ensure_decommission_clear_allowed, ensure_decommission_listing_disks_available,
         ensure_decommission_not_rebalancing, ensure_decommission_start_allowed, ensure_decommission_start_keeps_active_pool,
@@ -5279,11 +5473,12 @@ mod pools_tests {
         has_active_decommission_canceler, is_decommission_active, is_decommission_cancel_requested,
         load_decommission_entry_versions, local_decommission_queue_prefix, mark_decommission_bucket_done,
         merge_pool_status_refresh, missing_decommission_worker_prefix, observe_decommission_terminal_reload_result,
-        pool_meta_has_active_decommission, require_decommission_store, resolve_decommission_bucket_done_save_result,
-        resolve_decommission_bucket_state, resolve_decommission_check_after_list_result,
-        resolve_decommission_entry_cleanup_delete_result, resolve_decommission_entry_exact_versions,
-        resolve_decommission_entry_reload_result, resolve_decommission_listing_worker_result,
-        resolve_decommission_optional_bucket_config_result, resolve_decommission_pool_meta_reload_result,
+        pool_meta_has_active_decommission, record_decommission_entry_error, require_decommission_store,
+        resolve_decommission_bucket_done_save_result, resolve_decommission_bucket_state,
+        resolve_decommission_check_after_list_result, resolve_decommission_entry_cleanup_delete_result,
+        resolve_decommission_entry_exact_versions, resolve_decommission_entry_reload_result, resolve_decommission_listing_error,
+        resolve_decommission_listing_worker_result, resolve_decommission_optional_bucket_config_result,
+        resolve_decommission_partial_listing_entry, resolve_decommission_pool_meta_reload_result,
         resolve_decommission_preflight_heal_result, resolve_decommission_progress_save_result,
         resolve_decommission_spawn_failure_result, resolve_decommission_terminal_mark_after_error_result,
         resolve_decommission_terminal_mark_result, resolve_decommission_update_after_result,
@@ -5293,16 +5488,17 @@ mod pools_tests {
         should_preserve_decommission_canceled_state, should_reject_decommission_cancel_as_terminal,
         should_retry_decommission_cancel_reload, should_retry_decommission_listing, should_skip_canceled_decommission_routine,
         split_decommission_buckets, take_and_cancel_decommission_canceler, take_decommission_canceler,
-        touch_decommission_progress, track_decommission_current_object, track_decommission_current_object_stage,
-        validate_start_decommission_request, wait_decommission_listing_retry, wait_decommission_worker_drain,
-        with_decommission_entry_context,
+        track_decommission_current_object, track_decommission_current_object_stage, validate_start_decommission_request,
+        wait_decommission_listing_retry, wait_decommission_worker_drain, with_decommission_entry_context,
     };
     use crate::data_movement;
     use crate::disk::endpoint::Endpoint;
     use crate::error::{Error, StorageError};
     use crate::layout::endpoints::{EndpointServerPools, Endpoints, PoolEndpoints};
     use crate::services::rebalance::{RebalStatus, RebalanceInfo, RebalanceMeta, RebalanceStats};
-    use rustfs_filemeta::{FileInfo, FileInfoVersions, MetaCacheEntry, ObjectPartInfo};
+    use rustfs_filemeta::{
+        FileInfo, FileInfoVersions, MetaCacheEntries, MetaCacheEntry, MetadataResolutionParams, ObjectPartInfo,
+    };
     use rustfs_rio::Index;
     use std::sync::{
         Arc,
@@ -6321,6 +6517,65 @@ mod pools_tests {
         assert!(matches!(err, Error::SlowDown));
     }
 
+    #[test]
+    fn test_resolve_decommission_partial_listing_entry_rejects_unresolved_metadata() {
+        let err = resolve_decommission_partial_listing_entry(
+            MetaCacheEntries(vec![None]),
+            MetadataResolutionParams {
+                dir_quorum: 2,
+                obj_quorum: 2,
+                bucket: "bucket-a".to_string(),
+                ..Default::default()
+            },
+            "bucket-a",
+            "prefix/",
+            1,
+            2,
+            3,
+        )
+        .expect_err("unresolved partial listing must fail closed");
+
+        let message = err.to_string();
+        assert!(message.contains("decommission listing could not resolve metadata"));
+        assert!(message.contains("bucket-a/prefix/"));
+        assert!(message.contains("pool 2 set 3"));
+        assert!(message.contains("1 disk error(s)"));
+    }
+
+    #[tokio::test]
+    async fn test_record_decommission_entry_error_cancels_listing_and_preserves_first_error() {
+        let entry_error = Arc::new(tokio::sync::Mutex::new(None));
+        let rx = CancellationToken::new();
+
+        record_decommission_entry_error(&entry_error, &rx, Error::SlowDown).await;
+        record_decommission_entry_error(&entry_error, &rx, Error::OperationCanceled).await;
+
+        assert!(rx.is_cancelled());
+        assert!(matches!(*entry_error.lock().await, Some(Error::SlowDown)));
+    }
+
+    #[tokio::test]
+    async fn test_record_decommission_entry_error_ignores_already_canceled_listing() {
+        let entry_error = Arc::new(tokio::sync::Mutex::new(None));
+        let rx = CancellationToken::new();
+        rx.cancel();
+
+        record_decommission_entry_error(&entry_error, &rx, Error::SlowDown).await;
+
+        assert!(entry_error.lock().await.is_none());
+    }
+
+    #[test]
+    fn test_resolve_decommission_listing_error_preserves_real_listing_failure() {
+        let err = resolve_decommission_listing_error(Some(Error::SlowDown), Some(Error::OperationCanceled))
+            .expect("listing failure should be returned");
+        assert!(matches!(err, Error::SlowDown));
+
+        let err = resolve_decommission_listing_error(Some(Error::OperationCanceled), Some(Error::SlowDown))
+            .expect("entry failure should be returned");
+        assert!(matches!(err, Error::SlowDown));
+    }
+
     #[test]
     fn test_resolve_decommission_check_after_list_result_returns_list_result_without_entry_error() {
         let err = resolve_decommission_check_after_list_result(Err(Error::OperationCanceled), None)
@@ -6538,7 +6793,7 @@ mod pools_tests {
     }
 
     #[test]
-    fn test_touch_decommission_progress_updates_last_update_and_save_baseline() {
+    fn test_track_decommission_stage_does_not_advance_checkpoint_state() {
         let mut meta = PoolMeta {
             pools: vec![PoolStatus {
                 id: 0,
@@ -6553,11 +6808,13 @@ mod pools_tests {
             ..Default::default()
         };
 
-        touch_decommission_progress(&mut meta, 0).expect("valid decommission progress should be touched");
+        track_decommission_current_object_stage(&mut meta, 0, "bucket", "object", "migrate_object")
+            .expect("valid decommission progress should be tracked");
 
-        assert!(meta.pools[0].last_update > OffsetDateTime::UNIX_EPOCH);
+        assert_eq!(meta.pools[0].last_update, OffsetDateTime::UNIX_EPOCH);
         let info = meta.pools[0].decommission.as_ref().expect("decommission info should exist");
-        assert_eq!(info.items_since_last_progress_save(), 0);
+        assert_eq!(info.items_since_last_progress_save(), 5);
+        assert_eq!(info.stage, "migrate_object");
     }
 
     #[test]
@@ -6632,6 +6889,134 @@ mod pools_tests {
         assert_eq!(info.items_since_last_progress_save(), 1);
     }
 
+    #[test]
+    fn test_pool_meta_update_after_does_not_advance_last_update_before_save() {
+        let last_update = OffsetDateTime::UNIX_EPOCH;
+        let mut meta = PoolMeta {
+            pools: vec![PoolStatus {
+                id: 0,
+                cmd_line: "pool-0".to_string(),
+                last_update,
+                decommission: Some(PoolDecommissionInfo {
+                    start_time: Some(last_update),
+                    items_decommissioned: DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD,
+                    ..Default::default()
+                }),
+            }],
+            ..Default::default()
+        };
+
+        assert!(
+            meta.update_after(0, DECOMMISSION_PROGRESS_SAVE_INTERVAL)
+                .expect("item threshold should request a checkpoint")
+        );
+        assert_eq!(meta.pools[0].last_update, last_update);
+    }
+
+    #[test]
+    fn test_decommission_progress_checkpoint_commits_exact_snapshot_watermark() {
+        let start_time = OffsetDateTime::UNIX_EPOCH;
+        let checkpoint_at = start_time + Duration::seconds(30);
+        let mut meta = PoolMeta {
+            pools: vec![PoolStatus {
+                id: 0,
+                cmd_line: "pool-0".to_string(),
+                last_update: start_time,
+                decommission: Some(PoolDecommissionInfo {
+                    start_time: Some(start_time),
+                    items_decommissioned: DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD,
+                    ..Default::default()
+                }),
+            }],
+            ..Default::default()
+        };
+
+        let checkpoint = meta
+            .decommission_progress_checkpoint(0, DECOMMISSION_PROGRESS_SAVE_INTERVAL, checkpoint_at)
+            .expect("valid decommission state should produce a checkpoint")
+            .expect("item threshold should produce a checkpoint");
+        meta.count_item(0, 1, false);
+
+        assert!(meta.commit_decommission_progress_checkpoint(0, checkpoint));
+        let info = meta.pools[0].decommission.as_ref().expect("decommission info should exist");
+        assert_eq!(info.progress_save_item_baseline, checkpoint.counted_items);
+        assert_eq!(info.items_since_last_progress_save(), 1);
+        assert_eq!(meta.pools[0].last_update, checkpoint_at);
+    }
+
+    #[test]
+    fn test_decommission_progress_checkpoint_backoff_does_not_advance_baseline() {
+        let start_time = OffsetDateTime::UNIX_EPOCH;
+        let checkpoint_at = start_time + Duration::seconds(30);
+        let retry_after = checkpoint_at + DECOMMISSION_PROGRESS_SAVE_RETRY_BACKOFF;
+        let mut meta = PoolMeta {
+            pools: vec![PoolStatus {
+                id: 0,
+                cmd_line: "pool-0".to_string(),
+                last_update: start_time,
+                decommission: Some(PoolDecommissionInfo {
+                    start_time: Some(start_time),
+                    items_decommissioned: DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD,
+                    ..Default::default()
+                }),
+            }],
+            ..Default::default()
+        };
+
+        let checkpoint = meta
+            .decommission_progress_checkpoint(0, DECOMMISSION_PROGRESS_SAVE_INTERVAL, checkpoint_at)
+            .expect("valid decommission state should produce a checkpoint")
+            .expect("item threshold should produce a checkpoint");
+        meta.defer_decommission_progress_checkpoint(0, checkpoint, retry_after);
+
+        assert!(
+            meta.decommission_progress_checkpoint(0, DECOMMISSION_PROGRESS_SAVE_INTERVAL, checkpoint_at)
+                .expect("retry backoff check should succeed")
+                .is_none()
+        );
+        assert_eq!(meta.pools[0].last_update, start_time);
+        assert_eq!(
+            meta.pools[0]
+                .decommission
+                .as_ref()
+                .expect("decommission info should exist")
+                .progress_save_item_baseline,
+            0
+        );
+    }
+
+    #[test]
+    fn test_decommission_progress_checkpoint_count_scales_with_threshold() {
+        let start_time = OffsetDateTime::UNIX_EPOCH;
+        let checkpoint_at = start_time;
+        let mut meta = PoolMeta {
+            pools: vec![PoolStatus {
+                id: 0,
+                cmd_line: "pool-0".to_string(),
+                last_update: start_time,
+                decommission: Some(PoolDecommissionInfo {
+                    start_time: Some(start_time),
+                    ..Default::default()
+                }),
+            }],
+            ..Default::default()
+        };
+        let mut checkpoint_count = 0;
+
+        for _ in 0..(DECOMMISSION_PROGRESS_SAVE_ITEM_THRESHOLD * 10) {
+            meta.count_item(0, 1, false);
+            if let Some(checkpoint) = meta
+                .decommission_progress_checkpoint(0, DECOMMISSION_PROGRESS_SAVE_INTERVAL, checkpoint_at)
+                .expect("valid decommission state should produce a checkpoint")
+            {
+                checkpoint_count += 1;
+                assert!(meta.commit_decommission_progress_checkpoint(0, checkpoint));
+            }
+        }
+
+        assert_eq!(checkpoint_count, 10);
+    }
+
     #[test]
     fn test_ensure_decommission_not_rebalancing_rejects_running_rebalance() {
         let err = ensure_decommission_not_rebalancing(true).expect_err("rebalance running should be rejected");

crates/heal/Cargo.toml

@@ -91,7 +91,6 @@ metrics = { workspace = true }
 base64 = { workspace = true }
 bytes = { workspace = true }
 crc-fast = { workspace = true }
-sha2 = { workspace = true }
 
 [dev-dependencies]
 serde_json = { workspace = true, features = ["raw_value"] }

crates/heal/src/heal/channel.rs

@@ -1640,6 +1640,60 @@ mod tests {
         assert_eq!(payload["items"].as_array().expect("items should be an array").len(), 0);
     }
 
+    #[tokio::test]
+    async fn test_process_query_request_reports_displaced_terminal_detail() {
+        let heal_manager = Arc::new(HealManager::new(
+            Arc::new(MockStorage),
+            Some(HealConfig {
+                queue_size: 1,
+                ..HealConfig::default()
+            }),
+        ));
+        let mut displaced = HealRequest::new(
+            HealType::Bucket {
+                bucket: "displaced-channel".to_string(),
+            },
+            HealOptions::default(),
+            HealPriority::Low,
+        );
+        displaced.id = "displaced-channel-task".to_string();
+        let displaced_id = displaced.id.clone();
+        heal_manager
+            .submit_heal_request(displaced)
+            .await
+            .expect("initial channel task should queue");
+        heal_manager
+            .submit_heal_request(HealRequest::new(
+                HealType::Bucket {
+                    bucket: "successor-channel".to_string(),
+                },
+                HealOptions::default(),
+                HealPriority::High,
+            ))
+            .await
+            .expect("successor channel task should displace the initial task");
+
+        let processor = HealChannelProcessor::new(heal_manager);
+        let (tx, rx) = oneshot::channel();
+        processor
+            .process_query_request("displaced-channel".to_string(), displaced_id, None, tx)
+            .await
+            .expect("displaced query should process");
+        let response = rx
+            .await
+            .expect("query response should be returned")
+            .expect("displaced query should remain successful");
+        let payload: serde_json::Value = serde_json::from_slice(response.data.as_deref().expect("status payload should exist"))
+            .expect("status payload should be json");
+        assert_eq!(payload["summary"], "stopped");
+        assert!(
+            response
+                .error
+                .as_deref()
+                .is_some_and(|detail| detail.contains("reason=displaced"))
+        );
+    }
+
     #[tokio::test]
     async fn test_process_query_request_reports_running_for_queued_task() {
         let heal_manager = create_test_heal_manager();

crates/heal/src/heal/erasure_healer.rs

@@ -373,11 +373,6 @@ impl ErasureSetHealer {
         set_disk_id: &str,
         buckets: &[String],
     ) -> Result<(ResumeManager, CheckpointManager)> {
-        if self.replacement_task_id.is_none() && CheckpointManager::is_blocked(&self.disk, task_id).await {
-            return Err(Error::TaskExecutionFailed {
-                message: format!("Resume task {task_id} has a blocked checkpoint"),
-            });
-        }
         // check if resume state exists
         let has_resume_state = if self.replacement_task_id.is_some() {
             ResumeManager::has_replacement_intent(&self.disk, task_id).await

crates/heal/src/heal/manager.rs

@@ -40,6 +40,7 @@ use tracing::{debug, error, info, warn};
 use super::{DiskError, Endpoint, HealDiskExt as _, local_disk_map_read};
 
 const KEEP_HEAL_TASK_STATUS_DURATION: Duration = Duration::from_secs(10 * 60);
+const DISPLACED_HEAL_REASON: &str = "reason=displaced; retry_hint=submit_again";
 const LOG_COMPONENT_HEAL: &str = "heal";
 const LOG_SUBSYSTEM_DISK_SCANNER: &str = "disk_scanner";
 const LOG_SUBSYSTEM_MANAGER: &str = "manager";
@@ -120,26 +121,30 @@ struct MrfRepairNoticeTarget {
     version_id: Option<[u8; 16]>,
 }
 
-#[derive(Debug, Clone, PartialEq, Eq)]
+#[derive(Debug, Clone)]
 struct HealAdmissionDecision {
     result: HealAdmissionResult,
-    displaced_task_id: Option<String>,
+    displaced_request: Option<HealRequest>,
 }
 
 impl HealAdmissionDecision {
     const fn new(result: HealAdmissionResult) -> Self {
         Self {
             result,
-            displaced_task_id: None,
+            displaced_request: None,
         }
     }
 
-    fn accepted_with_displacement(displaced_task_id: String) -> Self {
+    fn accepted_with_displacement(displaced_request: HealRequest) -> Self {
         Self {
             result: HealAdmissionResult::Accepted,
-            displaced_task_id: Some(displaced_task_id),
+            displaced_request: Some(displaced_request),
         }
     }
+
+    fn displaced_task_id(&self) -> Option<&str> {
+        self.displaced_request.as_ref().map(|request| request.id.as_str())
+    }
 }
 
 fn lock_mrf_repair_notice_targets(
@@ -151,6 +156,55 @@ fn lock_mrf_repair_notice_targets(
     }
 }
 
+fn lock_displaced_terminals(
+    registry: &StdMutex<HashMap<String, Arc<CompletedHealStatus>>>,
+) -> StdMutexGuard<'_, HashMap<String, Arc<CompletedHealStatus>>> {
+    match registry.lock() {
+        Ok(guard) => guard,
+        Err(poisoned) => poisoned.into_inner(),
+    }
+}
+
+fn record_displaced_terminal(
+    registry: &StdMutex<HashMap<String, Arc<CompletedHealStatus>>>,
+    request: &HealRequest,
+) -> Arc<CompletedHealStatus> {
+    let terminal = Arc::new(CompletedHealStatus {
+        heal_type: request.heal_type.clone(),
+        status: HealTaskStatus::Failed {
+            error: format!("heal task displaced by a higher-priority request ({DISPLACED_HEAL_REASON})"),
+        },
+        result_items_truncated: false,
+        completed_at: SystemTime::now(),
+        seqed_items: Vec::new(),
+        next_seq: 0,
+        min_seq: 0,
+    });
+    let mut terminals = lock_displaced_terminals(registry);
+    prune_completed_heal_statuses(&mut terminals);
+    terminals.insert(request.id.clone(), Arc::clone(&terminal));
+    terminal
+}
+
+async fn remove_displaced_task_aliases(
+    aliases: &Arc<Mutex<HashMap<String, HealTaskAlias>>>,
+    terminals: &StdMutex<HashMap<String, Arc<CompletedHealStatus>>>,
+    task_id: &str,
+    terminal: &Arc<CompletedHealStatus>,
+) {
+    let mut aliases = aliases.lock().await;
+    let alias_ids = aliases
+        .iter()
+        .filter_map(|(alias_id, alias)| (alias.task_id == task_id).then_some(alias_id.clone()))
+        .collect::<Vec<_>>();
+    let mut displaced_terminals = lock_displaced_terminals(terminals);
+    prune_completed_heal_statuses(&mut displaced_terminals);
+    for alias_id in alias_ids {
+        displaced_terminals.insert(alias_id, Arc::clone(terminal));
+    }
+    aliases.retain(|alias_id, alias| alias_id != task_id && alias.task_id != task_id);
+}
+
 async fn remove_task_aliases_for_task(registry: &Arc<Mutex<HashMap<String, HealTaskAlias>>>, task_id: &str) {
     registry
         .lock()
@@ -618,6 +672,14 @@ pub struct HealManager {
     /// are shared so the lookup helper can hand a completed entry to a
     /// caller without cloning the retained result window.
     completed_heals: Arc<Mutex<HashMap<String, Arc<CompletedHealStatus>>>>,
+    /// Terminals for requests removed by priority displacement. An Accepted
+    /// task ID remains queryable for the same process lifetime and the normal
+    /// ten-minute status TTL; clients should treat `reason=displaced` as a
+    /// terminal result and submit a fresh request. This sidecar is synchronous
+    /// so admission can publish the terminal while the queue transition is
+    /// still under its lock, without awaiting another tokio lock. Queue state
+    /// is process-local, so this guarantee does not extend across restart.
+    displaced_terminals: Arc<StdMutex<HashMap<String, Arc<CompletedHealStatus>>>>,
     /// Client tokens merged into an existing task id.
     task_aliases: Arc<Mutex<HashMap<String, HealTaskAlias>>>,
     /// Heal tasks waiting for a retry backoff to expire.
@@ -659,6 +721,7 @@ struct HealQueueContext<'a> {
     heal_queue: &'a Arc<Mutex<PriorityHealQueue>>,
     active_heals: &'a Arc<Mutex<HashMap<String, Arc<HealTask>>>>,
     completed_heals: &'a Arc<Mutex<HashMap<String, Arc<CompletedHealStatus>>>>,
+    displaced_terminals: &'a Arc<StdMutex<HashMap<String, Arc<CompletedHealStatus>>>>,
     task_aliases: &'a Arc<Mutex<HashMap<String, HealTaskAlias>>>,
     retrying_heals: &'a Arc<Mutex<HashMap<String, RetryingHeal>>>,
     mrf_repair_notice_targets: &'a Arc<StdMutex<HashMap<String, Vec<MrfRepairNoticeTarget>>>>,
@@ -874,7 +937,7 @@ impl HealManager {
                         result = "accepted_by_displacement",
                         "Heal queue request accepted by displacement"
                     });
-                    return HealAdmissionDecision::accepted_with_displacement(displaced.id);
+                    return HealAdmissionDecision::accepted_with_displacement(displaced);
                 }
 
                 demote_to_debug_when!(per_object_request, warn, target: "rustfs::heal::manager", {
@@ -1105,6 +1168,7 @@ impl HealManager {
             active_heals: Arc::new(Mutex::new(HashMap::new())),
             heal_queue: Arc::new(Mutex::new(PriorityHealQueue::new())),
             completed_heals: Arc::new(Mutex::new(HashMap::new())),
+            displaced_terminals: Arc::new(StdMutex::new(HashMap::new())),
             task_aliases: Arc::new(Mutex::new(HashMap::new())),
             retrying_heals: Arc::new(Mutex::new(HashMap::new())),
             mrf_repair_notice_targets: Arc::new(StdMutex::new(HashMap::new())),
@@ -1209,6 +1273,10 @@ impl HealManager {
         active_heals.clear();
         publish_active_heal_count(&active_heals);
         self.completed_heals.lock().await.clear();
+        // Do not let the synchronous guard live across the following async lock.
+        {
+            lock_displaced_terminals(&self.displaced_terminals).clear();
+        }
         self.task_aliases.lock().await.clear();
         self.retrying_heals.lock().await.clear();
         lock_mrf_repair_notice_targets(&self.mrf_repair_notice_targets).clear();
@@ -1459,7 +1527,11 @@ impl HealManager {
             task_id = queued_id.to_owned();
         }
         let should_notify = matches!(admission, HealAdmissionResult::Accepted) && config.event_driven_scheduler_enable;
-        let displaced_task_id = admission_decision.displaced_task_id;
+        let displaced_task_id = admission_decision.displaced_task_id().map(ToOwned::to_owned);
+        let displaced_terminal = admission_decision
+            .displaced_request
+            .as_ref()
+            .map(|request| record_displaced_terminal(&self.displaced_terminals, request));
         if matches!(admission, HealAdmissionResult::Accepted | HealAdmissionResult::Merged)
             && let Some(target) = mrf_notice_target
         {
@@ -1473,8 +1545,12 @@ impl HealManager {
         drop(queue);
         drop(active_heals);
 
-        if let Some(displaced_task_id) = displaced_task_id {
-            self.remove_aliases_for_task(&displaced_task_id).await;
+        if let (Some(displaced_task_id), Some(displaced_terminal)) = (displaced_task_id, displaced_terminal) {
+            // The queue has already removed the displaced request, so the
+            // synchronous terminal sidecar was published before aliases and
+            // MRF ownership are cleaned up.
+            remove_displaced_task_aliases(&self.task_aliases, &self.displaced_terminals, &displaced_task_id, &displaced_terminal)
+                .await;
         }
 
         if should_notify {
@@ -1549,6 +1625,15 @@ impl HealManager {
             }
         }
 
+        if terminal_completed.is_none() {
+            let mut displaced_terminals = lock_displaced_terminals(&self.displaced_terminals);
+            prune_completed_heal_statuses(&mut displaced_terminals);
+            terminal_completed = displaced_terminals
+                .get(canonical_task_id)
+                .filter(|terminal| matches_path(&terminal.heal_type))
+                .cloned();
+        }
+
         match terminal_completed {
             Some(completed) => TaskStateLookup::Completed(completed),
             None => TaskStateLookup::NotFound,
@@ -1669,9 +1754,19 @@ impl HealManager {
 
         let mut completed_heals = self.completed_heals.lock().await;
         prune_completed_heal_statuses(&mut completed_heals);
-        completed_heals
+        if completed_heals
             .values()
             .any(|completed| heal_type_matches_path(&completed.heal_type, heal_path))
+        {
+            return true;
+        }
+        drop(completed_heals);
+
+        let mut displaced_terminals = lock_displaced_terminals(&self.displaced_terminals);
+        prune_completed_heal_statuses(&mut displaced_terminals);
+        displaced_terminals
+            .values()
+            .any(|terminal| heal_type_matches_path(&terminal.heal_type, heal_path))
     }
 
     /// Get task progress

crates/heal/src/heal/manager/auto_scan.rs

@@ -21,6 +21,7 @@ impl HealManager {
         let heal_queue = self.heal_queue.clone();
         let active_heals = self.active_heals.clone();
         let task_aliases = self.task_aliases.clone();
+        let displaced_terminals = self.displaced_terminals.clone();
         let mrf_repair_notice_targets = self.mrf_repair_notice_targets.clone();
         let storage = self.storage.clone();
         let replacement_recovery_anchors = self.replacement_recovery_anchors.clone();
@@ -481,6 +482,10 @@ impl HealManager {
                             let admission = admission_decision.result;
                             let should_notify =
                                 matches!(admission, HealAdmissionResult::Accepted) && config.event_driven_scheduler_enable;
+                            let displaced_terminal = admission_decision
+                                .displaced_request
+                                .as_ref()
+                                .map(|request| record_displaced_terminal(&displaced_terminals, request));
                             if matches!(admission, HealAdmissionResult::Accepted)
                                 && let Some(anchor) = recovery_anchor
                             {
@@ -491,8 +496,16 @@ impl HealManager {
                             }
                             drop(queue);
                             drop(config);
-                            if let Some(displaced_task_id) = admission_decision.displaced_task_id {
-                                remove_task_aliases_for_task(&task_aliases, &displaced_task_id).await;
+                            if let (Some(displaced_task_id), Some(displaced_terminal)) =
+                                (admission_decision.displaced_task_id().map(ToOwned::to_owned), displaced_terminal)
+                            {
+                                remove_displaced_task_aliases(
+                                    &task_aliases,
+                                    &displaced_terminals,
+                                    &displaced_task_id,
+                                    &displaced_terminal,
+                                )
+                                .await;
                                 lock_mrf_repair_notice_targets(&mrf_repair_notice_targets).remove(&displaced_task_id);
                             }
                             if matches!(admission, HealAdmissionResult::Accepted) {

crates/heal/src/heal/manager/scheduler.rs

@@ -21,6 +21,7 @@ impl HealManager {
         let heal_queue = self.heal_queue.clone();
         let active_heals = self.active_heals.clone();
         let completed_heals = self.completed_heals.clone();
+        let displaced_terminals = self.displaced_terminals.clone();
         let task_aliases = self.task_aliases.clone();
         let retrying_heals = self.retrying_heals.clone();
         let mrf_repair_notice_targets = self.mrf_repair_notice_targets.clone();
@@ -53,6 +54,7 @@ impl HealManager {
                             heal_queue: &heal_queue,
                             active_heals: &active_heals,
                             completed_heals: &completed_heals,
+                            displaced_terminals: &displaced_terminals,
                             task_aliases: &task_aliases,
                             retrying_heals: &retrying_heals,
                             mrf_repair_notice_targets: &mrf_repair_notice_targets,
@@ -71,6 +73,7 @@ impl HealManager {
                             heal_queue: &heal_queue,
                             active_heals: &active_heals,
                             completed_heals: &completed_heals,
+                            displaced_terminals: &displaced_terminals,
                             task_aliases: &task_aliases,
                             retrying_heals: &retrying_heals,
                             mrf_repair_notice_targets: &mrf_repair_notice_targets,
@@ -98,6 +101,7 @@ impl HealManager {
             heal_queue,
             active_heals,
             completed_heals,
+            displaced_terminals,
             task_aliases,
             retrying_heals,
             mrf_repair_notice_targets,
@@ -183,6 +187,7 @@ impl HealManager {
                 let active_heals_clone = active_heals.clone();
                 let heal_queue_clone = heal_queue.clone();
                 let completed_heals_clone = completed_heals.clone();
+                let displaced_terminals_clone = displaced_terminals.clone();
                 let task_aliases_clone = task_aliases.clone();
                 let retrying_heals_clone = retrying_heals.clone();
                 let mrf_repair_notice_targets_clone = mrf_repair_notice_targets.clone();
@@ -363,6 +368,7 @@ impl HealManager {
                         let retry_heal_queue = heal_queue_clone.clone();
                         let retrying_heals_for_spawn = retrying_heals_clone.clone();
                         let retry_task_aliases = task_aliases_clone.clone();
+                        let retry_displaced_terminals = displaced_terminals_clone.clone();
                         let retry_mrf_repair_notice_targets = mrf_repair_notice_targets_clone.clone();
                         let retry_completed_heals = completed_heals_clone.clone();
                         let retry_notify = notify_clone.clone();
@@ -430,6 +436,14 @@ impl HealManager {
                                 let admission = admission_decision.result;
                                 let should_notify = matches!(admission, HealAdmissionResult::Accepted)
                                     && retry_config.event_driven_scheduler_enable;
+                                // Publish the terminal synchronously while the
+                                // queue transition is protected. The subsequent
+                                // queue -> retrying handoff retains the lock order
+                                // used by operations_snapshot.
+                                let displaced_terminal = admission_decision
+                                    .displaced_request
+                                    .as_ref()
+                                    .map(|request| record_displaced_terminal(&retry_displaced_terminals, request));
                                 match admission {
                                     HealAdmissionResult::Accepted => {
                                         // Transfer ownership while holding queue -> retrying,
@@ -437,10 +451,18 @@ impl HealManager {
                                         #[cfg(test)]
                                         pause_retry_ownership_transition(&retry_request_id, true).await;
                                         retrying_heals_for_spawn.lock().await.remove(&retry_request_id);
-                                        let displaced_task_id = admission_decision.displaced_task_id;
+                                        let displaced_task_id = admission_decision.displaced_task_id().map(ToOwned::to_owned);
                                         drop(queue);
-                                        if let Some(displaced_task_id) = displaced_task_id {
-                                            remove_task_aliases_for_task(&retry_task_aliases, &displaced_task_id).await;
+                                        if let (Some(displaced_task_id), Some(displaced_terminal)) =
+                                            (displaced_task_id, displaced_terminal)
+                                        {
+                                            remove_displaced_task_aliases(
+                                                &retry_task_aliases,
+                                                &retry_displaced_terminals,
+                                                &displaced_task_id,
+                                                &displaced_terminal,
+                                            )
+                                            .await;
                                             remove_mrf_repair_notice_targets(
                                                 &retry_mrf_repair_notice_targets,
                                                 &displaced_task_id,

crates/heal/src/heal/manager/tests.rs

@@ -84,6 +84,7 @@ async fn process_manager_queue_once(manager: &HealManager) {
         heal_queue: &manager.heal_queue,
         active_heals: &manager.active_heals,
         completed_heals: &manager.completed_heals,
+        displaced_terminals: &manager.displaced_terminals,
         task_aliases: &manager.task_aliases,
         retrying_heals: &manager.retrying_heals,
         mrf_repair_notice_targets: &manager.mrf_repair_notice_targets,
@@ -2778,7 +2779,10 @@ async fn test_high_priority_request_displaces_lower_priority_when_queue_full() {
         HealAdmissionResult::Accepted
     );
     assert_eq!(manager.get_queue_length().await, 1);
-    assert!(matches!(manager.get_task_status(&low_id).await, Err(Error::TaskNotFound { .. })));
+    assert!(matches!(
+        manager.get_task_status(&low_id).await,
+        Ok(HealTaskStatus::Failed { error }) if error.contains("reason=displaced")
+    ));
     assert_eq!(
         manager
             .get_task_status(&high_id)
@@ -2788,6 +2792,263 @@ async fn test_high_priority_request_displaces_lower_priority_when_queue_full() {
     );
 }
 
+#[tokio::test]
+async fn displaced_task_remains_queryable() {
+    let manager = HealManager::new(
+        Arc::new(MockStorage),
+        Some(HealConfig {
+            queue_size: 1,
+            ..HealConfig::default()
+        }),
+    );
+    let mut displaced = HealRequest::new(
+        HealType::Bucket {
+            bucket: "displaced-bucket".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::Low,
+    );
+    displaced.id = "displaced-task".to_string();
+    let displaced_id = displaced.id.clone();
+    manager
+        .submit_heal_request(displaced)
+        .await
+        .expect("displaced request should queue");
+
+    let successor = HealRequest::new(
+        HealType::Bucket {
+            bucket: "successor-bucket".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::High,
+    );
+    manager
+        .submit_heal_request(successor)
+        .await
+        .expect("successor should displace low work");
+
+    let report = manager
+        .get_task_report(&displaced_id)
+        .await
+        .expect("displaced report should remain queryable");
+    assert!(matches!(report.status, HealTaskStatus::Failed { ref error } if error.contains("reason=displaced")));
+}
+
+#[tokio::test]
+async fn displaced_archive_failure_keeps_queryable_terminal() {
+    let manager = HealManager::new(Arc::new(MockStorage), None);
+    let mut request = HealRequest::new(
+        HealType::Bucket {
+            bucket: "archive-failure".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::Low,
+    );
+    request.id = "archive-failure-task".to_string();
+    let request_id = request.id.clone();
+    // The synchronous sidecar is the authoritative fallback when the normal
+    // completed-task archive has no entry (the failure window that must not
+    // turn an Accepted ID into NotFound).
+    record_displaced_terminal(&manager.displaced_terminals, &request);
+    assert!(manager.completed_heals.lock().await.is_empty());
+    assert!(matches!(
+        manager.get_task_status(&request_id).await,
+        Ok(HealTaskStatus::Failed { error }) if error.contains("reason=displaced")
+    ));
+}
+
+#[tokio::test]
+async fn scheduler_retry_displacement_keeps_evicted_task_queryable() {
+    let manager = Arc::new(HealManager::new(
+        Arc::new(MockStorage),
+        Some(HealConfig {
+            queue_size: 1,
+            event_driven_scheduler_enable: false,
+            ..HealConfig::default()
+        }),
+    ));
+    let mut retry_request = HealRequest::object("retry-transition".to_string(), "object".to_string(), None);
+    retry_request.priority = HealPriority::High;
+    let retry_id = retry_request.id.clone();
+    manager
+        .submit_heal_request(retry_request)
+        .await
+        .expect("retry request should queue");
+
+    // Process exactly one queue cycle so the retry task is spawned without a
+    // background scheduler consuming the filler request before the retry wakes.
+    process_manager_queue_once(&manager).await;
+    tokio::time::timeout(Duration::from_secs(1), async {
+        loop {
+            if manager.retrying_heals.lock().await.contains_key(&retry_id) {
+                break;
+            }
+            tokio::task::yield_now().await;
+        }
+    })
+    .await
+    .expect("retry request should enter backoff");
+
+    let filler = HealRequest::new(
+        HealType::Bucket {
+            bucket: "retry-displaced-filler".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::Low,
+    );
+    let filler_id = filler.id.clone();
+    manager
+        .submit_heal_request(filler)
+        .await
+        .expect("filler request should occupy the queue");
+
+    tokio::time::timeout(Duration::from_secs(5), async {
+        loop {
+            if matches!(
+                manager.get_task_status(&filler_id).await,
+                Ok(HealTaskStatus::Failed { ref error }) if error.contains("reason=displaced")
+            ) {
+                break;
+            }
+            tokio::time::sleep(Duration::from_millis(10)).await;
+        }
+    })
+    .await
+    .expect("retry admission should displace the filler request");
+    assert_eq!(manager.get_queue_length().await, 1);
+    assert_eq!(
+        manager.get_task_status(&retry_id).await.expect("retry should be queued"),
+        HealTaskStatus::Pending
+    );
+}
+
+#[tokio::test]
+async fn concurrent_displacers_produce_one_terminal_generation() {
+    let manager = Arc::new(HealManager::new(
+        Arc::new(MockStorage),
+        Some(HealConfig {
+            queue_size: 1,
+            ..HealConfig::default()
+        }),
+    ));
+    let mut displaced = HealRequest::new(
+        HealType::Bucket {
+            bucket: "concurrent-displaced".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::Low,
+    );
+    displaced.id = "concurrent-displaced-task".to_string();
+    let displaced_id = displaced.id.clone();
+    manager
+        .submit_heal_request(displaced)
+        .await
+        .expect("initial request should queue");
+
+    let first = HealRequest::new(
+        HealType::Bucket {
+            bucket: "concurrent-successor-a".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::High,
+    );
+    let second = HealRequest::new(
+        HealType::Bucket {
+            bucket: "concurrent-successor-b".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::High,
+    );
+    let (first_result, second_result) = tokio::join!(manager.submit_heal_request(first), manager.submit_heal_request(second));
+    let accepted = [&first_result, &second_result]
+        .into_iter()
+        .filter(|result| matches!(result, Ok(HealAdmissionResult::Accepted)))
+        .count();
+    assert_eq!(accepted, 1, "exactly one concurrent displacer should win the full queue");
+    assert!(
+        first_result.is_ok() && second_result.is_ok(),
+        "the losing request should receive a typed Full result"
+    );
+    let terminals = lock_displaced_terminals(&manager.displaced_terminals);
+    assert_eq!(terminals.len(), 1);
+    assert!(terminals.contains_key(&displaced_id));
+}
+
+#[tokio::test]
+async fn successor_chain_is_bounded_and_authorized() {
+    let manager = HealManager::new(
+        Arc::new(MockStorage),
+        Some(HealConfig {
+            queue_size: 1,
+            ..HealConfig::default()
+        }),
+    );
+    let mut original = HealRequest::new(
+        HealType::Bucket {
+            bucket: "authorized-original".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::Low,
+    );
+    original.id = "authorized-original-task".to_string();
+    let original_id = original.id.clone();
+    manager.submit_heal_request(original).await.expect("original should queue");
+    let mut duplicate = HealRequest::new(
+        HealType::Bucket {
+            bucket: "authorized-original".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::Low,
+    );
+    duplicate.id = "authorized-duplicate-task".to_string();
+    let duplicate_id = duplicate.id.clone();
+    manager
+        .submit_heal_request(duplicate)
+        .await
+        .expect("same-target duplicate should merge");
+    let successor = HealRequest::new(
+        HealType::Bucket {
+            bucket: "authorized-successor".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::High,
+    );
+    let successor_id = successor.id.clone();
+    manager.submit_heal_request(successor).await.expect("successor should queue");
+    assert!(manager.task_aliases.lock().await.is_empty());
+    assert!(matches!(manager.get_task_status(&original_id).await, Ok(HealTaskStatus::Failed { .. })));
+    assert!(matches!(manager.get_task_status(&duplicate_id).await, Ok(HealTaskStatus::Failed { .. })));
+    assert_eq!(
+        manager
+            .get_task_status(&successor_id)
+            .await
+            .expect("successor should remain queued"),
+        HealTaskStatus::Pending
+    );
+}
+
+#[tokio::test]
+async fn displaced_terminal_expires_after_bounded_ttl() {
+    let manager = HealManager::new(Arc::new(MockStorage), None);
+    let mut request = HealRequest::new(
+        HealType::Bucket {
+            bucket: "expires".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::Low,
+    );
+    request.id = "expires-task".to_string();
+    let request_id = request.id.clone();
+    record_displaced_terminal(&manager.displaced_terminals, &request);
+    {
+        let mut terminals = lock_displaced_terminals(&manager.displaced_terminals);
+        let entry =
+            Arc::get_mut(terminals.get_mut(&request_id).expect("terminal should be retained")).expect("test owns terminal entry");
+        entry.completed_at = SystemTime::now() - KEEP_HEAL_TASK_STATUS_DURATION - Duration::from_secs(1);
+    }
+    assert!(matches!(manager.get_task_status(&request_id).await, Err(Error::TaskNotFound { .. })));
+}
+
 #[tokio::test]
 async fn test_displacing_registered_mrf_task_drops_notice_ownership() {
     let storage: Arc<dyn HealStorageAPI> = Arc::new(MockStorage);

crates/heal/src/heal/resume.rs

@@ -51,7 +51,6 @@ const RESUME_STATE_FILE: &str = "ahm_resume_state.json";
 const REPLACEMENT_INTENT_FILE: &str = "ahm_replacement_intent.json";
 const RESUME_PROGRESS_FILE: &str = "ahm_progress.json";
 pub(super) const RESUME_CHECKPOINT_FILE: &str = "ahm_checkpoint.json";
-pub(super) const RESUME_CHECKPOINT_BLOCKED_FILE: &str = "ahm_checkpoint.blocked";
 const REPLACEMENT_COMPLETION_PROOF_FILE: &str = "ahm_replacement_completion_proof.json";
 const REPLACEMENT_RECOVERY_DIR: &str = "ahm-replacement";
 const REPLACEMENT_INTENT_SEAL_FILE: &str = "ahm_replacement_intent_seal";

crates/heal/src/heal/resume/checkpoint.rs

@@ -13,25 +13,21 @@
 // limitations under the License.
 
 use crate::{Error, Result};
-use base64::Engine as _;
 use serde::{Deserialize, Serialize};
-use sha2::{Digest, Sha256};
 use std::collections::HashSet;
 use std::path::Path;
 use std::sync::{Arc, Mutex};
 use std::time::{SystemTime, UNIX_EPOCH};
-use tokio::sync::{Mutex as AsyncMutex, RwLock};
+use tokio::sync::RwLock;
 use tracing::{debug, warn};
 
-use super::super::storage_api::owner::{EcstoreConditionalFileUpdate, EcstoreDiskAPI, EcstoreDiskBytes};
-use super::super::{BUCKET_META_PREFIX, DiskStore, HealDiskExt, RUSTFS_META_BUCKET};
+use super::super::{BUCKET_META_PREFIX, DiskStore, HealDiskExt as _, RUSTFS_META_BUCKET};
 use super::{
-    LOG_COMPONENT_HEAL, LOG_SUBSYSTEM_RESUME, PersistThrottle, RESUME_CHECKPOINT_BLOCKED_FILE, RESUME_CHECKPOINT_FILE,
-    delete_resume_file, path_to_str, validate_resume_task_id,
+    LOG_COMPONENT_HEAL, LOG_SUBSYSTEM_RESUME, PersistThrottle, RESUME_CHECKPOINT_FILE, delete_resume_file, path_to_str,
+    validate_resume_task_id,
 };
 
 const EVENT_HEAL_CHECKPOINT_STATE: &str = "heal_checkpoint_state";
-const RESUME_CHECKPOINT_DIGEST_FILE: &str = "ahm_checkpoint.sha256";
 
 /// Current on-disk schema version for `ResumeCheckpoint`. Same rationale as
 /// `CURRENT_RESUME_SCHEMA`: pre-per-version dedup identities are not comparable
@@ -120,111 +116,17 @@ pub struct CheckpointManager {
     disk: DiskStore,
     checkpoint: Arc<RwLock<ResumeCheckpoint>>,
     throttle: Mutex<PersistThrottle>,
-    save_lock: AsyncMutex<()>,
-    last_saved: Mutex<Option<EcstoreDiskBytes>>,
 }
 
 impl CheckpointManager {
-    fn blocked_path(task_id: &str) -> std::path::PathBuf {
-        Path::new(BUCKET_META_PREFIX).join(format!("{task_id}_{RESUME_CHECKPOINT_BLOCKED_FILE}"))
-    }
-
-    /// Return whether a checkpoint was permanently isolated after a malformed
-    /// or unsupported snapshot was observed.
-    pub(crate) async fn is_blocked(disk: &DiskStore, task_id: &str) -> bool {
-        if validate_resume_task_id(task_id).is_err() {
-            return false;
-        }
-        let blocked_path = Self::blocked_path(task_id);
-        let Ok(path) = path_to_str(&blocked_path) else {
-            return false;
-        };
-        match HealDiskExt::read_all(disk.as_ref(), RUSTFS_META_BUCKET, path).await {
-            Ok(_) => true,
-            Err(crate::heal::DiskError::FileNotFound) => false,
-            Err(_) => true,
-        }
-    }
-
-    /// Validate the checkpoint while enumerating resumable state. This reads
-    /// the checkpoint once and also isolates malformed or unsupported data.
-    pub(crate) async fn is_resumable(disk: &DiskStore, task_id: &str) -> Result<bool> {
-        validate_resume_task_id(task_id)?;
-        if Self::is_blocked(disk, task_id).await {
-            return Err(Error::InvalidCheckpoint(format!("Resume task {task_id} has a blocked checkpoint")));
-        }
-        let file_path = Path::new(BUCKET_META_PREFIX).join(format!("{task_id}_{RESUME_CHECKPOINT_FILE}"));
-        let Ok(path) = path_to_str(&file_path) else {
-            return Err(Error::InvalidCheckpoint("Resume checkpoint path is not valid UTF-8".to_string()));
-        };
-        match HealDiskExt::read_all(disk.as_ref(), RUSTFS_META_BUCKET, path).await {
-            Ok(bytes) if bytes.is_empty() => Ok(true),
-            Ok(bytes) => Self::load_from_data(disk.clone(), task_id, bytes.to_vec())
-                .await
-                .map(|_| true),
-            Err(crate::heal::DiskError::FileNotFound) => Ok(true),
-            Err(error) => Err(error.into()),
-        }
-    }
-
-    async fn block_invalid_snapshot(disk: &DiskStore, task_id: &str) {
-        // This marker is intentionally version-agnostic: an unsupported reader
-        // must stop selector retries until an operator cleans up the snapshot.
-        let blocked_path = Self::blocked_path(task_id);
-        let Ok(path) = path_to_str(&blocked_path) else {
-            return;
-        };
-        let result = EcstoreDiskAPI::compare_and_update_file(
-            disk.as_ref(),
-            RUSTFS_META_BUCKET,
-            path,
-            None,
-            Some(EcstoreDiskBytes::from_static(b"blocked")),
-        )
-        .await;
-        match result {
-            Ok(EcstoreConditionalFileUpdate::Updated | EcstoreConditionalFileUpdate::Mismatch) => {}
-            Ok(EcstoreConditionalFileUpdate::Missing) => warn!(
-                target: "rustfs::heal::resume",
-                event = EVENT_HEAL_CHECKPOINT_STATE,
-                component = LOG_COMPONENT_HEAL,
-                subsystem = LOG_SUBSYSTEM_RESUME,
-                task_id,
-                state = "blocked_marker_write_failed",
-                error = "marker target disappeared",
-                "Heal checkpoint could not persist its blocked marker"
-            ),
-            Err(error) => warn!(
-                target: "rustfs::heal::resume",
-                event = EVENT_HEAL_CHECKPOINT_STATE,
-                component = LOG_COMPONENT_HEAL,
-                subsystem = LOG_SUBSYSTEM_RESUME,
-                task_id,
-                state = "blocked_marker_write_failed",
-                error = %error,
-                "Heal checkpoint could not persist its blocked marker"
-            ),
-        }
-    }
-
     /// create new checkpoint manager
     pub async fn new(disk: DiskStore, task_id: String) -> Result<Self> {
         validate_resume_task_id(&task_id)?;
-        let checkpoint_volume = format!("{RUSTFS_META_BUCKET}/{BUCKET_META_PREFIX}");
-        if let Err(error) = EcstoreDiskAPI::make_volume(disk.as_ref(), &checkpoint_volume).await
-            && error != crate::heal::DiskError::VolumeExists
-        {
-            return Err(Error::TaskExecutionFailed {
-                message: format!("Failed to create checkpoint volume: {error}"),
-            });
-        }
         let checkpoint = ResumeCheckpoint::new(task_id);
         let manager = Self {
             disk,
             checkpoint: Arc::new(RwLock::new(checkpoint)),
             throttle: Mutex::new(PersistThrottle::new()),
-            save_lock: AsyncMutex::new(()),
-            last_saved: Mutex::new(None),
         };
 
         // save initial checkpoint
@@ -238,7 +140,6 @@ impl CheckpointManager {
                 error = %e,
                 "Heal checkpoint persistence failed"
             );
-            return Err(e);
         }
         Ok(manager)
     }
@@ -247,22 +148,11 @@ impl CheckpointManager {
     pub async fn load_from_disk(disk: DiskStore, task_id: &str) -> Result<Self> {
         validate_resume_task_id(task_id)?;
         let checkpoint_data = Self::read_checkpoint_file(&disk, task_id).await?;
-        Self::load_from_data(disk, task_id, checkpoint_data).await
-    }
-
-    async fn load_from_data(disk: DiskStore, task_id: &str, checkpoint_data: Vec<u8>) -> Result<Self> {
-        validate_resume_task_id(task_id)?;
-        let mut checkpoint: ResumeCheckpoint = match serde_json::from_slice(&checkpoint_data) {
-            Ok(checkpoint) => checkpoint,
-            Err(error) => {
-                Self::block_invalid_snapshot(&disk, task_id).await;
-                return Err(Error::TaskExecutionFailed {
-                    message: format!("Failed to deserialize checkpoint: {error}"),
-                });
-            }
-        };
+        let mut checkpoint: ResumeCheckpoint =
+            serde_json::from_slice(&checkpoint_data).map_err(|e| Error::TaskExecutionFailed {
+                message: format!("Failed to deserialize checkpoint: {e}"),
+            })?;
         if checkpoint.task_id != task_id {
-            Self::block_invalid_snapshot(&disk, task_id).await;
             return Err(Error::TaskExecutionFailed {
                 message: "Resume checkpoint task id does not match filename".to_string(),
             });
@@ -273,7 +163,6 @@ impl CheckpointManager {
         // identities. Discard the stale sets and position, then stamp the
         // current schema so the scan restarts cleanly.
         if checkpoint.schema_version > CURRENT_CHECKPOINT_SCHEMA {
-            Self::block_invalid_snapshot(&disk, task_id).await;
             return Err(Error::TaskExecutionFailed {
                 message: format!(
                     "Checkpoint schema {} is newer than supported schema {CURRENT_CHECKPOINT_SCHEMA}",
@@ -305,8 +194,6 @@ impl CheckpointManager {
             disk,
             checkpoint: Arc::new(RwLock::new(checkpoint)),
             throttle: Mutex::new(PersistThrottle::new()),
-            save_lock: AsyncMutex::new(()),
-            last_saved: Mutex::new(Some(EcstoreDiskBytes::from(checkpoint_data))),
         })
     }
 
@@ -317,7 +204,7 @@ impl CheckpointManager {
         }
         let file_path = Path::new(BUCKET_META_PREFIX).join(format!("{task_id}_{RESUME_CHECKPOINT_FILE}"));
         match path_to_str(&file_path) {
-            Ok(path_str) => match HealDiskExt::read_all(disk.as_ref(), RUSTFS_META_BUCKET, path_str).await {
+            Ok(path_str) => match disk.read_all(RUSTFS_META_BUCKET, path_str).await {
                 Ok(data) => !data.is_empty(),
                 Err(_) => false,
             },
@@ -405,8 +292,6 @@ impl CheckpointManager {
 
         let checkpoint_file = Path::new(BUCKET_META_PREFIX).join(format!("{task_id}_{RESUME_CHECKPOINT_FILE}"));
         delete_resume_file(&self.disk, &checkpoint_file).await?;
-        delete_resume_file(&self.disk, &Self::digest_path(&task_id)).await?;
-        delete_resume_file(&self.disk, &Self::blocked_path(&task_id)).await?;
 
         debug!(
             target: "rustfs::heal::resume",
@@ -422,139 +307,21 @@ impl CheckpointManager {
 
     /// save checkpoint to disk
     async fn save_checkpoint(&self) -> Result<()> {
-        // Serialize saves and take the snapshot only after acquiring the lock:
-        // a slower writer must not publish a snapshot taken before a newer one.
-        let _save_guard = self.save_lock.lock().await;
-        let checkpoint = self.checkpoint.read().await.clone();
+        let checkpoint = self.checkpoint.read().await;
         validate_resume_task_id(&checkpoint.task_id)?;
-        let checkpoint_data =
-            EcstoreDiskBytes::from(serde_json::to_vec(&checkpoint).map_err(|e| Error::TaskExecutionFailed {
-                message: format!("Failed to serialize checkpoint: {e}"),
-            })?);
+        let checkpoint_data = serde_json::to_vec(&*checkpoint).map_err(|e| Error::TaskExecutionFailed {
+            message: format!("Failed to serialize checkpoint: {e}"),
+        })?;
 
         let file_path = Path::new(BUCKET_META_PREFIX).join(format!("{}_{}", checkpoint.task_id, RESUME_CHECKPOINT_FILE));
 
         let path_str = path_to_str(&file_path)?;
-        let last_saved = self
-            .last_saved
-            .lock()
-            .map_err(|_| Error::TaskExecutionFailed {
-                message: "Checkpoint save state lock is poisoned; refusing to save".to_string(),
-            })?
-            .clone();
-        let update = EcstoreDiskAPI::compare_and_update_file(
-            self.disk.as_ref(),
-            RUSTFS_META_BUCKET,
-            path_str,
-            last_saved.clone(),
-            Some(checkpoint_data.clone()),
-        )
-        .await
-        .map_err(|e| Error::TaskExecutionFailed {
-            message: format!("Failed to save checkpoint: {e}"),
-        })?;
-
-        let expected = match update {
-            EcstoreConditionalFileUpdate::Updated => None,
-            EcstoreConditionalFileUpdate::Missing => {
-                return Err(Error::TaskExecutionFailed {
-                    message: "Checkpoint was removed after this manager saved it; refusing to recreate it".to_string(),
-                });
-            }
-            EcstoreConditionalFileUpdate::Mismatch => {
-                // A healthy manager normally completes the CAS above without
-                // another read or JSON parse. Inspect only after a mismatch so
-                // corruption and future schemas cannot be overwritten blindly.
-                let existing = match HealDiskExt::read_all(self.disk.as_ref(), RUSTFS_META_BUCKET, path_str).await {
-                    Ok(existing) => existing,
-                    Err(crate::heal::DiskError::FileNotFound) => {
-                        return Err(Error::TaskExecutionFailed {
-                            message: "Checkpoint was removed after this manager saved it; refusing to recreate it".to_string(),
-                        });
-                    }
-                    Err(error) => {
-                        return Err(Error::TaskExecutionFailed {
-                            message: format!("Failed to inspect checkpoint after CAS mismatch: {error}"),
-                        });
-                    }
-                };
-
-                if existing.is_empty() && last_saved.is_none() {
-                    Some(existing)
-                } else {
-                    let current: ResumeCheckpoint = match serde_json::from_slice(&existing) {
-                        Ok(current) => current,
-                        Err(error) => {
-                            Self::block_invalid_snapshot(&self.disk, &checkpoint.task_id).await;
-                            return Err(Error::TaskExecutionFailed {
-                                message: format!("Existing checkpoint is corrupt: {error}"),
-                            });
-                        }
-                    };
-                    if current.task_id != checkpoint.task_id {
-                        Self::block_invalid_snapshot(&self.disk, &checkpoint.task_id).await;
-                        return Err(Error::TaskExecutionFailed {
-                            message: "Existing checkpoint task id does not match filename".to_string(),
-                        });
-                    }
-                    if current.schema_version > CURRENT_CHECKPOINT_SCHEMA {
-                        Self::block_invalid_snapshot(&self.disk, &checkpoint.task_id).await;
-                        return Err(Error::TaskExecutionFailed {
-                            message: format!(
-                                "Existing checkpoint schema {} is newer than supported schema {CURRENT_CHECKPOINT_SCHEMA}",
-                                current.schema_version
-                            ),
-                        });
-                    }
-                    if last_saved.as_ref().is_none_or(|saved| saved.as_ref() != existing.as_ref()) {
-                        return Err(Error::TaskExecutionFailed {
-                            message: "Checkpoint changed since this manager loaded it; refusing to overwrite newer progress"
-                                .to_string(),
-                        });
-                    }
-                    Some(existing)
-                }
-            }
-        };
-
-        if let Some(expected) = expected {
-            match EcstoreDiskAPI::compare_and_update_file(
-                self.disk.as_ref(),
-                RUSTFS_META_BUCKET,
-                path_str,
-                Some(expected),
-                Some(checkpoint_data.clone()),
-            )
+        self.disk
+            .write_all(RUSTFS_META_BUCKET, path_str, checkpoint_data.into())
             .await
             .map_err(|e| Error::TaskExecutionFailed {
-                message: format!("Failed to save checkpoint after CAS mismatch: {e}"),
-            })? {
-                EcstoreConditionalFileUpdate::Updated => {}
-                EcstoreConditionalFileUpdate::Missing | EcstoreConditionalFileUpdate::Mismatch => {
-                    return Err(Error::TaskExecutionFailed {
-                        message: "Checkpoint changed while saving; refusing to overwrite newer progress".to_string(),
-                    });
-                }
-            }
-        }
-
-        let digest_path = Self::digest_path(&checkpoint.task_id);
-        let digest = base64::engine::general_purpose::STANDARD.encode(Sha256::digest(checkpoint_data.as_ref()));
-        HealDiskExt::write_all(
-            self.disk.as_ref(),
-            RUSTFS_META_BUCKET,
-            path_to_str(&digest_path)?,
-            EcstoreDiskBytes::from(digest.into_bytes()),
-        )
-        .await
-        .map_err(|e| Error::TaskExecutionFailed {
-            message: format!("Failed to save checkpoint digest: {e}"),
-        })?;
-
-        let mut last_saved = self.last_saved.lock().map_err(|_| Error::TaskExecutionFailed {
-            message: "Checkpoint save state lock is poisoned after save".to_string(),
-        })?;
-        *last_saved = Some(checkpoint_data);
+                message: format!("Failed to save checkpoint: {e}"),
+            })?;
 
         debug!(
             target: "rustfs::heal::resume",
@@ -574,35 +341,11 @@ impl CheckpointManager {
         let file_path = Path::new(BUCKET_META_PREFIX).join(format!("{task_id}_{RESUME_CHECKPOINT_FILE}"));
 
         let path_str = path_to_str(&file_path)?;
-        let checkpoint = HealDiskExt::read_all(disk.as_ref(), RUSTFS_META_BUCKET, path_str)
+        disk.read_all(RUSTFS_META_BUCKET, path_str)
             .await
             .map(|bytes| bytes.to_vec())
             .map_err(|e| Error::TaskExecutionFailed {
                 message: format!("Failed to read checkpoint file: {e}"),
-            })?;
-        let digest_path = Self::digest_path(task_id);
-        let digest_path = path_to_str(&digest_path)?;
-        match HealDiskExt::read_all(disk.as_ref(), RUSTFS_META_BUCKET, digest_path).await {
-            Ok(expected) => {
-                let actual = base64::engine::general_purpose::STANDARD.encode(Sha256::digest(&checkpoint));
-                if expected.as_ref() != actual.as_bytes() {
-                    Self::block_invalid_snapshot(disk, task_id).await;
-                    return Err(Error::InvalidCheckpoint(format!(
-                        "Resume checkpoint digest does not match task {task_id}"
-                    )));
-                }
-            }
-            Err(crate::heal::DiskError::FileNotFound) => {}
-            Err(error) => {
-                return Err(Error::TaskExecutionFailed {
-                    message: format!("Failed to read checkpoint digest: {error}"),
-                });
-            }
-        }
-        Ok(checkpoint)
-    }
-
-    fn digest_path(task_id: &str) -> std::path::PathBuf {
-        Path::new(BUCKET_META_PREFIX).join(format!("{task_id}_{RESUME_CHECKPOINT_DIGEST_FILE}"))
+            })
     }
 }

crates/heal/src/heal/resume/tests.rs

@@ -1675,285 +1675,6 @@ async fn future_resume_and_checkpoint_schemas_are_rejected() {
     temp_dir.close().expect("remove schema test directory");
 }
 
-#[tokio::test]
-async fn checkpoint_save_does_not_replace_a_non_empty_truncated_snapshot() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let task_id = ResumeUtils::generate_task_id();
-    let manager = CheckpointManager::new(disk.clone(), task_id.clone())
-        .await
-        .expect("create checkpoint manager");
-    let checkpoint_path = format!("{BUCKET_META_PREFIX}/{task_id}_{RESUME_CHECKPOINT_FILE}");
-    let truncated = b"{\"schema_version\":5,\"task_id\":";
-    disk.write_all(RUSTFS_META_BUCKET, &checkpoint_path, truncated.as_slice().into())
-        .await
-        .expect("write truncated checkpoint fixture");
-
-    let error = manager
-        .update_position(2, 7)
-        .await
-        .expect_err("a truncated checkpoint must fail closed during save");
-    assert!(error.to_string().contains("Existing checkpoint is corrupt"));
-    assert_eq!(
-        disk.read_all(RUSTFS_META_BUCKET, &checkpoint_path)
-            .await
-            .expect("read truncated checkpoint fixture"),
-        truncated.as_slice()
-    );
-    assert!(CheckpointManager::is_blocked(&disk, &task_id).await);
-    temp_dir.close().expect("remove checkpoint save test directory");
-}
-
-#[tokio::test]
-async fn checkpoint_save_does_not_replace_a_future_schema_snapshot() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let task_id = ResumeUtils::generate_task_id();
-    let manager = CheckpointManager::new(disk.clone(), task_id.clone())
-        .await
-        .expect("create checkpoint manager");
-    let checkpoint_path = format!("{BUCKET_META_PREFIX}/{task_id}_{RESUME_CHECKPOINT_FILE}");
-    let mut future = ResumeCheckpoint::new(task_id.clone());
-    future.schema_version = CURRENT_CHECKPOINT_SCHEMA + 1;
-    let future_bytes = serde_json::to_vec(&future).expect("serialize future checkpoint fixture");
-    disk.write_all(RUSTFS_META_BUCKET, &checkpoint_path, future_bytes.clone().into())
-        .await
-        .expect("write future checkpoint fixture");
-
-    let error = manager
-        .update_position(2, 7)
-        .await
-        .expect_err("a future schema must fail closed during save");
-    assert!(error.to_string().contains("Existing checkpoint schema"));
-    assert_eq!(
-        disk.read_all(RUSTFS_META_BUCKET, &checkpoint_path)
-            .await
-            .expect("read future checkpoint fixture"),
-        future_bytes
-    );
-    assert!(CheckpointManager::is_blocked(&disk, &task_id).await);
-    temp_dir.close().expect("remove future schema test directory");
-}
-
-#[tokio::test]
-async fn checkpoint_digest_rejects_same_length_progress_tampering() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let task_id = ResumeUtils::generate_task_id();
-    let manager = CheckpointManager::new(disk.clone(), task_id.clone())
-        .await
-        .expect("create checkpoint manager");
-    manager
-        .add_processed_object("victim-a".to_string())
-        .await
-        .expect("persist checkpoint progress");
-    manager.update_position(1, 1).await.expect("flush checkpoint progress");
-    let checkpoint_path = format!("{BUCKET_META_PREFIX}/{task_id}_{RESUME_CHECKPOINT_FILE}");
-    let original = disk
-        .read_all(RUSTFS_META_BUCKET, &checkpoint_path)
-        .await
-        .expect("read checkpoint fixture");
-    let tampered = original
-        .windows(b"victim-a".len())
-        .position(|window| window == b"victim-a")
-        .map(|index| {
-            let mut bytes = original.to_vec();
-            bytes[index..index + b"victim-a".len()].copy_from_slice(b"victim-b");
-            bytes
-        })
-        .expect("checkpoint should contain the processed object");
-    disk.write_all(RUSTFS_META_BUCKET, &checkpoint_path, tampered.into())
-        .await
-        .expect("write tampered checkpoint fixture");
-
-    assert!(CheckpointManager::load_from_disk(disk.clone(), &task_id).await.is_err());
-    assert!(CheckpointManager::is_blocked(&disk, &task_id).await);
-    temp_dir.close().expect("remove digest test directory");
-}
-
-#[tokio::test]
-async fn new_checkpoint_manager_rebuilds_an_empty_snapshot() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let task_id = ResumeUtils::generate_task_id();
-    let checkpoint_path = format!("{BUCKET_META_PREFIX}/{task_id}_{RESUME_CHECKPOINT_FILE}");
-    disk.write_all(RUSTFS_META_BUCKET, &checkpoint_path, EcstoreDiskBytes::new())
-        .await
-        .expect("write empty checkpoint fixture");
-
-    let manager = CheckpointManager::new(disk.clone(), task_id.clone())
-        .await
-        .expect("a new manager must rebuild an empty checkpoint");
-    manager
-        .update_position(3, 11)
-        .await
-        .expect("rebuilt checkpoint must remain writable");
-    assert!(CheckpointManager::has_checkpoint(&disk, &task_id).await);
-    temp_dir.close().expect("remove empty checkpoint test directory");
-}
-
-#[tokio::test]
-async fn deleted_checkpoint_is_not_recreated_by_an_old_manager() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let task_id = ResumeUtils::generate_task_id();
-    let manager = CheckpointManager::new(disk.clone(), task_id.clone())
-        .await
-        .expect("create checkpoint manager");
-    manager.cleanup().await.expect("delete checkpoint fixture");
-
-    let error = manager
-        .update_position(1, 2)
-        .await
-        .expect_err("an old manager must not resurrect a deleted checkpoint");
-    assert!(error.to_string().contains("removed after this manager saved it"));
-    assert!(!CheckpointManager::has_checkpoint(&disk, &task_id).await);
-    temp_dir.close().expect("remove deleted checkpoint test directory");
-}
-
-#[tokio::test]
-async fn an_empty_blocked_marker_still_blocks_resume_selection() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let task_id = ResumeUtils::generate_task_id();
-    let manager = CheckpointManager::new(disk.clone(), task_id.clone())
-        .await
-        .expect("create checkpoint manager");
-    let blocked_path = format!("{BUCKET_META_PREFIX}/{task_id}_{RESUME_CHECKPOINT_BLOCKED_FILE}");
-    disk.write_all(RUSTFS_META_BUCKET, &blocked_path, EcstoreDiskBytes::new())
-        .await
-        .expect("write empty blocked marker fixture");
-
-    assert!(CheckpointManager::is_blocked(&disk, &task_id).await);
-    assert!(CheckpointManager::is_resumable(&disk, &task_id).await.is_err());
-    // Recovery requires replacing/cleaning the snapshot, then removing the
-    // marker; ordinary selector retries are intentionally not an unlock path.
-    manager.cleanup().await.expect("clean blocked checkpoint");
-    assert!(!CheckpointManager::is_blocked(&disk, &task_id).await);
-    temp_dir.close().expect("remove empty blocked marker test directory");
-}
-
-#[tokio::test]
-async fn resumable_selector_skips_healthy_tasks_with_blocked_markers() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let tasks = [
-        (ResumeUtils::generate_task_id(), EcstoreDiskBytes::new()),
-        (ResumeUtils::generate_task_id(), EcstoreDiskBytes::from_static(b"blocked")),
-    ];
-    for (task_id, marker) in &tasks {
-        ResumeManager::new(
-            disk.clone(),
-            task_id.clone(),
-            "erasure_set".to_string(),
-            "pool_0_set_0".to_string(),
-            vec!["bucket".to_string()],
-        )
-        .await
-        .expect("create healthy resume state");
-        CheckpointManager::new(disk.clone(), task_id.clone())
-            .await
-            .expect("create healthy checkpoint");
-        let checkpoint_path = format!("{BUCKET_META_PREFIX}/{task_id}_{RESUME_CHECKPOINT_FILE}");
-        let checkpoint_bytes = disk
-            .read_all(RUSTFS_META_BUCKET, &checkpoint_path)
-            .await
-            .expect("read healthy checkpoint before blocking");
-        let marker_path = format!("{BUCKET_META_PREFIX}/{task_id}_{RESUME_CHECKPOINT_BLOCKED_FILE}");
-        disk.write_all(RUSTFS_META_BUCKET, &marker_path, marker.clone())
-            .await
-            .expect("write blocked marker");
-
-        assert!(ResumeUtils::get_resumable_tasks(&disk).await.is_err());
-        assert_eq!(
-            disk.read_all(RUSTFS_META_BUCKET, &checkpoint_path)
-                .await
-                .expect("read healthy checkpoint after blocking"),
-            checkpoint_bytes
-        );
-    }
-    temp_dir.close().expect("remove blocked selector test directory");
-}
-
-#[tokio::test]
-async fn stale_checkpoint_manager_cannot_overwrite_newer_progress() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let task_id = ResumeUtils::generate_task_id();
-    let first = CheckpointManager::new(disk.clone(), task_id.clone())
-        .await
-        .expect("create first checkpoint manager");
-    let second = CheckpointManager::load_from_disk(disk.clone(), &task_id)
-        .await
-        .expect("load second checkpoint manager");
-
-    second
-        .update_position(4, 20)
-        .await
-        .expect("persist newer checkpoint progress");
-    let error = first
-        .update_position(1, 3)
-        .await
-        .expect_err("stale checkpoint manager must not overwrite newer progress");
-    assert!(error.to_string().contains("newer progress"));
-
-    let persisted = CheckpointManager::load_from_disk(disk.clone(), &task_id)
-        .await
-        .expect("load newer checkpoint progress")
-        .get_checkpoint()
-        .await;
-    assert_eq!(persisted.current_bucket_index, 4);
-    assert_eq!(persisted.current_object_index, 20);
-    temp_dir.close().expect("remove stale manager test directory");
-}
-
-#[tokio::test]
-async fn resumable_selector_isolates_future_and_corrupt_checkpoints() {
-    let (temp_dir, disk) = schema_test_disk().await;
-    let future_task = ResumeUtils::generate_task_id();
-    let corrupt_task = ResumeUtils::generate_task_id();
-    for task_id in [&future_task, &corrupt_task] {
-        ResumeManager::new(
-            disk.clone(),
-            task_id.to_string(),
-            "erasure_set".to_string(),
-            "pool_0_set_0".to_string(),
-            vec!["bucket".to_string()],
-        )
-        .await
-        .expect("create resumable state fixture");
-    }
-
-    let future_path = format!("{BUCKET_META_PREFIX}/{future_task}_{RESUME_CHECKPOINT_FILE}");
-    let mut future = ResumeCheckpoint::new(future_task.clone());
-    future.schema_version = CURRENT_CHECKPOINT_SCHEMA + 1;
-    let future_bytes = serde_json::to_vec(&future).expect("serialize future checkpoint fixture");
-    disk.write_all(RUSTFS_META_BUCKET, &future_path, future_bytes.clone().into())
-        .await
-        .expect("write future checkpoint fixture");
-    let corrupt_path = format!("{BUCKET_META_PREFIX}/{corrupt_task}_{RESUME_CHECKPOINT_FILE}");
-    let corrupt_bytes = b"{truncated";
-    disk.write_all(RUSTFS_META_BUCKET, &corrupt_path, corrupt_bytes.as_slice().into())
-        .await
-        .expect("write corrupt checkpoint fixture");
-
-    assert!(CheckpointManager::is_resumable(&disk, &future_task).await.is_err());
-    assert!(CheckpointManager::is_resumable(&disk, &corrupt_task).await.is_err());
-    assert!(ResumeUtils::get_resumable_tasks(&disk).await.is_err());
-    for (task_id, path, bytes) in [
-        (&future_task, future_path, future_bytes),
-        (&corrupt_task, corrupt_path, corrupt_bytes.to_vec()),
-    ] {
-        assert_eq!(
-            disk.read_all(RUSTFS_META_BUCKET, &path)
-                .await
-                .expect("read isolated checkpoint bytes"),
-            bytes
-        );
-        let blocked_path = format!("{BUCKET_META_PREFIX}/{task_id}_{RESUME_CHECKPOINT_BLOCKED_FILE}");
-        assert!(
-            !disk
-                .read_all(RUSTFS_META_BUCKET, &blocked_path)
-                .await
-                .expect("read checkpoint blocked marker")
-                .is_empty()
-        );
-    }
-    temp_dir.close().expect("remove selector isolation test directory");
-}
-
 #[test]
 fn test_persist_throttle_batches_until_threshold() {
     let mut throttle = PersistThrottle::new();

crates/heal/src/heal/resume/utils.rs

@@ -21,7 +21,7 @@ use uuid::Uuid;
 use super::super::{BUCKET_META_PREFIX, DiskError, DiskStore, HealDiskExt as _, RUSTFS_META_BUCKET};
 use super::replacement::{ReplacementPhase, ReplacementRecoveryRecord};
 use super::{
-    CheckpointManager, EVENT_HEAL_RESUME_STATE, LOG_COMPONENT_HEAL, LOG_SUBSYSTEM_RESUME, REPLACEMENT_COMPLETION_PROOF_FILE,
+    EVENT_HEAL_RESUME_STATE, LOG_COMPONENT_HEAL, LOG_SUBSYSTEM_RESUME, REPLACEMENT_COMPLETION_PROOF_FILE,
     REPLACEMENT_INTENT_FILE, RESUME_STATE_FILE, ResumeManager, ResumeStateFile, is_replacement_intent, path_to_str,
     replacement_recovery_corruption_for_state_load, replacement_recovery_dir, validate_resume_task_id,
 };
@@ -67,7 +67,6 @@ impl ResumeUtils {
                 // Extract task ID from filename: {task_id}_ahm_resume_state.json
                 if let Some(task_id) = entry.strip_suffix(&format!("_{RESUME_STATE_FILE}"))
                     && validate_resume_task_id(task_id).is_ok()
-                    && CheckpointManager::is_resumable(disk, task_id).await?
                 {
                     task_ids.push(task_id.to_string());
                 }

crates/kms/tests/vault_ha_failover_live.rs

@@ -16,14 +16,14 @@
 //!
 //! `scripts/test/vault_ha_kms_live.sh` owns the official Vault containers and
 //! kills the active node while this test continuously decrypts through a
-//! surviving standby. KV2 and Transit requests must remain successful, use a
-//! bounded number of attempts, and leave the circuit and in-flight gauges at
-//! zero after a new leader is elected.
+//! surviving standby. KV2 and Transit must recover after the bounded circuit
+//! interval, use a bounded number of attempts, and leave the circuit and
+//! in-flight gauges at zero after a new leader is elected.
 
 use std::collections::HashMap;
 use std::path::{Path, PathBuf};
-use std::sync::Arc;
 use std::sync::atomic::{AtomicBool, AtomicU64, Ordering};
+use std::sync::{Arc, Mutex};
 use std::time::Duration;
 
 use metrics_util::MetricKind;
@@ -43,6 +43,11 @@ const OPERATION_ATTEMPTS: &str = "rustfs_kms_backend_operation_attempts";
 const IN_FLIGHT: &str = "rustfs_kms_backend_in_flight";
 const CIRCUIT_OPEN: &str = "rustfs_kms_backend_circuit_open";
 const MAX_ATTEMPTS: u32 = 10;
+const ATTEMPT_TIMEOUT: Duration = Duration::from_secs(2);
+const HEALTHY_PROGRESS_TIMEOUT: Duration = Duration::from_secs(20);
+// The circuit remains open for 30s after five failed attempts.
+const POST_FAILOVER_PROGRESS_TIMEOUT: Duration = Duration::from_secs(35);
+const FAILOVER_ERROR_POLL_INTERVAL: Duration = Duration::from_millis(100);
 
 type MetricEntry = (
     metrics_util::CompositeKey,
@@ -64,7 +69,7 @@ fn config(backend: KmsBackend, backend_config: BackendConfig) -> KmsConfig {
         backend,
         backend_config,
         allow_insecure_dev_defaults: true,
-        timeout: Duration::from_secs(2),
+        timeout: ATTEMPT_TIMEOUT,
         retry_attempts: MAX_ATTEMPTS,
         enable_cache: false,
         ..KmsConfig::default()
@@ -164,14 +169,31 @@ fn retryable_failures(snapshot: &[MetricEntry], operation: &str) -> u64 {
         .sum()
 }
 
-async fn wait_for_count(counter: &AtomicU64, minimum: u64, description: &str) {
-    tokio::time::timeout(Duration::from_secs(20), async {
+async fn wait_for_count(
+    counter: &AtomicU64,
+    failure: &Mutex<Option<String>>,
+    minimum: u64,
+    description: &str,
+    timeout: Duration,
+) {
+    tokio::time::timeout(timeout, async {
         while counter.load(Ordering::SeqCst) < minimum {
+            if let Some(error) = failure.lock().expect("decrypt failure lock poisoned").as_ref() {
+                panic!(
+                    "{description} worker failed after {} successful decrypts: {error}",
+                    counter.load(Ordering::SeqCst)
+                );
+            }
             tokio::time::sleep(Duration::from_millis(25)).await;
         }
     })
     .await
-    .unwrap_or_else(|_| panic!("timed out waiting for {description}"));
+    .unwrap_or_else(|_| {
+        panic!(
+            "timed out after {timeout:?} waiting for {description}: completed {}, expected {minimum}",
+            counter.load(Ordering::SeqCst)
+        )
+    });
 }
 
 async fn wait_for_file(path: &Path, description: &str) {
@@ -189,7 +211,8 @@ async fn decrypt_loop<B: KmsBackendTrait + Send + Sync + 'static>(
     request: DecryptRequest,
     expected: Vec<u8>,
     completed: Arc<AtomicU64>,
-    failed: Arc<AtomicBool>,
+    allow_failover_errors: Arc<AtomicBool>,
+    failure: Arc<Mutex<Option<String>>>,
     stop: CancellationToken,
 ) {
     while !stop.is_cancelled() {
@@ -197,8 +220,18 @@ async fn decrypt_loop<B: KmsBackendTrait + Send + Sync + 'static>(
             Ok(response) if response.plaintext == expected => {
                 completed.fetch_add(1, Ordering::SeqCst);
             }
-            Ok(_) | Err(_) => {
-                failed.store(true, Ordering::SeqCst);
+            Ok(_) => {
+                *failure.lock().expect("decrypt failure lock poisoned") =
+                    Some("decrypt returned unexpected plaintext".to_string());
+                return;
+            }
+            Err(rustfs_kms::KmsError::BackendError { .. } | rustfs_kms::KmsError::OperationTimedOut { .. })
+                if allow_failover_errors.load(Ordering::SeqCst) =>
+            {
+                tokio::time::sleep(FAILOVER_ERROR_POLL_INTERVAL).await;
+            }
+            Err(error) => {
+                *failure.lock().expect("decrypt failure lock poisoned") = Some(error.to_string());
                 return;
             }
         }
@@ -296,7 +329,9 @@ async fn exercise_failover(snapshotter: &Snapshotter) {
     );
 
     let stop = CancellationToken::new();
-    let failed = Arc::new(AtomicBool::new(false));
+    let allow_failover_errors = Arc::new(AtomicBool::new(false));
+    let kv2_failure = Arc::new(Mutex::new(None));
+    let transit_failure = Arc::new(Mutex::new(None));
     let kv2_completed = Arc::new(AtomicU64::new(0));
     let transit_completed = Arc::new(AtomicU64::new(0));
     let kv2_worker = tokio::spawn(decrypt_loop(
@@ -304,7 +339,8 @@ async fn exercise_failover(snapshotter: &Snapshotter) {
         kv2_request,
         kv2_data_key.plaintext_key,
         Arc::clone(&kv2_completed),
-        Arc::clone(&failed),
+        Arc::clone(&allow_failover_errors),
+        Arc::clone(&kv2_failure),
         stop.clone(),
     ));
     let transit_worker = tokio::spawn(decrypt_loop(
@@ -312,12 +348,21 @@ async fn exercise_failover(snapshotter: &Snapshotter) {
         transit_request,
         transit_data_key.plaintext_key,
         Arc::clone(&transit_completed),
-        Arc::clone(&failed),
+        Arc::clone(&allow_failover_errors),
+        Arc::clone(&transit_failure),
         stop.clone(),
     ));
 
-    wait_for_count(&kv2_completed, 2, "two healthy KV2 decrypts").await;
-    wait_for_count(&transit_completed, 2, "two healthy Transit decrypts").await;
+    wait_for_count(&kv2_completed, &kv2_failure, 2, "two healthy KV2 decrypts", HEALTHY_PROGRESS_TIMEOUT).await;
+    wait_for_count(
+        &transit_completed,
+        &transit_failure,
+        2,
+        "two healthy Transit decrypts",
+        HEALTHY_PROGRESS_TIMEOUT,
+    )
+    .await;
+    allow_failover_errors.store(true, Ordering::SeqCst);
     std::fs::write(&marker, b"ready").expect("publish failover readiness marker");
 
     wait_for_file(&elected, "the replacement Vault leader").await;
@@ -326,18 +371,39 @@ async fn exercise_failover(snapshotter: &Snapshotter) {
 
     let kv2_after_election = kv2_completed.load(Ordering::SeqCst) + 2;
     let transit_after_election = transit_completed.load(Ordering::SeqCst) + 2;
-    wait_for_count(&kv2_completed, kv2_after_election, "post-failover KV2 decrypts").await;
-    wait_for_count(&transit_completed, transit_after_election, "post-failover Transit decrypts").await;
+    wait_for_count(
+        &kv2_completed,
+        &kv2_failure,
+        kv2_after_election,
+        "post-failover KV2 decrypts",
+        POST_FAILOVER_PROGRESS_TIMEOUT,
+    )
+    .await;
+    wait_for_count(
+        &transit_completed,
+        &transit_failure,
+        transit_after_election,
+        "post-failover Transit decrypts",
+        POST_FAILOVER_PROGRESS_TIMEOUT,
+    )
+    .await;
 
     stop.cancel();
     kv2_worker.await.expect("KV2 decrypt worker must join");
     transit_worker.await.expect("Transit decrypt worker must join");
-    assert!(!failed.load(Ordering::SeqCst), "no decrypt may fail or return different plaintext");
+    assert!(
+        kv2_failure.lock().expect("KV2 failure lock poisoned").is_none(),
+        "no KV2 decrypt may fail or return different plaintext"
+    );
+    assert!(
+        transit_failure.lock().expect("Transit failure lock poisoned").is_none(),
+        "no Transit decrypt may fail or return different plaintext"
+    );
 }
 
 #[test]
 #[ignore = "requires a real three-node Vault Raft cluster; run scripts/test/vault_ha_kms_live.sh"]
-fn vault_raft_leader_failure_preserves_kv2_and_transit_decrypts() {
+fn vault_raft_leader_failure_recovers_kv2_and_transit_decrypts() {
     let recorder = DebuggingRecorder::new();
     let snapshotter = recorder.snapshotter();
     metrics::with_local_recorder(&recorder, || {
@@ -349,11 +415,6 @@ fn vault_raft_leader_failure_preserves_kv2_and_transit_decrypts() {
     });
     let snapshot = snapshotter.snapshot().into_vec();
 
-    assert_eq!(
-        counter_value(&snapshot, OPERATIONS_TOTAL, &[("outcome", "circuit_open")]),
-        0,
-        "a bounded leader election must not open the circuit"
-    );
     assert_eq!(
         counter_value(&snapshot, OPERATIONS_TOTAL, &[("outcome", "budget_exhausted")]),
         0,

crates/protos/src/generated/proto_gen/node_service.rs

@@ -722,6 +722,10 @@ pub struct DeleteVersionsResponse {
     pub errors: ::prost::alloc::vec::Vec<::prost::alloc::string::String>,
     #[prost(message, optional, tag = "3")]
     pub error: ::core::option::Option<Error>,
+    /// Senders dual-write the legacy strings and typed entries. Receivers prefer typed entries
+    /// when present and fall back to strings for peers that predate this field. Code zero means success.
+    #[prost(message, repeated, tag = "4")]
+    pub item_errors: ::prost::alloc::vec::Vec<Error>,
 }
 #[derive(Clone, PartialEq, Eq, Hash, ::prost::Message)]
 pub struct ReadMultipleRequest {

crates/protos/src/lib.rs

@@ -2106,6 +2106,9 @@ pub enum ChannelClass {
     Bulk,
 }
 
+// Keep multiplexed unary RPCs below h2's per-connection small-frame budget.
+const INTERNODE_RPC_CONCURRENCY_LIMIT: usize = 64;
+
 /// Whether control/bulk channel isolation is enabled (env-gated, default off for safe rollout).
 fn channel_isolation_enabled() -> bool {
     rustfs_utils::get_env_bool(
@@ -2188,6 +2191,7 @@ async fn build_channel(dial_addr: &str, cache_key: &str) -> Result<Channel, Box<
     let mut connector = Endpoint::from_shared(dial_addr.to_string())?
         // Fast connection timeout for dead peer detection
         .connect_timeout(connect_timeout)
+        .concurrency_limit(INTERNODE_RPC_CONCURRENCY_LIMIT)
         // TCP-level keepalive - OS will probe connection
         .tcp_keepalive(Some(tcp_keepalive))
         // Disable Nagle so latency-sensitive control-plane RPCs (locks/health) are not batched

crates/protos/src/node.proto

@@ -493,6 +493,9 @@ message DeleteVersionsResponse {
   bool success = 1;
   repeated string errors = 2;
   optional Error error = 3;
+  // Senders dual-write the legacy strings and typed entries. Receivers prefer typed entries
+  // when present and fall back to strings for peers that predate this field. Code zero means success.
+  repeated Error item_errors = 4;
 }
 
 message ReadMultipleRequest {

crates/scanner/src/data_usage_define.rs

@@ -125,6 +125,34 @@ 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,
@@ -146,6 +174,11 @@ 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"));
 

crates/scanner/src/data_usage_define/persistence.rs

@@ -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 Self::revision_for_path(store, &backup_path).await {
+            None => match read_config_revision(store, &backup_path).await {
                 Ok(revision) => Some(revision),
                 Err(err) => {
                     counter!(METRIC_CACHE_BACKUP_REVISION_FAILURE_TOTAL).increment(1);
@@ -336,33 +336,6 @@ 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()
     }

crates/scanner/src/lib.rs

@@ -75,7 +75,10 @@ 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::{ScannerCycleScheduleStatus, init_data_scanner, scanner_cycle_schedule_status, scanner_topology_digest};
+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_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,

crates/scanner/src/scanner.rs

@@ -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_with_revision,
+    DataUsageCache, DataUsageCacheRevision, LEGACY_DATA_USAGE_OBJ_NAME_PATH, read_config_revision, read_config_with_revision,
 };
 use crate::runtime_config::{
     ScannerRuntimeConfig, ScannerRuntimeConfigSource, refresh_scanner_runtime_config_from_global, scanner_bitrot_cycle,
@@ -54,9 +54,7 @@ use rustfs_config::{ENV_SCANNER_CYCLE, ENV_SCANNER_SPEED, ENV_SCANNER_START_DELA
 use rustfs_data_usage::observed_data_usage_is_newer;
 use serde::{Deserialize, Serialize};
 use sha2::{Digest as _, Sha256};
-#[cfg(test)]
-use tokio::sync::Notify;
-use tokio::sync::mpsc;
+use tokio::sync::{Notify, mpsc};
 use tokio::time::{Duration, Instant};
 use tokio_util::sync::CancellationToken;
 use tokio_util::task::AbortOnDropHandle;
@@ -104,6 +102,13 @@ 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);
@@ -125,6 +130,12 @@ 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;
 
@@ -576,19 +587,21 @@ 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;
             }
 
-            if let Err(e) = run_data_scanner_with_maintenance_state(
+            let run_result = run_data_scanner_with_maintenance_state(
                 ctx_clone.clone(),
                 storeapi_clone.clone(),
                 startup_features,
                 startup_maintenance_generation,
             )
-            .await
-            {
+            .await;
+            if let Err(e) = &run_result {
                 error!(
                     target: "rustfs::scanner",
                     event = EVENT_SCANNER_CYCLE_STATE,
@@ -599,11 +612,52 @@ 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,
-                _ = tokio::time::sleep(randomized_cycle_delay()) => {}
+                _ = SCANNER_CYCLE_RECOVERY_WAKE.notified() => {},
+                _ = tokio::time::sleep(retry_delay) => {}
             }
         }
     });
@@ -1606,40 +1660,22 @@ async fn run_data_scanner_with_maintenance_state(
         observe_scanner_activity(&storeapi, distributed, &mut scanner_activity_seen).await;
     }
 
-    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 (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 usage_floor = match persisted_usage_floor(storeapi.clone()).await {
         Ok(floor) => floor,
         Err(err) => {
@@ -2219,7 +2255,12 @@ 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(crate) use cycle_state::{current_scanner_leader_epoch, decode_persisted_scanner_cycle_fence};
+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 use heal_info::{BackgroundHealInfo, read_background_heal_info, save_background_heal_info};
 pub use usage_store::store_data_usage_in_backend;
 

crates/scanner/src/scanner/leadership.rs

@@ -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) else {
+        let Some(claimed_epoch) = persisted_epoch.checked_add(1).filter(|epoch| *epoch < u64::MAX) else {
             error!(
                 target: "rustfs::scanner",
                 event = EVENT_SCANNER_PERSIST_STATE,

crates/scanner/src/scanner/tests.rs

@@ -15,11 +15,12 @@
 use super::*;
 use crate::EcstoreResult;
 use crate::{
-    Endpoint, EndpointServerPools, Endpoints, InstanceContext, PoolEndpoints, ScannerGetObjectReader as GetObjectReader,
-    ScannerObjectInfo as ObjectInfo, ScannerObjectOptions as ObjectOptions, ScannerPutObjReader as PutObjReader,
-    init_bucket_metadata_sys_for_scanner_tests, init_ecstore_config_for_scanner_tests, init_local_disks_with_instance_ctx,
+    DATA_USAGE_BLOOM_RECOVERY_PATH, Endpoint, EndpointServerPools, Endpoints, InstanceContext, PoolEndpoints,
+    ScannerGetObjectReader as GetObjectReader, ScannerObjectInfo as ObjectInfo, ScannerObjectOptions as ObjectOptions,
+    ScannerPutObjReader as PutObjReader, init_bucket_metadata_sys_for_scanner_tests, init_ecstore_config_for_scanner_tests,
+    init_local_disks_with_instance_ctx,
 };
-use std::collections::HashMap;
+use std::collections::{HashMap, HashSet};
 use std::io::Cursor;
 use std::task::Poll;
 use temp_env::{with_var, with_var_unset};
@@ -117,6 +118,15 @@ async fn scanner_cycle_lock_fence_bounds_uncooperative_shutdown() {
     assert!(cycle_ctx.is_cancelled());
 }
 
+#[tokio::test]
+async fn scanner_cycle_recovery_wake_survives_wait_registration_race() {
+    notify_scanner_cycle_recovery_wake();
+
+    tokio::time::timeout(Duration::from_secs(1), SCANNER_CYCLE_RECOVERY_WAKE.notified())
+        .await
+        .expect("recovery wake should retain a permit until the waiter registers");
+}
+
 struct ScannerDefaultSpeedGuard;
 
 impl ScannerDefaultSpeedGuard {
@@ -151,6 +161,7 @@ impl Drop for ScannerDefaultCycleGuard {
 struct MemoryConfigStore {
     objects: Mutex<HashMap<String, Vec<u8>>>,
     revisions: Mutex<HashMap<String, u64>>,
+    non_regular_objects: Mutex<HashSet<String>>,
     fail_put_number: Mutex<HashMap<String, usize>>,
     object_not_found_put_number: Mutex<HashMap<String, usize>>,
     error_after_commit_put_number: Mutex<HashMap<String, usize>>,
@@ -191,12 +202,16 @@ impl crate::storage_api::scanner_io::ObjectIO for MemoryConfigStore {
             .get(&key)
             .cloned()
             .ok_or(EcstoreError::FileNotFound)?;
-        let revision = *self.revisions.lock().await.entry(key).or_insert(1);
+        let data_len = i64::try_from(data.len()).expect("memory test object length should fit in i64");
+        let revision = *self.revisions.lock().await.entry(key.clone()).or_insert(1);
+        let is_dir = self.non_regular_objects.lock().await.contains(&key);
 
         Ok(GetObjectReader {
             stream: Box::new(Cursor::new(data)),
             object_info: ObjectInfo {
                 etag: Some(format!("memory-{revision}")),
+                size: data_len,
+                is_dir,
                 ..Default::default()
             },
             buffered_body: None,
@@ -797,6 +812,10 @@ fn scanner_cycle_state_decodes_legacy_and_fenced_formats() {
     let (fenced_cycle, fenced_epoch) = decode_scanner_cycle_state(&fenced).expect("fenced cycle state should decode");
     assert_eq!(fenced_cycle.next, 13);
     assert_eq!(fenced_epoch, 7);
+
+    let mut trailing = fenced;
+    trailing.push(0);
+    assert!(decode_scanner_cycle_state(&trailing).is_err());
 }
 
 #[test]
@@ -823,6 +842,840 @@ fn scanner_startup_fails_closed_on_nonempty_corrupt_cycle_state() {
     assert!(encode_scanner_cycle_state(&exhausted, 7).is_err());
 }
 
+#[tokio::test]
+async fn corrupt_cycle_state_is_quarantined_once() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let state_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_NAME_PATH.as_str());
+    store.objects.lock().await.insert(state_key.clone(), vec![1]);
+    store.revisions.lock().await.insert(state_key.clone(), 7);
+
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store.clone()).await,
+        ScannerCycleStateStartup::Blocked
+    ));
+    let marker_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str());
+    let marker_data = store
+        .objects
+        .lock()
+        .await
+        .get(&marker_key)
+        .cloned()
+        .expect("corrupt state must leave a durable recovery marker");
+    let marker: ScannerCycleRecoveryMarker = serde_json::from_slice(&marker_data).expect("marker should be valid JSON");
+    assert_eq!(marker.primary_revision, "memory-7");
+    assert_eq!(marker.path, DATA_USAGE_BLOOM_NAME_PATH.as_str());
+    assert_eq!(marker.quarantine_path, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str());
+    assert_eq!(marker.classification, "corrupt");
+
+    // A second startup sees the matching marker before consuming the poison body.
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store.clone()).await,
+        ScannerCycleStateStartup::Blocked
+    ));
+
+    // Replacing the primary object advances its revision; the stale marker must
+    // not quarantine the newer, valid state.
+    let cycle = CurrentCycle {
+        next: 9,
+        ..Default::default()
+    };
+    let encoded = encode_scanner_cycle_state(&cycle, 3).expect("valid state should encode");
+    store.objects.lock().await.insert(state_key.clone(), encoded);
+    store.revisions.lock().await.insert(state_key, 8);
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store).await,
+        ScannerCycleStateStartup::Ready {
+            cycle: CurrentCycle { next: 9, .. },
+            leader_epoch: 3,
+            ..
+        }
+    ));
+}
+
+#[tokio::test]
+async fn empty_cycle_state_object_is_quarantined_as_corrupt() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let state_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_NAME_PATH.as_str());
+    store.objects.lock().await.insert(state_key.clone(), Vec::new());
+    store.revisions.lock().await.insert(state_key, 6);
+
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store).await,
+        ScannerCycleStateStartup::Blocked
+    ));
+    assert_eq!(scanner_cycle_recovery_status().classification.as_deref(), Some("corrupt"));
+    assert!(
+        scanner_cycle_recovery_status()
+            .reason
+            .as_deref()
+            .is_some_and(|reason| reason.contains("empty"))
+    );
+}
+
+#[tokio::test]
+async fn future_cycle_state_schema_is_recovery_required() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let state_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_NAME_PATH.as_str());
+    let mut future = 17_u64.to_le_bytes().to_vec();
+    future.extend_from_slice(b"RSCYC999");
+    future.extend_from_slice(&4_u64.to_le_bytes());
+    future.extend_from_slice(&[0x90]);
+    store.objects.lock().await.insert(state_key.clone(), future);
+    store.revisions.lock().await.insert(state_key, 13);
+
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store).await,
+        ScannerCycleStateStartup::Blocked
+    ));
+    assert_eq!(scanner_cycle_recovery_status().classification.as_deref(), Some("future_schema"));
+}
+
+#[tokio::test]
+async fn concurrent_leaders_cannot_quarantine_newer_cycle_state() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let state_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_NAME_PATH.as_str());
+    store.objects.lock().await.insert(state_key.clone(), vec![1]);
+    store.revisions.lock().await.insert(state_key, 4);
+
+    let (first, second) = tokio::join!(
+        load_scanner_cycle_state_for_startup(store.clone()),
+        load_scanner_cycle_state_for_startup(store.clone()),
+    );
+    assert!(matches!(first, ScannerCycleStateStartup::Blocked));
+    assert!(matches!(second, ScannerCycleStateStartup::Blocked));
+
+    let marker_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str());
+    let marker_data = store
+        .objects
+        .lock()
+        .await
+        .get(&marker_key)
+        .cloned()
+        .expect("one contender must publish the recovery marker");
+    let marker: ScannerCycleRecoveryMarker = serde_json::from_slice(&marker_data).expect("marker should decode");
+    assert_eq!(marker.primary_revision, "memory-4");
+}
+
+#[tokio::test]
+async fn cleanup_pending_marker_blocks_a_rewritten_primary_after_restart() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let state_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_NAME_PATH.as_str());
+    let marker_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str());
+    let encoded = encode_scanner_cycle_state(
+        &CurrentCycle {
+            next: 12,
+            ..Default::default()
+        },
+        8,
+    )
+    .expect("valid state should encode");
+    store.objects.lock().await.insert(state_key.clone(), encoded);
+    store.revisions.lock().await.insert(state_key, 22);
+    let marker = ScannerCycleRecoveryMarker {
+        schema_version: 1,
+        primary_revision: "memory-21".to_string(),
+        generation: 11,
+        leader_epoch: 7,
+        classification: "corrupt".to_string(),
+        first_detected_at_unix_secs: 1,
+        last_attempt_at_unix_secs: 2,
+        retry_count: 1,
+        reason: "reset in progress".to_string(),
+        path: DATA_USAGE_BLOOM_NAME_PATH.clone(),
+        quarantine_path: DATA_USAGE_BLOOM_RECOVERY_PATH.clone(),
+        state: "cleanup-pending".to_string(),
+    };
+    store
+        .objects
+        .lock()
+        .await
+        .insert(marker_key.clone(), serde_json::to_vec(&marker).expect("marker should encode"));
+    store.revisions.lock().await.insert(marker_key, 3);
+
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store).await,
+        ScannerCycleStateStartup::Blocked
+    ));
+    assert_eq!(scanner_cycle_recovery_status().state, "cleanup-pending");
+}
+
+#[test]
+fn full_rescan_reset_accepts_unknown_marker_fields_without_trusting_cursor() {
+    let marker = br#"{
+        "schema_version": 99,
+        "primary_revision": "memory-7",
+        "generation": 9000,
+        "leader_epoch": 9000,
+        "classification": "new-future-classification",
+        "first_detected_at_unix_secs": 1,
+        "last_attempt_at_unix_secs": 2,
+        "retry_count": 9,
+        "reason": "future marker",
+        "path": "buckets/.bloomcycle.bin",
+        "quarantine_path": "buckets/.bloomcycle.bin.recovery-required.json",
+        "future_field": {"cursor": "untrusted"}
+    }"#;
+    let decoded =
+        super::cycle_state::decode_recovery_marker_for_reset(marker, &DataUsageCacheRevision::Etag("memory-3".to_string()))
+            .expect("full-rescan compatibility decoder should accept additive fields");
+    assert_eq!(decoded.primary_revision, "memory-7");
+    assert_eq!(decoded.classification, "future_schema");
+    assert_eq!(decoded.generation, 0);
+    assert_eq!(decoded.leader_epoch, 0);
+    assert_eq!(decoded.state, "blocked");
+
+    let malformed =
+        super::cycle_state::decode_recovery_marker_for_reset(b"{not-json", &DataUsageCacheRevision::Etag("memory-4".to_string()))
+            .expect("a full-rescan reset must recover even when the marker is malformed");
+    assert!(malformed.primary_revision.is_empty());
+    assert_eq!(malformed.classification, "future_schema");
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rebuilds_after_malformed_marker_without_trusting_cursor() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), vec![0xff, 0x00, 0x01])
+        .await
+        .expect("corrupt cycle state should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), br#"{not-json"#.to_vec())
+        .await
+        .expect("malformed marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("full-rescan reset should recover malformed marker");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("rebuilt cycle state should remain durable");
+    let (cycle, leader_epoch) = decode_scanner_cycle_state(&state).expect("rebuilt cycle state should decode");
+    assert_eq!(cycle.next, 0, "reset must use the verified usage floor, not marker cursor");
+    assert_eq!(leader_epoch, 1);
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_ignores_epoch_from_malformed_future_primary() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    let mut future_primary = vec![0; 24];
+    future_primary[8..16].copy_from_slice(b"RSCY9999");
+    future_primary[16..24].copy_from_slice(&u64::MAX.to_le_bytes());
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), future_primary)
+        .await
+        .expect("future cycle state should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), br#"{not-json"#.to_vec())
+        .await
+        .expect("malformed marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("full-rescan reset should recover malformed future state");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("rebuilt cycle state should remain durable");
+    let (_, leader_epoch) = decode_scanner_cycle_state(&state).expect("rebuilt cycle state should decode");
+    assert_eq!(leader_epoch, 1, "invalid persisted bytes must not raise the recovery epoch");
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn ecstore_exact_recovery_marker_delete_honors_etag() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), b"marker-v1".to_vec())
+        .await
+        .expect("initial recovery marker should be persisted");
+    let (_, stale_revision) = read_config_with_revision(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str())
+        .await
+        .expect("initial marker revision should load");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), b"marker-v2".to_vec())
+        .await
+        .expect("replacement recovery marker should be persisted");
+
+    let delete_result = store
+        .delete_config_object(
+            RUSTFS_META_BUCKET,
+            DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(),
+            ObjectOptions {
+                http_preconditions: Some(stale_revision.preconditions()),
+                ..Default::default()
+            },
+        )
+        .await;
+    assert!(matches!(delete_result, Err(EcstoreError::PreconditionFailed)));
+    assert_eq!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str())
+            .await
+            .expect("replacement marker should remain durable"),
+        b"marker-v2"
+    );
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rejects_corrupt_primary_under_stale_blocked_marker() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    let corrupt_primary = vec![0xff, 0x00, 0x01];
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), corrupt_primary.clone())
+        .await
+        .expect("corrupt cycle state should be persisted");
+    let (_, primary_revision) = read_config_with_revision(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("primary revision should load");
+    let marker = ScannerCycleRecoveryMarker {
+        schema_version: 1,
+        primary_revision: "memory-stale".to_string(),
+        generation: 1,
+        leader_epoch: 1,
+        classification: "corrupt".to_string(),
+        first_detected_at_unix_secs: 1,
+        last_attempt_at_unix_secs: 2,
+        retry_count: 1,
+        reason: "blocked primary changed".to_string(),
+        path: DATA_USAGE_BLOOM_NAME_PATH.clone(),
+        quarantine_path: DATA_USAGE_BLOOM_RECOVERY_PATH.clone(),
+        state: "blocked".to_string(),
+    };
+    let marker_data = serde_json::to_vec(&marker).expect("blocked marker should encode");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), marker_data.clone())
+        .await
+        .expect("blocked marker should be persisted");
+
+    assert!(
+        reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+            .await
+            .is_err(),
+        "a strict marker must fail closed when its primary revision changed"
+    );
+    assert_eq!(
+        read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+            .await
+            .expect("primary should remain readable"),
+        corrupt_primary
+    );
+    assert_eq!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str())
+            .await
+            .expect("blocked marker should remain durable"),
+        marker_data
+    );
+    assert!(!matches!(primary_revision, DataUsageCacheRevision::Missing));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_preserves_valid_primary_when_marker_is_malformed() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    let primary = CurrentCycle {
+        next: 42,
+        ..Default::default()
+    };
+    let old_primary_data = encode_scanner_cycle_state(&primary, 7).expect("valid cycle state should encode");
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), old_primary_data.clone())
+        .await
+        .expect("valid cycle state should be persisted");
+    let (_, old_primary_revision) = read_config_with_revision(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("primary state revision should load");
+    let old_usage = DataUsageInfo {
+        scanner_epoch: Some(7),
+        scanner_cycle: Some(41),
+        ..Default::default()
+    };
+    let old_usage_data = serde_json::to_vec(&old_usage).expect("usage snapshot should encode");
+    save_config(store.clone(), DATA_USAGE_OBJ_NAME_PATH.as_str(), old_usage_data.clone())
+        .await
+        .expect("usage snapshot should be persisted");
+    let (_, old_usage_revision) = read_config_with_revision(store.clone(), DATA_USAGE_OBJ_NAME_PATH.as_str())
+        .await
+        .expect("usage snapshot revision should load");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), b"{not-json".to_vec())
+        .await
+        .expect("malformed marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("reset should clear a stale malformed marker");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("valid primary should remain durable");
+    let (cycle, leader_epoch) = decode_scanner_cycle_state(&state).expect("primary cycle state should decode");
+    assert_eq!(cycle.next, 42, "reset must not regress an independently fenced primary");
+    assert_eq!(leader_epoch, 8, "reset must advance the preserved primary epoch");
+    let stale_primary_save = save_config_with_preconditions(
+        store.clone(),
+        DATA_USAGE_BLOOM_NAME_PATH.as_str(),
+        old_primary_data,
+        old_primary_revision.preconditions(),
+    )
+    .await;
+    assert!(matches!(stale_primary_save, Err(EcstoreError::PreconditionFailed)));
+    let usage = read_config(store.clone(), DATA_USAGE_OBJ_NAME_PATH.as_str())
+        .await
+        .expect("usage epoch fence should remain durable");
+    assert_eq!(
+        serde_json::from_slice::<DataUsageInfo>(&usage)
+            .expect("fenced usage should decode")
+            .scanner_epoch,
+        Some(8)
+    );
+    let stale_save = save_config_with_preconditions(
+        store.clone(),
+        DATA_USAGE_OBJ_NAME_PATH.as_str(),
+        old_usage_data,
+        old_usage_revision.preconditions(),
+    )
+    .await;
+    assert!(matches!(stale_save, Err(EcstoreError::PreconditionFailed)));
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_resumes_cleanup_pending_preserved_primary() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    let completed_at = Utc::now();
+    let primary = CurrentCycle {
+        current: 3,
+        next: 42,
+        cycle_completed: vec![completed_at],
+        started: completed_at,
+    };
+    save_config(
+        store.clone(),
+        DATA_USAGE_BLOOM_NAME_PATH.as_str(),
+        encode_scanner_cycle_state(&primary, 7).expect("valid cycle state should encode"),
+    )
+    .await
+    .expect("valid cycle state should be persisted");
+    let usage = DataUsageInfo {
+        scanner_epoch: Some(7),
+        scanner_cycle: Some(41),
+        ..Default::default()
+    };
+    save_config(
+        store.clone(),
+        DATA_USAGE_OBJ_NAME_PATH.as_str(),
+        serde_json::to_vec(&usage).expect("usage snapshot should encode"),
+    )
+    .await
+    .expect("usage snapshot should be persisted");
+    let marker = ScannerCycleRecoveryMarker {
+        schema_version: 1,
+        primary_revision: "memory-old".to_string(),
+        generation: 41,
+        leader_epoch: 7,
+        classification: "corrupt".to_string(),
+        first_detected_at_unix_secs: 1,
+        last_attempt_at_unix_secs: 2,
+        retry_count: 1,
+        reason: "reset in progress".to_string(),
+        path: DATA_USAGE_BLOOM_NAME_PATH.clone(),
+        quarantine_path: DATA_USAGE_BLOOM_RECOVERY_PATH.clone(),
+        state: "cleanup-pending".to_string(),
+    };
+    save_config(
+        store.clone(),
+        DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(),
+        serde_json::to_vec(&marker).expect("marker should encode"),
+    )
+    .await
+    .expect("cleanup marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("reset should resume a cleanup-pending preserved primary");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("preserved cycle state should remain durable");
+    let (cycle, leader_epoch) = decode_scanner_cycle_state(&state).expect("cycle state should decode");
+    assert_eq!(cycle.current, 3, "cleanup retry must preserve the in-progress cursor");
+    assert_eq!(cycle.next, 42);
+    assert_eq!(cycle.cycle_completed, vec![completed_at]);
+    assert_eq!(cycle.started, completed_at);
+    assert_eq!(leader_epoch, 8);
+    let usage = read_config(store.clone(), DATA_USAGE_OBJ_NAME_PATH.as_str())
+        .await
+        .expect("usage epoch fence should remain durable");
+    assert_eq!(
+        serde_json::from_slice::<DataUsageInfo>(&usage)
+            .expect("usage should decode")
+            .scanner_epoch,
+        Some(8)
+    );
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rebuilds_oversized_regular_primary_with_malformed_marker() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), vec![0; 1024 * 1024 + 1])
+        .await
+        .expect("oversized cycle state should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), b"{not-json".to_vec())
+        .await
+        .expect("malformed marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("explicit full-rescan reset should replace an oversized regular primary");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("rebuilt cycle state should remain durable");
+    let (cycle, leader_epoch) = decode_scanner_cycle_state(&state).expect("rebuilt cycle state should decode");
+    assert_eq!(cycle.next, 0);
+    assert_eq!(leader_epoch, 1);
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rebuilds_oversized_primary_after_cleanup_marker() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), vec![0; 1024 * 1024 + 1])
+        .await
+        .expect("oversized cycle state should be persisted");
+    let (_, primary_revision) = read_config_with_revision(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("primary revision should load");
+    let marker = ScannerCycleRecoveryMarker {
+        schema_version: 1,
+        primary_revision: match primary_revision {
+            DataUsageCacheRevision::Etag(etag) => etag,
+            DataUsageCacheRevision::Missing => panic!("primary revision should be present"),
+        },
+        generation: 1,
+        leader_epoch: 1,
+        classification: "corrupt".to_string(),
+        first_detected_at_unix_secs: 1,
+        last_attempt_at_unix_secs: 2,
+        retry_count: 1,
+        reason: "reset in progress".to_string(),
+        path: DATA_USAGE_BLOOM_NAME_PATH.clone(),
+        quarantine_path: DATA_USAGE_BLOOM_RECOVERY_PATH.clone(),
+        state: "cleanup-pending".to_string(),
+    };
+    save_config(
+        store.clone(),
+        DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(),
+        serde_json::to_vec(&marker).expect("cleanup marker should encode"),
+    )
+    .await
+    .expect("cleanup marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("cleanup retry should rebuild an oversized primary");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("rebuilt cycle state should remain durable");
+    let (cycle, leader_epoch) = decode_scanner_cycle_state(&state).expect("rebuilt cycle state should decode");
+    assert_eq!(cycle.next, 0);
+    assert_eq!(leader_epoch, 1);
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rebuilds_with_oversized_marker() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), vec![0xff, 0x00, 0x01])
+        .await
+        .expect("corrupt cycle state should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), vec![b'x'; 64 * 1024 + 1])
+        .await
+        .expect("oversized recovery marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("full-rescan reset should recover an oversized marker");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("rebuilt cycle state should remain durable");
+    let (_, leader_epoch) = decode_scanner_cycle_state(&state).expect("rebuilt cycle state should decode");
+    assert_eq!(leader_epoch, 1);
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rebuilds_with_empty_marker() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), vec![0xff, 0x00, 0x01])
+        .await
+        .expect("corrupt cycle state should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), Vec::new())
+        .await
+        .expect("empty recovery marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("full-rescan reset should recover an empty marker");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("rebuilt cycle state should remain durable");
+    let (_, leader_epoch) = decode_scanner_cycle_state(&state).expect("rebuilt cycle state should decode");
+    assert_eq!(leader_epoch, 1);
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_keeps_cleanup_marker_when_preserved_epoch_is_exhausted() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    let primary = CurrentCycle {
+        next: 42,
+        ..Default::default()
+    };
+    save_config(
+        store.clone(),
+        DATA_USAGE_BLOOM_NAME_PATH.as_str(),
+        encode_scanner_cycle_state(&primary, u64::MAX).expect("valid cycle state should encode"),
+    )
+    .await
+    .expect("valid cycle state should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), b"{not-json".to_vec())
+        .await
+        .expect("malformed marker should be persisted");
+
+    assert!(
+        reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+            .await
+            .is_err()
+    );
+
+    let marker = read_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str())
+        .await
+        .expect("cleanup marker should remain durable");
+    assert_eq!(
+        serde_json::from_slice::<ScannerCycleRecoveryMarker>(&marker)
+            .expect("cleanup marker should decode")
+            .state,
+        "cleanup-pending"
+    );
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store).await,
+        ScannerCycleStateStartup::Blocked
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rejects_preserved_epoch_that_would_be_terminal() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    let primary = CurrentCycle {
+        next: 42,
+        ..Default::default()
+    };
+    save_config(
+        store.clone(),
+        DATA_USAGE_BLOOM_NAME_PATH.as_str(),
+        encode_scanner_cycle_state(&primary, u64::MAX - 1).expect("valid cycle state should encode"),
+    )
+    .await
+    .expect("valid cycle state should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), b"{not-json".to_vec())
+        .await
+        .expect("malformed marker should be persisted");
+
+    assert!(
+        reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+            .await
+            .is_err(),
+        "reset must not persist the terminal leader epoch"
+    );
+
+    let marker = read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str())
+        .await
+        .expect("cleanup marker should remain durable");
+    assert_eq!(
+        serde_json::from_slice::<ScannerCycleRecoveryMarker>(&marker)
+            .expect("cleanup marker should decode")
+            .state,
+        "cleanup-pending"
+    );
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rejects_usage_floor_that_would_be_terminal() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), vec![0xff, 0x00, 0x01])
+        .await
+        .expect("corrupt cycle state should be persisted");
+    save_config(
+        store.clone(),
+        DATA_USAGE_OBJ_NAME_PATH.as_str(),
+        serde_json::to_vec(&DataUsageInfo {
+            scanner_epoch: Some(u64::MAX - 1),
+            ..Default::default()
+        })
+        .expect("usage floor should encode"),
+    )
+    .await
+    .expect("usage floor should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), b"{not-json".to_vec())
+        .await
+        .expect("malformed marker should be persisted");
+
+    assert!(
+        reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+            .await
+            .is_err(),
+        "reset must not persist the terminal leader epoch"
+    );
+    assert_eq!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str())
+            .await
+            .expect("recovery marker should remain durable"),
+        b"{not-json"
+    );
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rebuilds_empty_primary_with_malformed_marker() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    save_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str(), Vec::new())
+        .await
+        .expect("empty cycle state should be persisted");
+    save_config(store.clone(), DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(), b"{not-json".to_vec())
+        .await
+        .expect("malformed marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("explicit full-rescan reset should replace an empty primary");
+
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("rebuilt cycle state should remain durable");
+    let (cycle, leader_epoch) = decode_scanner_cycle_state(&state).expect("rebuilt cycle state should decode");
+    assert_eq!(cycle.next, 0);
+    assert_eq!(leader_epoch, 1);
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn full_rescan_reset_rebuilds_when_primary_cycle_state_is_missing() {
+    let (_temp_dir, store) = setup_scanner_cycle_store().await;
+    let marker = ScannerCycleRecoveryMarker {
+        schema_version: 1,
+        primary_revision: "memory-missing".to_string(),
+        generation: u64::MAX,
+        leader_epoch: u64::MAX,
+        classification: "corrupt".to_string(),
+        first_detected_at_unix_secs: 1,
+        last_attempt_at_unix_secs: 2,
+        retry_count: 0,
+        reason: "missing primary".to_string(),
+        path: DATA_USAGE_BLOOM_NAME_PATH.clone(),
+        quarantine_path: DATA_USAGE_BLOOM_RECOVERY_PATH.clone(),
+        state: "blocked".to_string(),
+    };
+    save_config(
+        store.clone(),
+        DATA_USAGE_BLOOM_RECOVERY_PATH.as_str(),
+        serde_json::to_vec(&marker).expect("marker should encode"),
+    )
+    .await
+    .expect("marker should be persisted");
+
+    reset_scanner_cycle_recovery(CancellationToken::new(), store.clone())
+        .await
+        .expect("full-rescan reset should recreate missing primary");
+    let state = read_config(store.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str())
+        .await
+        .expect("missing primary should be rebuilt");
+    let (cycle, leader_epoch) = decode_scanner_cycle_state(&state).expect("rebuilt cycle state should decode");
+    assert_eq!(cycle.next, 0);
+    assert_eq!(leader_epoch, 1);
+    assert!(matches!(
+        read_config(store, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()).await,
+        Err(EcstoreError::ConfigNotFound)
+    ));
+}
+
+#[tokio::test]
+async fn corrupt_cycle_state_rename_or_marker_failure_stays_recovery_required() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let state_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_NAME_PATH.as_str());
+    let marker_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str());
+    store.objects.lock().await.insert(state_key.clone(), vec![1]);
+    store.revisions.lock().await.insert(state_key, 9);
+    store.fail_put_number.lock().await.insert(marker_key, 1);
+
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store.clone()).await,
+        ScannerCycleStateStartup::Transient(_)
+    ));
+    let status = scanner_cycle_recovery_status();
+    assert_eq!(status.state, "recovery-required");
+    assert!(status.retryable);
+    assert!(
+        store
+            .objects
+            .lock()
+            .await
+            .contains_key(&memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_NAME_PATH.as_str()))
+    );
+}
+
+#[tokio::test]
+async fn oversized_or_symlinked_cycle_state_is_rejected() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_NAME_PATH.as_str());
+    store.objects.lock().await.insert(key.clone(), vec![0; 1024 * 1024 + 1]);
+    store.revisions.lock().await.insert(key.clone(), 11);
+
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store.clone()).await,
+        ScannerCycleStateStartup::Blocked
+    ));
+    assert_eq!(scanner_cycle_recovery_status().classification.as_deref(), Some("corrupt"));
+    assert!(
+        scanner_cycle_recovery_status()
+            .reason
+            .as_deref()
+            .is_some_and(|reason| reason.contains("oversized"))
+    );
+
+    let marker_key = memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_BLOOM_RECOVERY_PATH.as_str());
+    store.objects.lock().await.remove(&marker_key);
+    store.objects.lock().await.insert(key.clone(), vec![1]);
+    store.revisions.lock().await.insert(key.clone(), 12);
+    store.non_regular_objects.lock().await.insert(key);
+    // The object contract exposes a non-regular object as `is_dir`; local
+    // backends reject symlink/reparse entries before they become an object.
+    assert!(matches!(
+        load_scanner_cycle_state_for_startup(store).await,
+        ScannerCycleStateStartup::Blocked
+    ));
+}
+
 #[tokio::test]
 async fn scanner_startup_uses_primary_and_backup_usage_floor() {
     let store = Arc::new(MemoryConfigStore::default());
@@ -855,6 +1708,31 @@ async fn scanner_startup_uses_primary_and_backup_usage_floor() {
     assert_eq!(epoch, 11);
 }
 
+#[tokio::test]
+async fn scanner_usage_floor_ignores_older_backup_after_primary_epoch_fence() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let backup_path = format!("{}.bkp", DATA_USAGE_OBJ_NAME_PATH.as_str());
+    for (path, epoch, cycle) in [(DATA_USAGE_OBJ_NAME_PATH.as_str(), 8, 100), (backup_path.as_str(), 7, 10_000)] {
+        store.objects.lock().await.insert(
+            memory_config_key(RUSTFS_META_BUCKET, path),
+            serde_json::to_vec(&DataUsageInfo {
+                scanner_epoch: Some(epoch),
+                scanner_cycle: Some(cycle),
+                ..Default::default()
+            })
+            .expect("usage snapshot should encode"),
+        );
+    }
+
+    assert_eq!(
+        persisted_usage_floor(store).await.expect("usage floor should load"),
+        PersistedUsageFloor {
+            next_cycle: 101,
+            leader_epoch: 8,
+        }
+    );
+}
+
 #[test]
 fn scanner_startup_treats_incomplete_usage_snapshot_as_cold() {
     let mut legacy = complete_usage_with_bucket_count(Some(std::time::SystemTime::now()), 1);
@@ -987,6 +1865,15 @@ async fn scanner_usage_floor_fails_closed_on_corrupt_or_exhausted_usage_state()
 
     assert!(persisted_usage_floor(store.clone()).await.is_err());
 
+    store.objects.lock().await.insert(
+        memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_OBJ_NAME_PATH.as_str()),
+        br#"{}"#.to_vec(),
+    );
+    assert!(
+        persisted_usage_floor(store.clone()).await.is_err(),
+        "a structurally incomplete usage snapshot must not be treated as an empty floor"
+    );
+
     store.objects.lock().await.insert(
         memory_config_key(RUSTFS_META_BUCKET, DATA_USAGE_OBJ_NAME_PATH.as_str()),
         serde_json::to_vec(&DataUsageInfo {
@@ -1244,6 +2131,22 @@ async fn test_leadership_claim_preserves_usage_epoch_floor_across_old_epoch_conf
     assert_eq!(store.put_counts.lock().await.get(&key), Some(&3));
 }
 
+#[tokio::test]
+async fn test_leadership_claim_rejects_terminal_epoch() {
+    let store = Arc::new(MemoryConfigStore::default());
+    let ctx = CancellationToken::new();
+    let mut revision = DataUsageCacheRevision::Missing;
+    let mut cycle = CurrentCycle {
+        next: 12,
+        ..Default::default()
+    };
+    let mut persisted_epoch = u64::MAX - 1;
+
+    assert!(!claim_scanner_leadership(&ctx, store.clone(), &mut cycle, &mut revision, &mut persisted_epoch).await);
+    assert_eq!(persisted_epoch, u64::MAX - 1);
+    assert!(read_config(store, &DATA_USAGE_BLOOM_NAME_PATH).await.is_err());
+}
+
 #[tokio::test]
 async fn test_leadership_claim_confirms_commit_after_returned_error() {
     let store = Arc::new(MemoryConfigStore::default());
@@ -2975,6 +3878,24 @@ fn superseded_retry_backoff_grows_from_the_default_cycle() {
     }
 }
 
+#[tokio::test(start_paused = true)]
+async fn corrupt_cycle_state_backoff_uses_virtual_clock() {
+    let mut backoff = ScannerRetryBackoff::default();
+    backoff.record_retryable_cycle(true);
+    let first_delay = backoff
+        .retry_interval(Duration::from_secs(60))
+        .expect("the first recovery retry should be scheduled");
+    assert_eq!(first_delay, Duration::from_secs(5));
+
+    let deadline = Instant::now() + first_delay;
+    assert!(Instant::now() < deadline);
+    tokio::time::advance(first_delay).await;
+    assert!(Instant::now() >= deadline);
+
+    backoff.record_retryable_cycle(true);
+    assert_eq!(backoff.retry_interval(Duration::from_secs(60)), Some(Duration::from_secs(10)));
+}
+
 #[test]
 fn scanner_cycle_wait_plan_drives_growth_resets_and_bitrot_cap() {
     let runtime_config = ScannerRuntimeConfig {

rustfs/src/admin/handlers/mod.rs

@@ -126,6 +126,7 @@ mod tests {
         let _list_remote_target_handler = replication::ListRemoteTargetHandler {};
         let _remove_remote_target_handler = replication::RemoveRemoteTargetHandler {};
         let _scanner_status_handler = scanner::ScannerStatusHandler {};
+        let _scanner_cycle_state_reset_handler = scanner::ScannerCycleStateResetHandler {};
         let _ilm_expiry_status_handler = scanner::IlmExpiryStatusHandler {};
         let _manual_transition_handler = ilm_transition::ManualTransitionRunHandler {};
         let _manual_transition_status_handler = ilm_transition::ManualTransitionJobStatusHandler {};

rustfs/src/admin/handlers/scanner.rs

@@ -13,8 +13,11 @@
 // limitations under the License.
 
 use crate::admin::auth::authorize_admin_request;
+use crate::admin::handlers::supervise_admin_mutation;
 use crate::admin::router::{AdminOperation, Operation, S3Router};
-use crate::admin::runtime_sources::current_scanner_metrics_report;
+use crate::admin::runtime_sources::{
+    app_context_from_req, current_object_store_handle_for_context, current_scanner_metrics_report,
+};
 use crate::module_switches::{ENV_SCANNER_ENABLED, scanner_enabled_from_env};
 use crate::server::ADMIN_PREFIX;
 use chrono::Utc;
@@ -22,11 +25,13 @@ use http::{HeaderMap, HeaderValue};
 use hyper::{Method, StatusCode};
 use matchit::Params;
 use rustfs_common::metrics::{ScannerLifecycleExpirySnapshot, ScannerMaintenanceControlSnapshot, ScannerMetricsReport};
+use rustfs_config::MAX_ADMIN_REQUEST_BODY_SIZE;
 use rustfs_credentials::Credentials;
 use rustfs_policy::policy::action::{Action, AdminAction};
 use s3s::header::CONTENT_TYPE;
 use s3s::{Body, S3Error, S3ErrorCode, S3Request, S3Response, S3Result, s3_error};
-use serde::Serialize;
+use serde::{Deserialize, Serialize};
+use tokio_util::sync::CancellationToken;
 
 const JSON_CONTENT_TYPE: &str = "application/json";
 
@@ -38,6 +43,13 @@ struct ScannerStatusResponse {
     metrics: ScannerMetricsReport,
     cycle_schedule: rustfs_scanner::ScannerCycleScheduleStatus,
     runtime_config: rustfs_scanner::runtime_config::ScannerRuntimeConfigStatus,
+    cycle_recovery: rustfs_scanner::ScannerCycleRecoveryStatus,
+}
+
+#[derive(Debug, Deserialize)]
+#[serde(deny_unknown_fields)]
+struct ScannerCycleResetRequest {
+    mode: String,
 }
 
 #[derive(Debug, Serialize)]
@@ -117,6 +129,7 @@ fn scanner_status_response(
         metrics,
         cycle_schedule,
         runtime_config,
+        cycle_recovery: rustfs_scanner::scanner::scanner_cycle_recovery_status(),
     }
 }
 
@@ -144,6 +157,11 @@ pub fn register_scanner_route(r: &mut S3Router<AdminOperation>) -> std::io::Resu
         format!("{ADMIN_PREFIX}/v3/scanner/status").as_str(),
         AdminOperation(&ScannerStatusHandler {}),
     )?;
+    r.insert(
+        Method::POST,
+        format!("{ADMIN_PREFIX}/v3/scanner/cycle-state/reset").as_str(),
+        AdminOperation(&ScannerCycleStateResetHandler {}),
+    )?;
     r.insert(
         Method::GET,
         format!("{ADMIN_PREFIX}/v3/ilm/expiry/status").as_str(),
@@ -163,6 +181,13 @@ async fn validate_scanner_status_request(req: &S3Request<Body>) -> S3Result<Cred
     authorize_admin_request(req, vec![Action::AdminAction(AdminAction::ServerInfoAdminAction)]).await
 }
 
+async fn validate_scanner_reset_request(req: &S3Request<Body>) -> S3Result<Credentials> {
+    if req.credentials.is_none() {
+        return Err(s3_error!(InvalidRequest, "missing credentials"));
+    }
+    authorize_admin_request(req, vec![Action::AdminAction(AdminAction::ConfigUpdateAdminAction)]).await
+}
+
 fn json_response(body: Vec<u8>) -> S3Result<S3Response<(StatusCode, Body)>> {
     let mut headers = HeaderMap::new();
     let content_type = HeaderValue::from_str(JSON_CONTENT_TYPE)
@@ -192,6 +217,37 @@ impl Operation for ScannerStatusHandler {
 
 pub struct IlmExpiryStatusHandler {}
 
+pub struct ScannerCycleStateResetHandler {}
+
+#[async_trait::async_trait]
+impl Operation for ScannerCycleStateResetHandler {
+    async fn call(&self, mut req: S3Request<Body>, _params: Params<'_, '_>) -> S3Result<S3Response<(StatusCode, Body)>> {
+        let _cred = validate_scanner_reset_request(&req).await?;
+        let body = req
+            .input
+            .store_all_limited(MAX_ADMIN_REQUEST_BODY_SIZE)
+            .await
+            .map_err(|err| S3Error::with_message(S3ErrorCode::InvalidRequest, format!("invalid reset request body: {err}")))?;
+        let reset = serde_json::from_slice::<ScannerCycleResetRequest>(&body)
+            .map_err(|err| S3Error::with_message(S3ErrorCode::InvalidRequest, format!("invalid reset request body: {err}")))?;
+        if reset.mode != "full-rescan" {
+            return Err(S3Error::with_message(S3ErrorCode::InvalidRequest, "reset mode must be full-rescan"));
+        }
+        let context = app_context_from_req(&req)
+            .ok_or_else(|| S3Error::with_message(S3ErrorCode::InternalError, "storage layer not initialized"))?;
+        let store = current_object_store_handle_for_context(Some(context.as_ref()))
+            .ok_or_else(|| S3Error::with_message(S3ErrorCode::InternalError, "storage layer not initialized"))?;
+        supervise_admin_mutation("scanner cycle state reset", async move {
+            rustfs_scanner::scanner::reset_scanner_cycle_recovery(CancellationToken::new(), store)
+                .await
+                .map_err(|err| S3Error::with_message(S3ErrorCode::InternalError, err.to_string()))?;
+            Ok::<_, S3Error>(())
+        })
+        .await?;
+        json_response(br#"{"status":"reset","mode":"full-rescan"}"#.to_vec())
+    }
+}
+
 #[async_trait::async_trait]
 impl Operation for IlmExpiryStatusHandler {
     async fn call(&self, req: S3Request<Body>, _params: Params<'_, '_>) -> S3Result<S3Response<(StatusCode, Body)>> {
@@ -237,6 +293,38 @@ mod tests {
         assert_eq!(err.message(), Some("missing credentials"));
     }
 
+    #[tokio::test]
+    async fn scanner_reset_gate_rejects_missing_credentials() {
+        let req = S3Request {
+            input: Body::from(String::new()),
+            method: Method::POST,
+            uri: http::Uri::from_static("/rustfs/admin/v3/scanner/cycle-state/reset"),
+            headers: HeaderMap::new(),
+            extensions: http::Extensions::new(),
+            credentials: None,
+            region: None,
+            service: None,
+            trailing_headers: None,
+        };
+
+        let err = validate_scanner_reset_request(&req)
+            .await
+            .expect_err("a reset request without credentials must be rejected");
+        assert_eq!(err.code(), &S3ErrorCode::InvalidRequest);
+        assert_eq!(err.message(), Some("missing credentials"));
+    }
+
+    #[test]
+    fn admin_reset_requires_full_rescan_or_verified_cursor() {
+        let full_rescan: ScannerCycleResetRequest =
+            serde_json::from_str(r#"{"mode":"full-rescan"}"#).expect("full rescan must be accepted");
+        assert_eq!(full_rescan.mode, "full-rescan");
+        let cursor: ScannerCycleResetRequest =
+            serde_json::from_str(r#"{"mode":"cursor"}"#).expect("mode validation belongs to the handler");
+        assert_ne!(cursor.mode, "full-rescan");
+        assert!(serde_json::from_str::<ScannerCycleResetRequest>(r#"{"mode":"full-rescan","cursor":"untrusted"}"#).is_err());
+    }
+
     #[test]
     fn scanner_disabled_reason_reports_startup_env_key() {
         assert_eq!(scanner_disabled_reason(true), None);
@@ -304,6 +392,11 @@ mod tests {
         assert_eq!(encoded["cycle_schedule"]["effective_interval_seconds"], 0);
         assert_eq!(encoded["cycle_schedule"]["clean_idle_backoff_enabled"], false);
         assert_eq!(encoded["cycle_schedule"]["clean_idle_backoff_multiplier"], 1);
+        assert_eq!(encoded["cycle_recovery"]["state"], "healthy");
+        assert_eq!(
+            encoded["cycle_recovery"]["quarantine_path"],
+            rustfs_scanner::DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()
+        );
     }
 
     #[test]

rustfs/src/admin/route_policy.rs

@@ -428,6 +428,12 @@ pub const ADMIN_ROUTE_POLICY_SPECS: &[AdminRouteSpec] = &[
     admin(HttpMethod::Get, "/rustfs/admin/v3/config", CONFIG_UPDATE, RouteRiskLevel::High),
     admin(HttpMethod::Put, "/rustfs/admin/v3/config", CONFIG_UPDATE, RouteRiskLevel::High),
     admin(HttpMethod::Get, "/rustfs/admin/v3/scanner/status", SERVER_INFO, RouteRiskLevel::Sensitive),
+    admin(
+        HttpMethod::Post,
+        "/rustfs/admin/v3/scanner/cycle-state/reset",
+        CONFIG_UPDATE,
+        RouteRiskLevel::High,
+    ),
     admin(
         HttpMethod::Get,
         "/rustfs/admin/v3/ilm/expiry/status",
@@ -2020,6 +2026,12 @@ mod tests {
         assert_not_action(HttpMethod::Get, "/rustfs/admin/v3/ilm/expiry/status", SET_TIER);
     }
 
+    #[test]
+    fn route_policy_requires_config_update_for_scanner_cycle_reset() {
+        assert_action(HttpMethod::Post, "/rustfs/admin/v3/scanner/cycle-state/reset", CONFIG_UPDATE);
+        assert_not_action(HttpMethod::Post, "/rustfs/admin/v3/scanner/cycle-state/reset", SERVER_INFO);
+    }
+
     #[test]
     fn route_policy_uses_tier_actions_for_transition_routes() {
         assert_action(HttpMethod::Post, "/rustfs/admin/v3/ilm/transition/run", SET_TIER);

rustfs/src/admin/route_registration_test.rs

@@ -243,6 +243,7 @@ fn expected_admin_route_matrix() -> Vec<RouteMatrixEntry> {
         admin_route(Method::GET, "/v3/config"),
         admin_route(Method::PUT, "/v3/config"),
         admin_route(Method::GET, "/v3/scanner/status"),
+        admin_route(Method::POST, "/v3/scanner/cycle-state/reset"),
         admin_route(Method::GET, "/v3/audit/target/list"),
         admin_route_sample(
             Method::PUT,
@@ -879,6 +880,7 @@ fn test_register_routes_cover_representative_admin_paths() {
     assert_route(&router, Method::GET, &admin_path("/v3/config"));
     assert_route(&router, Method::PUT, &admin_path("/v3/config"));
     assert_route(&router, Method::GET, &admin_path("/v3/scanner/status"));
+    assert_route(&router, Method::POST, &admin_path("/v3/scanner/cycle-state/reset"));
     assert_route(&router, Method::GET, &admin_path("/v3/ilm/expiry/status"));
     assert_route(&router, Method::POST, &admin_path("/v3/ilm/transition/run"));
     assert_route(
@@ -1367,6 +1369,7 @@ fn test_admin_alias_paths_match_existing_admin_routes() {
         (Method::GET, compat_admin_alias_path("/v3/config")),
         (Method::PUT, compat_admin_alias_path("/v3/config")),
         (Method::GET, compat_admin_alias_path("/v3/scanner/status")),
+        (Method::POST, compat_admin_alias_path("/v3/scanner/cycle-state/reset")),
         (Method::GET, compat_admin_alias_path("/v3/ilm/expiry/status")),
     ] {
         assert!(

rustfs/src/storage/rpc/node_service/disk.rs

@@ -146,6 +146,29 @@ fn encode_file_info_msgpack(value: &FileInfo) -> std::result::Result<Vec<u8>, Di
     encode_msgpack_with_capacity(value, "FileInfo", FILE_INFO_MSGPACK_ENCODE_CAPACITY_HINT)
 }
 
+fn encode_delete_versions_errors(disk_errors: Vec<Option<DiskError>>) -> (Vec<String>, Vec<Error>) {
+    let mut errors = Vec::with_capacity(disk_errors.len());
+    let mut item_errors = Vec::with_capacity(disk_errors.len());
+    for error in disk_errors {
+        match error {
+            Some(error) => {
+                let code = match &error {
+                    DiskError::Io(source) if source.kind() == std::io::ErrorKind::NotFound => DiskError::FileNotFound.to_u32(),
+                    _ => error.to_u32(),
+                };
+                let error_info = error.to_string();
+                errors.push(error_info.clone());
+                item_errors.push(Error { code, error_info });
+            }
+            None => {
+                errors.push(String::new());
+                item_errors.push(Error::default());
+            }
+        }
+    }
+    (errors, item_errors)
+}
+
 fn encode_msgpack_named<T: serde::Serialize>(value: &T, value_name: &str) -> std::result::Result<Vec<u8>, DiskError> {
     let mut serializer = rmp_serde::Serializer::new(Vec::with_capacity(MSGPACK_ENCODE_CAPACITY_HINT)).with_struct_map();
     value
@@ -552,6 +575,7 @@ impl NodeService {
                             success: false,
                             errors: Vec::new(),
                             error: Some(DiskError::other(format!("decode FileInfoVersions failed: {err}")).into()),
+                            item_errors: Vec::new(),
                         }));
                     }
                 };
@@ -563,30 +587,26 @@ impl NodeService {
                         success: false,
                         errors: Vec::new(),
                         error: Some(DiskError::other(format!("decode DeleteOptions failed: {err}")).into()),
+                        item_errors: Vec::new(),
                     }));
                 }
             };
 
-            let errors = disk
-                .delete_versions(&request.volume, versions, opts)
-                .await
-                .into_iter()
-                .map(|error| match error {
-                    Some(e) => e.to_string(),
-                    None => "".to_string(),
-                })
-                .collect();
+            let (errors, item_errors) =
+                encode_delete_versions_errors(disk.delete_versions(&request.volume, versions, opts).await);
 
             Ok(Response::new(DeleteVersionsResponse {
                 success: true,
                 errors,
                 error: None,
+                item_errors,
             }))
         } else {
             Ok(Response::new(DeleteVersionsResponse {
                 success: false,
                 errors: Vec::new(),
                 error: Some(DiskError::other("cannot find disk".to_string()).into()),
+                item_errors: Vec::new(),
             }))
         }
     }
@@ -1612,8 +1632,8 @@ impl NodeService {
 mod tests {
     use super::{
         compat_response_json, decode_msgpack_or_json, decode_rename_data_request_file_info,
-        encode_batch_read_version_response_payloads, encode_file_info_msgpack, encode_msgpack, encode_msgpack_named,
-        encode_read_multiple_response_payloads, encode_rename_data_response_payloads,
+        encode_batch_read_version_response_payloads, encode_delete_versions_errors, encode_file_info_msgpack, encode_msgpack,
+        encode_msgpack_named, encode_read_multiple_response_payloads, encode_rename_data_response_payloads,
     };
     use crate::storage::rpc::node_service::make_server;
     use crate::storage::storage_api::ReadMultipleResp;
@@ -1632,6 +1652,18 @@ mod tests {
         count: u32,
     }
 
+    #[test]
+    fn delete_versions_response_dual_writes_typed_item_errors() {
+        let raw_not_found = super::DiskError::Io(std::io::Error::from(std::io::ErrorKind::NotFound));
+        let (errors, item_errors) = encode_delete_versions_errors(vec![Some(raw_not_found), None]);
+
+        assert!(errors[0].starts_with("io error "));
+        assert!(errors[1].is_empty());
+        assert_eq!(item_errors[0].code, super::DiskError::FileNotFound.to_u32());
+        assert_eq!(item_errors[0].error_info, errors[0]);
+        assert_eq!(item_errors[1].code, 0);
+    }
+
     #[tokio::test]
     #[serial]
     async fn handle_read_version_records_attribution_for_missing_disk() {

scripts/test/vault_ha_kms_live.sh

@@ -241,7 +241,7 @@ env \
   RUSTFS_TEST_VAULT_FAILOVER_MARKER="$MARKER" \
   RUSTFS_TEST_VAULT_OLD_LEADER="$OLD_LEADER" \
   cargo test -p rustfs-kms --test vault_ha_failover_live \
-    vault_raft_leader_failure_preserves_kv2_and_transit_decrypts -- \
+    vault_raft_leader_failure_recovers_kv2_and_transit_decrypts -- \
     --ignored --nocapture --test-threads=1 &
 TEST_PID=$!