fix(ecstore): satisfy delete fence lint checks

* fix(ecstore): fail closed on unresolved decommission entries

* perf(ecstore): avoid successful listing name clone

.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.

crates/config/src/constants/runtime.rs

@@ -57,6 +57,13 @@ pub const DEFAULT_MAX_IO_EVENTS_PER_TICK: usize = 1024;
 pub const DEFAULT_EVENT_INTERVAL: u32 = 61;
 pub const DEFAULT_RNG_SEED: Option<u64> = None; // None means random
 
+/// Dedicated blocking thread pool for fsync/fdatasync operations.
+/// When > 1, fsync operations are isolated from the main blocking pool to
+/// prevent device-bound fsync from starving read operations (pread/stat/open).
+/// Default 0 means auto (no isolation, use main runtime).
+pub const ENV_FSYNC_BLOCKING_THREADS: &str = "RUSTFS_RUNTIME_FSYNC_BLOCKING_THREADS";
+pub const DEFAULT_FSYNC_BLOCKING_THREADS: usize = 0;
+
 // Dial9 Tokio Telemetry Default values
 pub const DEFAULT_RUNTIME_DIAL9_ENABLED: bool = false; // Disabled by default
 pub const DEFAULT_RUNTIME_DIAL9_OUTPUT_DIR: &str = "/var/log/rustfs/telemetry";

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(),
@@ -3275,6 +3416,9 @@ impl ECStore {
             )
             .await?;
 
+            let source_cleanup_mutation_fence = self
+                .acquire_decommission_source_cleanup_fence(bucket.as_str(), entry.name.as_str(), set.as_ref())
+                .await?;
             let cleanup_result = data_movement::cleanup_source_entry_if_unchanged(
                 set.clone(),
                 bucket.as_str(),
@@ -3286,6 +3430,7 @@ impl ECStore {
                     lifecycle_guard: bucket_incarnation_fence
                         .as_ref()
                         .and_then(|guard| guard.namespace_lock_guard()),
+                    object_mutation_fence: Some(&source_cleanup_mutation_fence),
                 },
                 "decommission",
             )
@@ -3334,34 +3479,42 @@ 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"
+                        );
+                    }
                 }
             }
         }
@@ -3379,6 +3532,22 @@ impl ECStore {
         Ok(())
     }
 
+    #[cfg(test)]
+    pub(crate) async fn decommission_entry_for_test(
+        self: &Arc<Self>,
+        idx: usize,
+        entry: MetaCacheEntry,
+        bucket: String,
+        set: Arc<SetDisks>,
+    ) -> Result<()> {
+        let worker_permit = Arc::new(Semaphore::new(1))
+            .acquire_owned()
+            .await
+            .map_err(|err| Error::other(format!("decommission test worker permit acquire failed: {err}")))?;
+        self.decommission_entry(CancellationToken::new(), idx, entry, bucket, set, worker_permit, None, None, None, None)
+            .await
+    }
+
     #[tracing::instrument(skip(self, rx))]
     async fn decommission_pool(
         self: &Arc<Self>,
@@ -3538,6 +3707,7 @@ 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 +3721,11 @@ 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 +3755,7 @@ 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 +4361,7 @@ 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 +4456,7 @@ 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)?;
@@ -4314,15 +4484,20 @@ impl ECStore {
     ) -> Result<()> {
         warn!("decommission_object: start {} {}", &bucket, &rd.object_info.name);
         let object_name = rd.object_info.name.clone();
-        let result = data_movement::migrate_object(
+        let mut migration = tokio::task::JoinSet::new();
+        migration.spawn(data_movement::migrate_decommission_object(
             self,
             pool_idx,
             bucket.clone(),
             rd,
             expected_bucket_incarnation_id,
             "decommission_object",
-        )
-        .await;
+        ));
+        let result = migration
+            .join_next()
+            .await
+            .ok_or_else(|| Error::other("decommission migration task was not started"))?
+            .map_err(|err| Error::other(format!("decommission migration task join error: {err}")))?;
         if result.is_ok() {
             warn!("decommission_object: migrated {} {}", &bucket, &object_name);
         }
@@ -5021,12 +5196,15 @@ 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 +5219,12 @@ 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 +5237,51 @@ 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 +5289,10 @@ impl SetDisks {
         )
         .await?;
 
+        if let Some(err) = entry_error.lock().await.clone() {
+            return Err(err);
+        }
+
         Ok(())
     }
 }
@@ -5264,11 +5483,11 @@ 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 +5498,12 @@ 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 +5513,17 @@ 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 +6542,65 @@ 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 +6818,7 @@ 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 +6833,13 @@ 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 +6914,134 @@ 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/ecstore/src/data_movement/mod.rs

@@ -26,7 +26,7 @@ use crate::storage_api_contracts::{
     namespace::NamespaceLocking as _,
     object::{HTTPPreconditions, ObjectOperations as _},
 };
-use crate::store::ECStore;
+use crate::store::{ECStore, ObjectLockDiagGuard, SourceCleanupMutationFence};
 use bytes::Bytes;
 use rustfs_filemeta::{FileInfo, FileInfoVersions, ObjectPartInfo};
 use rustfs_rio::{EtagResolvable, HashReader, HashReaderDetector, Index, TryGetIndex};
@@ -856,7 +856,6 @@ fn is_equivalent_data_movement_object(source: &ObjectInfo, target: &ObjectInfo)
 fn is_superseding_unversioned_data_movement_object(source: &ObjectInfo, target: &ObjectInfo) -> bool {
     is_unversioned_data_movement_object(source)
         && is_unversioned_data_movement_object(target)
-        && !target.delete_marker
         && source
             .mod_time
             .zip(target.mod_time)
@@ -1028,6 +1027,7 @@ pub(crate) enum SourceCleanupError {
 pub(crate) struct SourceCleanupBucketFence<'a> {
     pub(crate) expected_incarnation_id: Option<uuid::Uuid>,
     pub(crate) lifecycle_guard: Option<&'a rustfs_lock::NamespaceLockGuard>,
+    pub(crate) object_mutation_fence: Option<&'a SourceCleanupMutationFence>,
 }
 
 fn ensure_source_cleanup_versions_match(
@@ -1065,7 +1065,9 @@ pub(crate) async fn ensure_source_cleanup_versions_unchanged(
 struct SourceCleanupDeleteBarrierState {
     bucket: String,
     object: String,
+    fence_pending: tokio::sync::Notify,
     arrived: tokio::sync::Notify,
+    is_paused: AtomicBool,
     release: tokio::sync::Notify,
 }
 
@@ -1079,7 +1081,7 @@ pub(crate) struct SourceCleanupDeleteBarrier {
 }
 
 #[cfg(test)]
-static SOURCE_CLEANUP_DELETE_BARRIER: std::sync::OnceLock<std::sync::Mutex<Option<Arc<SourceCleanupDeleteBarrierState>>>> =
+static SOURCE_CLEANUP_DELETE_BARRIERS: std::sync::OnceLock<std::sync::Mutex<Vec<Arc<SourceCleanupDeleteBarrierState>>>> =
     std::sync::OnceLock::new();
 
 #[cfg(test)]
@@ -1092,15 +1094,22 @@ impl SourceCleanupDeleteBarrier {
         let state = Arc::new(SourceCleanupDeleteBarrierState {
             bucket: bucket.to_string(),
             object: object.to_string(),
+            fence_pending: tokio::sync::Notify::new(),
             arrived: tokio::sync::Notify::new(),
+            is_paused: AtomicBool::new(false),
             release: tokio::sync::Notify::new(),
         });
-        let mut slot = SOURCE_CLEANUP_DELETE_BARRIER
-            .get_or_init(|| std::sync::Mutex::new(None))
+        let mut barriers = SOURCE_CLEANUP_DELETE_BARRIERS
+            .get_or_init(|| std::sync::Mutex::new(Vec::new()))
             .lock()
             .expect("source cleanup delete barrier mutex should not poison");
-        assert!(slot.is_none(), "source cleanup delete barrier must be unique");
-        *slot = Some(Arc::clone(&state));
+        assert!(
+            !barriers
+                .iter()
+                .any(|barrier| barrier.bucket == bucket && barrier.object == object),
+            "source cleanup delete barrier must be unique per object"
+        );
+        barriers.push(Arc::clone(&state));
         Self { state }
     }
 
@@ -1110,35 +1119,58 @@ impl SourceCleanupDeleteBarrier {
             .expect("source cleanup should reach the pre-delete barrier");
     }
 
+    pub(crate) async fn wait_until_fence_pending(&self) {
+        tokio::time::timeout(StdDuration::from_secs(30), self.state.fence_pending.notified())
+            .await
+            .expect("source cleanup should attempt the fixed mutation fence");
+    }
+
+    pub(crate) fn is_paused(&self) -> bool {
+        self.state.is_paused.load(Ordering::Acquire)
+    }
+
     pub(crate) fn release(&self) {
         self.state.release.notify_one();
     }
 }
 
+#[cfg(test)]
+pub(crate) fn notify_source_cleanup_mutation_fence_pending(bucket: &str, object: &str) {
+    let barrier = SOURCE_CLEANUP_DELETE_BARRIERS
+        .get_or_init(|| std::sync::Mutex::new(Vec::new()))
+        .lock()
+        .expect("source cleanup delete barrier mutex should not poison")
+        .iter()
+        .find(|barrier| barrier.bucket == bucket && barrier.object == object)
+        .cloned();
+    if let Some(barrier) = barrier {
+        barrier.fence_pending.notify_one();
+    }
+}
+
 #[cfg(test)]
 impl Drop for SourceCleanupDeleteBarrier {
     fn drop(&mut self) {
         self.state.release.notify_one();
-        let mut slot = SOURCE_CLEANUP_DELETE_BARRIER
-            .get_or_init(|| std::sync::Mutex::new(None))
+        let mut barriers = SOURCE_CLEANUP_DELETE_BARRIERS
+            .get_or_init(|| std::sync::Mutex::new(Vec::new()))
             .lock()
             .expect("source cleanup delete barrier mutex should not poison");
-        if slot.as_ref().is_some_and(|state| Arc::ptr_eq(state, &self.state)) {
-            *slot = None;
-        }
+        barriers.retain(|state| !Arc::ptr_eq(state, &self.state));
     }
 }
 
 #[cfg(test)]
 async fn pause_source_cleanup_before_delete(bucket: &str, object: &str) {
-    let barrier = SOURCE_CLEANUP_DELETE_BARRIER
-        .get_or_init(|| std::sync::Mutex::new(None))
+    let barrier = SOURCE_CLEANUP_DELETE_BARRIERS
+        .get_or_init(|| std::sync::Mutex::new(Vec::new()))
         .lock()
         .expect("source cleanup delete barrier mutex should not poison")
-        .as_ref()
-        .filter(|barrier| barrier.bucket == bucket && barrier.object == object)
+        .iter()
+        .find(|barrier| barrier.bucket == bucket && barrier.object == object)
         .cloned();
     if let Some(barrier) = barrier {
+        barrier.is_paused.store(true, Ordering::Release);
         barrier.arrived.notify_one();
         barrier.release.notified().await;
     }
@@ -1154,11 +1186,20 @@ pub(crate) async fn cleanup_source_entry_if_unchanged(
     op_label: &str,
 ) -> std::result::Result<ObjectInfo, SourceCleanupError> {
     let cleanup_key = encode_dir_object(object);
-    let ns_lock = set.new_ns_lock(bucket, cleanup_key.as_str()).await?;
-    let _guard = ns_lock
-        .get_write_lock(get_lock_acquire_timeout())
-        .await
-        .map_err(Error::from)?;
+    let source_guard = if bucket_fence
+        .object_mutation_fence
+        .is_some_and(SourceCleanupMutationFence::source_lock_covered)
+    {
+        None
+    } else {
+        let ns_lock = set.new_ns_lock(bucket, cleanup_key.as_str()).await?;
+        Some(
+            ns_lock
+                .get_write_lock(get_lock_acquire_timeout())
+                .await
+                .map_err(Error::from)?,
+        )
+    };
 
     if bucket_fence
         .lifecycle_guard
@@ -1168,6 +1209,14 @@ pub(crate) async fn cleanup_source_entry_if_unchanged(
             "{op_label}: bucket incarnation fence was lost before source cleanup"
         ))));
     }
+    if bucket_fence
+        .object_mutation_fence
+        .is_some_and(SourceCleanupMutationFence::is_lock_lost)
+    {
+        return Err(SourceCleanupError::Storage(Error::other(format!(
+            "{op_label}: object mutation fence was lost before source cleanup"
+        ))));
+    }
 
     ensure_source_cleanup_versions_unchanged(set.clone(), bucket, object, expected, allowed_missing, op_label).await?;
 
@@ -1182,7 +1231,12 @@ pub(crate) async fn cleanup_source_entry_if_unchanged(
         expected_bucket_incarnation_id: bucket_fence.expected_incarnation_id,
         ..Default::default()
     };
-    opts.add_namespace_lock_guard(&_guard);
+    if let Some(source_guard) = source_guard.as_ref() {
+        opts.add_namespace_lock_guard(source_guard);
+    }
+    if let Some(object_mutation_fence) = bucket_fence.object_mutation_fence {
+        object_mutation_fence.add_namespace_lock_fence(&mut opts);
+    }
     if let Some(bucket_lifecycle_guard) = bucket_fence.lifecycle_guard {
         opts.add_bucket_lifecycle_lock_guard(bucket_lifecycle_guard);
     }
@@ -1330,6 +1384,37 @@ fn data_movement_part_upload_failure_stage(err: &Error) -> &'static str {
     }
 }
 
+pub(crate) async fn migrate_decommission_object(
+    store: Arc<ECStore>,
+    pool_idx: usize,
+    bucket: String,
+    rd: GetObjectReader,
+    source_bucket_incarnation_id: Option<uuid::Uuid>,
+    op_label: &str,
+) -> Result<()> {
+    let source = rd.object_info.clone();
+    let _mutation_fence = store
+        .acquire_decommission_object_mutation_fence(&bucket, &source.name)
+        .await?;
+    let current = find_data_movement_target_info(store.as_ref(), pool_idx, &bucket, &source)
+        .await?
+        .ok_or(Error::FileNotFound)?;
+    if !is_equivalent_data_movement_object_identity(&source, &current, true, false) {
+        return Err(Error::FileNotFound);
+    }
+
+    migrate_object_inner(
+        store,
+        pool_idx,
+        bucket,
+        rd,
+        source_bucket_incarnation_id,
+        op_label,
+        Some(&_mutation_fence),
+    )
+    .await
+}
+
 pub(crate) async fn migrate_object(
     store: Arc<ECStore>,
     pool_idx: usize,
@@ -1337,6 +1422,18 @@ pub(crate) async fn migrate_object(
     rd: GetObjectReader,
     source_bucket_incarnation_id: Option<uuid::Uuid>,
     op_label: &str,
+) -> Result<()> {
+    migrate_object_inner(store, pool_idx, bucket, rd, source_bucket_incarnation_id, op_label, None).await
+}
+
+async fn migrate_object_inner(
+    store: Arc<ECStore>,
+    pool_idx: usize,
+    bucket: String,
+    rd: GetObjectReader,
+    source_bucket_incarnation_id: Option<uuid::Uuid>,
+    op_label: &str,
+    mutation_fence: Option<&ObjectLockDiagGuard>,
 ) -> Result<()> {
     let object_info = rd.object_info.clone();
     let has_part_checksums = object_info
@@ -1350,7 +1447,7 @@ pub(crate) async fn migrate_object(
         let mut new_multipart_opts = data_movement_new_multipart_opts(&object_info, pool_idx);
         new_multipart_opts.expected_bucket_incarnation_id = source_bucket_incarnation_id;
         let (res, target_pool_idx, expected_bucket_incarnation_id) = match store
-            .handle_new_multipart_upload_with_pool_idx(&bucket, &object_info.name, &new_multipart_opts)
+            .handle_new_multipart_upload_with_pool_idx(&bucket, &object_info.name, &new_multipart_opts, mutation_fence)
             .await
         {
             Ok(res) => res,
@@ -1448,7 +1545,7 @@ pub(crate) async fn migrate_object(
             if let Err(err) = store
                 .clone()
                 .complete_multipart_upload_for_data_movement(
-                    target_pool_idx,
+                    (target_pool_idx, mutation_fence),
                     &bucket,
                     &object_info.name,
                     &res.upload_id,
@@ -1609,7 +1706,7 @@ pub(crate) async fn migrate_object(
     let mut put_opts = data_movement_put_object_opts(&object_info, pool_idx);
     put_opts.expected_bucket_incarnation_id = source_bucket_incarnation_id;
     let (target_pool_idx, put_result) = store
-        .put_object_for_data_movement(&bucket, &object_info.name, &mut data, &put_opts)
+        .put_object_for_data_movement(&bucket, &object_info.name, &mut data, &put_opts, mutation_fence)
         .await
         .map_err(|err| data_movement_stage_error(op_label, "prepare_put_object", &bucket, &object_info.name, err))?;
     if let Err(err) = put_result {
@@ -3541,25 +3638,47 @@ mod tests {
     }
 
     #[test]
-    fn test_precondition_conflict_rejects_newer_delete_marker() {
-        let source = ObjectInfo {
-            size: 128,
-            etag: Some("etag-source".to_string()),
-            mod_time: Some(OffsetDateTime::UNIX_EPOCH),
-            ..Default::default()
-        };
-        let target = ObjectInfo {
-            delete_marker: true,
-            etag: None,
-            mod_time: OffsetDateTime::UNIX_EPOCH.checked_add(time::Duration::SECOND),
-            ..source.clone()
-        };
+    fn test_precondition_conflict_accepts_only_newer_null_delete_marker() {
+        for version_id in [None, Some(Uuid::nil())] {
+            let source = ObjectInfo {
+                version_id,
+                size: 128,
+                etag: Some("etag-source".to_string()),
+                mod_time: Some(OffsetDateTime::UNIX_EPOCH),
+                ..Default::default()
+            };
+            let target = ObjectInfo {
+                delete_marker: true,
+                etag: None,
+                mod_time: OffsetDateTime::UNIX_EPOCH.checked_add(time::Duration::SECOND),
+                ..source.clone()
+            };
 
-        let should_resume =
-            resolve_data_movement_overwrite_resume_result(&Error::PreconditionFailed, Ok(Some(target)), &source, 0, 1)
-                .expect("delete marker conflict should be evaluated");
+            assert!(
+                resolve_data_movement_overwrite_resume_result(
+                    &Error::PreconditionFailed,
+                    Ok(Some(target.clone())),
+                    &source,
+                    0,
+                    1,
+                )
+                .expect("newer null delete marker should be evaluated")
+            );
 
-        assert!(!should_resume);
+            let mut same_time = target.clone();
+            same_time.mod_time = source.mod_time;
+            assert!(
+                !resolve_data_movement_overwrite_resume_result(&Error::PreconditionFailed, Ok(Some(same_time)), &source, 0, 1,)
+                    .expect("same-generation null delete marker should be rejected")
+            );
+
+            let mut versioned = target;
+            versioned.version_id = Some(Uuid::new_v4());
+            assert!(
+                !resolve_data_movement_overwrite_resume_result(&Error::PreconditionFailed, Ok(Some(versioned)), &source, 0, 1,)
+                    .expect("a UUID delete marker must not erase a null source version")
+            );
+        }
     }
 
     #[test]

crates/ecstore/src/disk/local.rs

@@ -858,6 +858,7 @@ const EVENT_DISK_LOCAL_DIRECT_IO_FALLBACK: &str = "disk_local_direct_io_fallback
 #[cfg(target_os = "linux")]
 const EVENT_DISK_LOCAL_URING_LATCH_OFF: &str = "disk_local_uring_latch_off";
 const EVENT_DISK_LOCAL_DELETE_FAILED: &str = "disk_local_delete_failed";
+const EVENT_DISK_LOCAL_DELETE_ROLLBACK_FAILED: &str = "disk_local_delete_rollback_failed";
 const EVENT_DISK_LOCAL_CHECK_PARTS: &str = "disk_local_check_parts";
 const EVENT_DISK_LOCAL_ACCESS_FAILED: &str = "disk_local_access_failed";
 const EVENT_DISK_LOCAL_VOLUME_SETUP_FAILED: &str = "disk_local_volume_setup_failed";
@@ -6106,6 +6107,43 @@ impl LocalDisk {
         Ok((bytes, modtime))
     }
 
+    async fn write_missing_delete_marker(
+        &self,
+        volume: &str,
+        path: &str,
+        fi: FileInfo,
+        object_dir: &Path,
+        xl_path: &Path,
+        rollback_dir: Option<Uuid>,
+    ) -> Result<()> {
+        if let Some(rollback_dir) = rollback_dir {
+            let rollback_path = object_dir.join(rollback_dir.to_string());
+            fs::create_dir_all(&rollback_path).await.map_err(to_file_error)?;
+            fs::write(rollback_path.join(DELETE_MARKER_ROLLBACK_FILE), [])
+                .await
+                .map_err(to_file_error)?;
+        }
+        if let Err(err) = self.write_metadata("", volume, path, fi).await {
+            if let Some(rollback_dir) = rollback_dir
+                && let Err(restore_err) = restore_delete_rollback(object_dir, xl_path, rollback_dir, &self.publication_root).await
+            {
+                warn!(
+                    event = EVENT_DISK_LOCAL_DELETE_ROLLBACK_FAILED,
+                    component = LOG_COMPONENT_ECSTORE,
+                    subsystem = LOG_SUBSYSTEM_DISK_LOCAL,
+                    result = "failed",
+                    volume,
+                    path,
+                    rollback_dir = %rollback_dir,
+                    error = ?restore_err,
+                    "Disk local delete rollback failed"
+                );
+            }
+            return Err(err);
+        }
+        Ok(())
+    }
+
     async fn delete_versions_internal(&self, volume: &str, path: &str, fis: &[FileInfo], opts: &DeleteOptions) -> Result<()> {
         let volume_dir = self.io_get_bucket_path(volume)?;
         let xlpath = self.io_get_object_path(volume, format!("{path}/{STORAGE_FORMAT_FILE}").as_str())?;
@@ -6123,7 +6161,20 @@ impl LocalDisk {
             return restore_metadata_backup(object_dir, &xlpath, rollback_dir, &self.publication_root).await;
         }
 
-        let (data, _) = self.read_all_data_with_dmtime(volume, volume_dir.as_path(), &xlpath).await?;
+        let (data, _) = match self.read_all_data_with_dmtime(volume, volume_dir.as_path(), &xlpath).await {
+            Ok(data) => data,
+            Err(DiskError::FileNotFound) => {
+                // `deleted` alone can be an explicit marker purge; only
+                // `mark_deleted` may create metadata that was not present.
+                let Some(delete_marker) = fis.iter().find(|fi| fi.deleted && fi.mark_deleted).cloned() else {
+                    return Err(DiskError::FileNotFound);
+                };
+                return self
+                    .write_missing_delete_marker(volume, path, delete_marker, object_dir, &xlpath, opts.old_data_dir)
+                    .await;
+            }
+            Err(err) => return Err(err),
+        };
 
         if data.is_empty() {
             return Err(DiskError::FileNotFound);
@@ -10422,29 +10473,9 @@ impl DiskAPI for LocalDisk {
                 }
 
                 if fi.deleted && force_del_marker {
-                    if let Some(rollback_dir) = rollback_dir {
-                        let rollback_path = file_path.join(rollback_dir.to_string());
-                        fs::create_dir_all(&rollback_path).await.map_err(to_file_error)?;
-                        fs::write(rollback_path.join(DELETE_MARKER_ROLLBACK_FILE), [])
-                            .await
-                            .map_err(to_file_error)?;
-                    }
-                    if let Err(err) = self.write_metadata("", volume, path, fi).await {
-                        if let Some(rollback_dir) = rollback_dir
-                            && let Err(restore_err) =
-                                restore_delete_rollback(file_path.as_path(), &xl_path, rollback_dir, &self.publication_root).await
-                        {
-                            warn!(
-                                volume,
-                                path,
-                                rollback_dir = %rollback_dir,
-                                error = ?restore_err,
-                                "failed to restore metadata after delete marker commit error"
-                            );
-                        }
-                        return Err(err);
-                    }
-                    return Ok(());
+                    return self
+                        .write_missing_delete_marker(volume, path, fi, file_path.as_path(), &xl_path, rollback_dir)
+                        .await;
                 }
 
                 return if fi.version_id.is_some() {

crates/ecstore/src/disk/os.rs

@@ -315,7 +315,7 @@ pub async fn fsync_dir(dir: impl AsRef<Path>) -> io::Result<()> {
     #[cfg(unix)]
     {
         let dir = dir.as_ref().to_path_buf();
-        tokio::task::spawn_blocking(move || fsync_dir_std(dir)).await?
+        fsync_spawn_blocking(move || fsync_dir_std(dir)).await?
     }
 
     #[cfg(not(unix))]
@@ -683,7 +683,7 @@ async fn fsync_open_dst_dir_group(group: &DstDirFsyncGroup) -> io::Result<()> {
     #[cfg(test)]
     let dir = group.dir.clone();
     let dir_file = group.dir_file.clone();
-    tokio::task::spawn_blocking(move || {
+    fsync_spawn_blocking(move || {
         #[cfg(test)]
         {
             if let Some(kind) = fsync_dir_recorder::take_grouped_failure(&dir) {
@@ -1080,6 +1080,44 @@ const TEST_GLOBAL_FILE_SYNCS: usize = 64;
 
 static FILE_SYNC_PERMITS: LazyLock<Semaphore> = LazyLock::new(|| Semaphore::new(global_file_sync_limit()));
 static DISK_FILE_SYNC_LIMITERS: LazyLock<Mutex<HashMap<PathBuf, Weak<Semaphore>>>> = LazyLock::new(|| Mutex::new(HashMap::new()));
+
+/// Dedicated tokio runtime for fsync/fdatasync blocking operations. When
+/// configured with >1 threads, isolates device-bound fsync from the main
+/// blocking pool so reads (pread/stat/open) are not starved. `None` means
+/// fall back to the main runtime (zero behavior change).
+static FSYNC_RUNTIME: LazyLock<Option<tokio::runtime::Runtime>> = LazyLock::new(|| {
+    let threads =
+        rustfs_utils::get_env_usize(rustfs_config::ENV_FSYNC_BLOCKING_THREADS, rustfs_config::DEFAULT_FSYNC_BLOCKING_THREADS);
+    if threads <= 1 {
+        return None;
+    }
+    let mut builder = tokio::runtime::Builder::new_multi_thread();
+    builder
+        .worker_threads(num_cpus::get().min(8))
+        .max_blocking_threads(threads)
+        .thread_name("rustfs-fsync")
+        .thread_stack_size(512 * 1024)
+        .enable_all();
+    match builder.build() {
+        Ok(rt) => {
+            tracing::info!(threads, "fsync dedicated blocking pool enabled");
+            Some(rt)
+        }
+        Err(err) => {
+            tracing::warn!(%err, "failed to build fsync runtime, falling back to main pool");
+            None
+        }
+    }
+});
+
+/// Spawn a blocking task on the fsync-dedicated runtime if configured,
+/// otherwise fall back to the main tokio blocking pool.
+fn fsync_spawn_blocking<T: Send + 'static>(f: impl FnOnce() -> T + Send + 'static) -> tokio::task::JoinHandle<T> {
+    match FSYNC_RUNTIME.as_ref() {
+        Some(rt) => rt.spawn_blocking(f),
+        None => tokio::task::spawn_blocking(f),
+    }
+}
 static DISK_VOLUME_MUTATION_LOCKS: LazyLock<Mutex<HashMap<PathBuf, Weak<RwLock<()>>>>> =
     LazyLock::new(|| Mutex::new(HashMap::new()));
 type NamespaceMutationLock = AsyncMutex<()>;
@@ -1217,7 +1255,7 @@ where
     F: FnOnce() -> io::Result<T> + Send + 'static,
 {
     let (disk_permit, global_permit) = acquire_file_sync_permits(disk_permits).await?;
-    let result = tokio::task::spawn_blocking(move || {
+    let result = fsync_spawn_blocking(move || {
         let _disk_permit = disk_permit;
         work()
     })
@@ -2146,7 +2184,7 @@ async fn run_blocking_namespace_file_sync_operation_with_global<T: Send + 'stati
         wait_started,
     );
     let disk_permit = admission.disk_permit.clone();
-    let result = tokio::task::spawn_blocking(move || {
+    let result = fsync_spawn_blocking(move || {
         let _lease = lease;
         let _disk_permit = disk_permit;
         operation()

crates/ecstore/src/object_api/types.rs

@@ -24,7 +24,7 @@ use crate::storage_api_contracts::{
 pub struct NamespaceLockFence {
     signals: Arc<Vec<Arc<rustfs_lock::distributed_lock::LockLostSignal>>>,
     #[cfg(test)]
-    forced_lost: Arc<std::sync::atomic::AtomicBool>,
+    forced_lost: Arc<Vec<Arc<std::sync::atomic::AtomicBool>>>,
 }
 
 impl Debug for NamespaceLockFence {
@@ -40,13 +40,17 @@ impl NamespaceLockFence {
         Self {
             signals: Arc::default(),
             #[cfg(test)]
-            forced_lost: Arc::new(std::sync::atomic::AtomicBool::new(false)),
+            forced_lost: Arc::new(vec![Arc::new(std::sync::atomic::AtomicBool::new(false))]),
         }
     }
 
     pub(crate) fn is_lock_lost(&self) -> bool {
         #[cfg(test)]
-        if self.forced_lost.load(std::sync::atomic::Ordering::Acquire) {
+        if self
+            .forced_lost
+            .iter()
+            .any(|lost| lost.load(std::sync::atomic::Ordering::Acquire))
+        {
             return true;
         }
         self.signals.iter().any(|signal| signal.is_lost())
@@ -57,27 +61,26 @@ impl NamespaceLockFence {
     }
 
     fn extend(&mut self, other: &Self) {
-        if Arc::ptr_eq(&self.signals, &other.signals) {
-            return;
+        if !Arc::ptr_eq(&self.signals, &other.signals) {
+            Arc::make_mut(&mut self.signals).extend(other.signals.iter().cloned());
         }
-        Arc::make_mut(&mut self.signals).extend(other.signals.iter().cloned());
         #[cfg(test)]
-        if other.forced_lost.load(std::sync::atomic::Ordering::Acquire) {
-            self.forced_lost.store(true, std::sync::atomic::Ordering::Release);
+        if !Arc::ptr_eq(&self.forced_lost, &other.forced_lost) {
+            Arc::make_mut(&mut self.forced_lost).extend(other.forced_lost.iter().cloned());
         }
     }
 
     #[cfg(test)]
     pub(crate) fn lost_for_test() -> Self {
         let fence = Self::new();
-        fence.forced_lost.store(true, std::sync::atomic::Ordering::Release);
+        fence.forced_lost[0].store(true, std::sync::atomic::Ordering::Release);
         fence
     }
 
     #[cfg(test)]
     pub(crate) fn loss_handle_for_test() -> (Self, Arc<std::sync::atomic::AtomicBool>) {
         let fence = Self::new();
-        (fence.clone(), Arc::clone(&fence.forced_lost))
+        (fence.clone(), Arc::clone(&fence.forced_lost[0]))
     }
 }
 
@@ -411,6 +414,13 @@ impl ObjectOptions {
         self.namespace_lock_fence.get_or_insert_with(NamespaceLockFence::new);
     }
 
+    #[cfg(test)]
+    pub(crate) fn add_namespace_lock_fence_for_test(&mut self, fence: &NamespaceLockFence) {
+        self.namespace_lock_fence
+            .get_or_insert_with(NamespaceLockFence::new)
+            .extend(fence);
+    }
+
     pub(crate) fn ensure_lifecycle_delete_all_journal(&mut self) {
         self.lifecycle_delete_all_journal
             .get_or_insert_with(|| Arc::new(parking_lot::Mutex::new(LifecycleDeleteAllJournalState::default())));

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

@@ -334,6 +334,7 @@ impl ECStore {
                             lifecycle_guard: bucket_incarnation_fence
                                 .as_ref()
                                 .and_then(|guard| guard.namespace_lock_guard()),
+                            ..Default::default()
                         },
                         "rebalance",
                     ),

crates/ecstore/src/set_disk/mod.rs

@@ -735,8 +735,12 @@ pub(crate) use core::io_primitives::disk_call_counters;
 mod ctx;
 mod metadata;
 mod ops;
+#[cfg(test)]
+pub(crate) use ops::multipart::NewMultipartUploadCommitObservation;
 #[cfg(any(test, feature = "test-util"))]
 pub use ops::multipart::{MultipartCommitBarrier, MultipartCommitPause};
+#[cfg(test)]
+pub(crate) use ops::object::DeleteObjectCommitBarrier;
 #[cfg(feature = "test-util")]
 pub(crate) use ops::object::TransitionCleanupStoreBarrier as SetDiskTransitionCleanupStoreBarrier;
 pub(crate) use ops::object::body_cache_plaintext_len;
@@ -3025,6 +3029,16 @@ pub struct SetDisks {
     storage_class_config_override: Arc<std::sync::RwLock<Option<Arc<storageclass::Config>>>>,
 }
 
+// DistributedLock sends the raw ObjectKey to its clients; LockRegistry clones
+// each endpoint's canonical Arc, so an exact Arc set identifies the lock domain.
+pub(crate) fn same_distributed_lock_domain(left: &[Arc<dyn LockClient>], right: &[Arc<dyn LockClient>]) -> bool {
+    left.iter()
+        .all(|left_client| right.iter().any(|right_client| Arc::ptr_eq(left_client, right_client)))
+        && right
+            .iter()
+            .all(|right_client| left.iter().any(|left_client| Arc::ptr_eq(left_client, right_client)))
+}
+
 const ERASURE_CACHE_MAX_ENTRIES: usize = 32;
 
 #[derive(Clone, Copy, Debug, Eq, Hash, PartialEq)]
@@ -3600,6 +3614,15 @@ impl SetDisks {
         &self.ctx
     }
 
+    /// Whether both sets' namespace-lock implementations cover the same object key.
+    pub(crate) async fn shares_namespace_lock_domain(&self, other: &Self) -> bool {
+        match (self.ctx.is_dist_erasure().await, other.ctx.is_dist_erasure().await) {
+            (false, false) => Arc::ptr_eq(&self.local_lock_manager, &other.local_lock_manager),
+            (true, true) => same_distributed_lock_domain(&self.lockers, &other.lockers),
+            _ => false,
+        }
+    }
+
     /// The lock manager this set actually uses (test-only; Phase 5 Slice 3).
     #[cfg(test)]
     pub(crate) fn local_lock_manager_for_test(&self) -> &Arc<rustfs_lock::GlobalLockManager> {
@@ -4584,11 +4607,11 @@ fn should_preserve_delete_replication_state(opts: &ObjectOptions) -> bool {
 }
 
 fn should_force_delete_marker_for_missing_version(opts: &ObjectOptions) -> bool {
-    opts.delete_marker || (opts.versioned && opts.version_id.is_none() && !opts.data_movement)
+    opts.delete_marker || ((opts.versioned || opts.version_suspended) && opts.version_id.is_none() && !opts.data_movement)
 }
 
 fn resolve_delete_version_state(opts: &ObjectOptions, goi: &ObjectInfo, version_found: bool) -> (bool, bool) {
-    let mut mark_delete = goi.version_id.is_some() || (opts.versioned && opts.version_id.is_none());
+    let mut mark_delete = goi.version_id.is_some() || ((opts.versioned || opts.version_suspended) && opts.version_id.is_none());
     let mut delete_marker = opts.versioned;
 
     if opts.version_id.is_some() {

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

@@ -32,6 +32,8 @@ use crate::crash_inject::{self, CrashPoint};
 use crate::multipart_listing::paginate_multipart_listing;
 use futures::{StreamExt, stream};
 use std::future::Future;
+#[cfg(test)]
+use std::sync::atomic::AtomicBool;
 #[cfg(any(test, feature = "test-util"))]
 use std::sync::atomic::{AtomicUsize, Ordering};
 use std::time::Duration;
@@ -65,6 +67,7 @@ impl StaleMultipartCleanupGuard {
 #[cfg(any(test, feature = "test-util"))]
 #[derive(Clone, Copy, PartialEq, Eq)]
 pub enum MultipartCommitPause {
+    NewUploadBeforeLockLost,
     PutPartBeforeLockAcquire,
     PutPartBeforeLockLost,
     PutPartAfterRename,
@@ -156,6 +159,72 @@ impl Drop for MultipartCommitBarrier {
     }
 }
 
+#[cfg(test)]
+struct NewMultipartUploadCommitObservationState {
+    bucket: String,
+    object: String,
+    committed: AtomicBool,
+}
+
+#[cfg(test)]
+pub(crate) struct NewMultipartUploadCommitObservation {
+    state: Arc<NewMultipartUploadCommitObservationState>,
+}
+
+#[cfg(test)]
+static NEW_MULTIPART_UPLOAD_COMMIT_OBSERVATION: std::sync::OnceLock<
+    std::sync::Mutex<Option<Arc<NewMultipartUploadCommitObservationState>>>,
+> = std::sync::OnceLock::new();
+
+#[cfg(test)]
+impl NewMultipartUploadCommitObservation {
+    pub(crate) fn install(bucket: &str, object: &str) -> Self {
+        let state = Arc::new(NewMultipartUploadCommitObservationState {
+            bucket: bucket.to_string(),
+            object: object.to_string(),
+            committed: AtomicBool::new(false),
+        });
+        let mut slot = NEW_MULTIPART_UPLOAD_COMMIT_OBSERVATION
+            .get_or_init(|| std::sync::Mutex::new(None))
+            .lock()
+            .expect("new multipart upload commit observation mutex should not poison");
+        assert!(slot.is_none(), "new multipart upload commit observation must be unique");
+        *slot = Some(Arc::clone(&state));
+        Self { state }
+    }
+
+    pub(crate) fn committed(&self) -> bool {
+        self.state.committed.load(Ordering::Acquire)
+    }
+}
+
+#[cfg(test)]
+impl Drop for NewMultipartUploadCommitObservation {
+    fn drop(&mut self) {
+        let mut slot = NEW_MULTIPART_UPLOAD_COMMIT_OBSERVATION
+            .get_or_init(|| std::sync::Mutex::new(None))
+            .lock()
+            .expect("new multipart upload commit observation mutex should not poison");
+        if slot.as_ref().is_some_and(|state| Arc::ptr_eq(state, &self.state)) {
+            *slot = None;
+        }
+    }
+}
+
+#[cfg(test)]
+fn observe_new_multipart_upload_commit(bucket: &str, object: &str) {
+    let state = NEW_MULTIPART_UPLOAD_COMMIT_OBSERVATION
+        .get_or_init(|| std::sync::Mutex::new(None))
+        .lock()
+        .expect("new multipart upload commit observation mutex should not poison")
+        .as_ref()
+        .filter(|state| state.bucket == bucket && state.object == object)
+        .cloned();
+    if let Some(state) = state {
+        state.committed.store(true, Ordering::Release);
+    }
+}
+
 #[cfg(any(test, feature = "test-util"))]
 async fn pause_multipart_commit(bucket: &str, object: &str, pause: MultipartCommitPause) {
     let barrier = {
@@ -1615,6 +1684,30 @@ impl crate::storage_api_contracts::multipart::MultipartOperations for SetDisks {
 
         let upload_path = Self::get_multipart_upload_dir(bucket, object, upload_uuid.as_str(), opts.data_movement);
 
+        #[cfg(any(test, feature = "test-util"))]
+        pause_multipart_commit(bucket, object, MultipartCommitPause::NewUploadBeforeLockLost).await;
+        if _object_lock_guard.as_ref().is_some_and(|guard| guard.is_lock_lost()) {
+            return Err(StorageError::NamespaceLockQuorumUnavailable {
+                mode: "new_multipart_upload_commit",
+                bucket: bucket.to_string(),
+                object: object.to_string(),
+                required: 1,
+                achieved: 0,
+            });
+        }
+        if opts
+            .namespace_lock_fence
+            .as_ref()
+            .is_some_and(NamespaceLockFence::is_lock_lost)
+        {
+            return Err(StorageError::NamespaceLockQuorumUnavailable {
+                mode: "new_multipart_upload_outer_lock",
+                bucket: bucket.to_string(),
+                object: object.to_string(),
+                required: 1,
+                achieved: 0,
+            });
+        }
         ensure_multipart_bucket_lifecycle_lock_held(bucket, object, opts)?;
         Self::write_unique_file_info(
             &shuffle_disks,
@@ -1626,6 +1719,8 @@ impl crate::storage_api_contracts::multipart::MultipartOperations for SetDisks {
         )
         .await
         .map_err(|e| to_object_err(e.into(), vec![bucket, object]))?;
+        #[cfg(test)]
+        observe_new_multipart_upload_commit(bucket, object);
 
         // evalDisks
 

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

@@ -2483,6 +2483,7 @@ impl SetDisks {
                         })
                         .await?,
                     );
+                    notify_put_object_commit_namespace_acquired(bucket, object);
                 }
                 #[cfg(not(any(test, feature = "test-util")))]
                 {
@@ -4630,6 +4631,7 @@ struct PutObjectCommitBarrierState {
     arrived: tokio::sync::Notify,
     release: tokio::sync::Notify,
     namespace_pending: tokio::sync::Notify,
+    namespace_acquired: std::sync::atomic::AtomicBool,
 }
 
 #[cfg(any(test, feature = "test-util"))]
@@ -4651,6 +4653,7 @@ impl PutObjectCommitBarrier {
             arrived: tokio::sync::Notify::new(),
             release: tokio::sync::Notify::new(),
             namespace_pending: tokio::sync::Notify::new(),
+            namespace_acquired: std::sync::atomic::AtomicBool::new(false),
         });
         let mut slot = PUT_OBJECT_COMMIT_BARRIER
             .get_or_init(|| std::sync::Mutex::new(Vec::new()))
@@ -4685,6 +4688,10 @@ impl PutObjectCommitBarrier {
             .await
             .expect("put object should wait for the namespace lock after leaving the commit barrier");
     }
+
+    pub fn namespace_acquired(&self) -> bool {
+        self.state.namespace_acquired.load(std::sync::atomic::Ordering::Acquire)
+    }
 }
 
 #[cfg(any(test, feature = "test-util"))]
@@ -4741,6 +4748,22 @@ fn notify_put_object_commit_namespace_pending(bucket: &str, object: &str) {
     }
 }
 
+#[cfg(any(test, feature = "test-util"))]
+fn notify_put_object_commit_namespace_acquired(bucket: &str, object: &str) {
+    let barrier = PUT_OBJECT_COMMIT_BARRIER
+        .get_or_init(|| std::sync::Mutex::new(Vec::new()))
+        .lock()
+        .expect("put object commit barrier mutex should not poison")
+        .iter()
+        .find(|barrier| {
+            barrier.bucket == bucket && barrier.object == object && barrier.pause == PutObjectCommitPause::BeforeNamespace
+        })
+        .cloned();
+    if let Some(barrier) = barrier {
+        barrier.namespace_acquired.store(true, std::sync::atomic::Ordering::Release);
+    }
+}
+
 #[cfg(test)]
 struct DeleteObjectCommitBarrierState {
     bucket: String,
@@ -4750,7 +4773,7 @@ struct DeleteObjectCommitBarrierState {
 }
 
 #[cfg(test)]
-struct DeleteObjectCommitBarrier {
+pub(crate) struct DeleteObjectCommitBarrier {
     state: Arc<DeleteObjectCommitBarrierState>,
 }
 
@@ -4760,7 +4783,7 @@ static DELETE_OBJECT_COMMIT_BARRIER: std::sync::OnceLock<std::sync::Mutex<Option
 
 #[cfg(test)]
 impl DeleteObjectCommitBarrier {
-    fn install(bucket: &str, object: &str) -> Self {
+    pub(crate) fn install(bucket: &str, object: &str) -> Self {
         let state = Arc::new(DeleteObjectCommitBarrierState {
             bucket: bucket.to_string(),
             object: object.to_string(),
@@ -4776,13 +4799,13 @@ impl DeleteObjectCommitBarrier {
         Self { state }
     }
 
-    async fn wait_until_paused(&self) {
+    pub(crate) async fn wait_until_paused(&self) {
         tokio::time::timeout(Duration::from_secs(30), self.state.arrived.notified())
             .await
             .expect("delete object should reach the deterministic commit barrier");
     }
 
-    fn release(&self) {
+    pub(crate) fn release(&self) {
         self.state.release.notify_one();
     }
 }
@@ -5877,6 +5900,7 @@ impl crate::storage_api_contracts::object::ObjectOperations for SetDisks {
             if dobj.version_id.is_none() && (version_suspended || versioned) {
                 vr.mod_time = Some(OffsetDateTime::now_utc());
                 vr.deleted = true;
+                vr.mark_deleted = true;
                 if versioned {
                     vr.version_id = Some(Uuid::new_v4());
                 }
@@ -11636,6 +11660,7 @@ mod transition_upload_integrity_tests {
                 crate::data_movement::SourceCleanupBucketFence {
                     expected_incarnation_id: None,
                     lifecycle_guard: Some(&bucket_guard),
+                    ..Default::default()
                 },
                 "test_data_movement",
             )

crates/ecstore/src/store/mod.rs

@@ -151,7 +151,7 @@ pub(crate) mod init_format;
 pub(crate) mod list_objects;
 mod multipart;
 mod object;
-pub(crate) use object::ObjectLockDiagGuard;
+pub(crate) use object::{ObjectLockDiagGuard, SourceCleanupMutationFence};
 pub use object::{
     PrepareSelectObjectSnapshotError, PreparedGetObjectReader, SelectObjectSnapshot, SelectObjectSnapshotReadError,
     SnapshotConsistencyError,

crates/ecstore/src/store/multipart.rs

@@ -400,7 +400,7 @@ impl ECStore {
         object: &str,
         opts: &ObjectOptions,
     ) -> Result<MultipartUploadResult> {
-        self.handle_new_multipart_upload_with_pool_idx(bucket, object, opts)
+        self.handle_new_multipart_upload_with_pool_idx(bucket, object, opts, None)
             .await
             .map(|(res, _, _)| res)
     }
@@ -410,20 +410,22 @@ impl ECStore {
         bucket: &str,
         object: &str,
         opts: &ObjectOptions,
+        mutation_fence: Option<&ObjectLockDiagGuard>,
     ) -> Result<(MultipartUploadResult, usize, Option<Uuid>)> {
         check_new_multipart_args(bucket, object)?;
-        let (opts, _bucket_lifecycle_guard) = self.guard_multipart_bucket_incarnation(bucket, opts).await?;
-        let opts = &opts;
+        let (mut opts, _bucket_lifecycle_guard) = self.guard_multipart_bucket_incarnation(bucket, opts).await?;
 
         if self.single_pool() {
+            self.apply_decommission_target_mutation_fence(0, object, &mut opts, mutation_fence)
+                .await;
             return self.pools[0]
-                .new_multipart_upload(bucket, object, opts)
+                .new_multipart_upload(bucket, object, &opts)
                 .await
                 .map(|res| (res, 0, opts.expected_bucket_incarnation_id));
         }
 
         if opts.data_movement && opts.version_id.is_some() {
-            let idx = self.select_data_movement_pool_idx(bucket, object, -1, opts, false).await?;
+            let idx = self.select_data_movement_pool_idx(bucket, object, -1, &opts, false).await?;
             if idx == opts.src_pool_idx {
                 return Err(StorageError::DataMovementOverwriteErr(
                     bucket.to_owned(),
@@ -431,7 +433,9 @@ impl ECStore {
                     opts.version_id.clone().unwrap_or_default(),
                 ));
             }
-            let res = self.pools[idx].new_multipart_upload(bucket, object, opts).await?;
+            self.apply_decommission_target_mutation_fence(idx, object, &mut opts, mutation_fence)
+                .await;
+            let res = self.pools[idx].new_multipart_upload(bucket, object, &opts).await?;
             return Ok((res, idx, opts.expected_bucket_incarnation_id));
         }
 
@@ -454,7 +458,9 @@ impl ECStore {
             .await?;
 
             if !res.uploads.is_empty() {
-                let res = self.pools[idx].new_multipart_upload(bucket, object, opts).await?;
+                self.apply_decommission_target_mutation_fence(idx, object, &mut opts, mutation_fence)
+                    .await;
+                let res = self.pools[idx].new_multipart_upload(bucket, object, &opts).await?;
                 return Ok((res, idx, opts.expected_bucket_incarnation_id));
             }
         }
@@ -467,7 +473,9 @@ impl ECStore {
             ));
         }
 
-        let res = self.pools[idx].new_multipart_upload(bucket, object, opts).await?;
+        self.apply_decommission_target_mutation_fence(idx, object, &mut opts, mutation_fence)
+            .await;
+        let res = self.pools[idx].new_multipart_upload(bucket, object, &opts).await?;
         Ok((res, idx, opts.expected_bucket_incarnation_id))
     }
 
@@ -704,13 +712,14 @@ impl ECStore {
 
     pub(crate) async fn complete_multipart_upload_for_data_movement(
         self: Arc<Self>,
-        target_pool_idx: usize,
+        target: (usize, Option<&ObjectLockDiagGuard>),
         bucket: &str,
         object: &str,
         upload_id: &str,
         uploaded_parts: Vec<CompletePart>,
         opts: &ObjectOptions,
     ) -> Result<ObjectInfo> {
+        let (target_pool_idx, mutation_fence) = target;
         check_complete_multipart_args(bucket, object, upload_id)?;
         if !opts.data_movement {
             return Err(Error::other("targeted multipart completion requires data_movement options"));
@@ -739,6 +748,8 @@ impl ECStore {
             snapshot.add_lock_fences(&mut opts);
             opts.object_lock_config_snapshot = Some(snapshot);
         }
+        self.apply_decommission_target_mutation_fence(target_pool_idx, object, &mut opts, mutation_fence)
+            .await;
         #[cfg(test)]
         pause_data_movement_multipart_before_selected_completion(bucket).await;
         let pool = self

crates/ecstore/src/store/object.rs

@@ -32,12 +32,13 @@ use crate::bucket::metadata_sys::{
 use crate::bucket::object_lock::objectlock_sys::{
     check_object_lock_for_deletion_with_state, ensure_recursive_force_delete_allowed_for_state,
 };
-use crate::bucket::replication::ReplicationObjectBridge;
+use crate::bucket::replication::{DeleteReplicationConfigSnapshot, ReplicationObjectBridge};
+use crate::bucket::versioning::VersioningApi;
 use crate::disk::OldCurrentSize;
 use crate::object_api::{NamespaceLockFence, ObjectLockConfigSnapshot};
 use crate::set_disk::{
-    get_lock_acquire_timeout, get_object_lock_diag_slow_acquire_threshold, get_object_lock_diag_slow_hold_threshold,
-    is_lock_optimization_enabled, is_object_lock_diag_enabled,
+    SetDisks, get_lock_acquire_timeout, get_object_lock_diag_slow_acquire_threshold, get_object_lock_diag_slow_hold_threshold,
+    is_lock_optimization_enabled, is_object_lock_diag_enabled, same_distributed_lock_domain,
 };
 use crate::storage_api_contracts::{
     namespace::NamespaceLocking as _,
@@ -352,6 +353,8 @@ impl fmt::Display for ObjectLockDiagMode {
 
 pub(crate) struct ObjectLockDiagGuard {
     guard: rustfs_lock::NamespaceLockGuard,
+    #[cfg(test)]
+    test_namespace_lock_fence: Option<NamespaceLockFence>,
     enabled: bool,
     op: &'static str,
     bucket: Option<String>,
@@ -373,6 +376,8 @@ impl ObjectLockDiagGuard {
     ) -> Self {
         Self {
             guard,
+            #[cfg(test)]
+            test_namespace_lock_fence: None,
             enabled,
             op,
             bucket,
@@ -393,6 +398,115 @@ impl ObjectLockDiagGuard {
     pub(crate) fn is_lock_lost(&self) -> bool {
         self.guard.is_lock_lost()
     }
+
+    pub(crate) fn add_namespace_lock_fence(&self, opts: &mut ObjectOptions) {
+        opts.ensure_namespace_lock_fence();
+        if let Some(signal) = self.lock_lost_signal() {
+            opts.add_namespace_lock_lost_signal(signal);
+        }
+        #[cfg(test)]
+        if let Some(fence) = self.test_namespace_lock_fence.as_ref() {
+            opts.add_namespace_lock_fence_for_test(fence);
+        }
+    }
+}
+
+#[cfg(test)]
+#[derive(Clone, Copy, PartialEq, Eq)]
+pub(crate) enum DecommissionMutationFenceTestPhase {
+    Migration,
+    SourceCleanup,
+}
+
+#[cfg(test)]
+struct DecommissionMutationFenceLossState {
+    bucket: String,
+    object: String,
+    phase: DecommissionMutationFenceTestPhase,
+    fence: NamespaceLockFence,
+    loss_handle: Arc<std::sync::atomic::AtomicBool>,
+}
+
+#[cfg(test)]
+pub(crate) struct DecommissionMutationFenceLossHook {
+    state: Arc<DecommissionMutationFenceLossState>,
+}
+
+#[cfg(test)]
+static DECOMMISSION_MUTATION_FENCE_LOSS_HOOK: std::sync::OnceLock<
+    std::sync::Mutex<Option<Arc<DecommissionMutationFenceLossState>>>,
+> = std::sync::OnceLock::new();
+
+#[cfg(test)]
+impl DecommissionMutationFenceLossHook {
+    pub(crate) fn install(bucket: &str, object: &str, phase: DecommissionMutationFenceTestPhase) -> Self {
+        let (fence, loss_handle) = NamespaceLockFence::loss_handle_for_test();
+        let state = Arc::new(DecommissionMutationFenceLossState {
+            bucket: bucket.to_string(),
+            object: object.to_string(),
+            phase,
+            fence,
+            loss_handle,
+        });
+        let mut slot = DECOMMISSION_MUTATION_FENCE_LOSS_HOOK
+            .get_or_init(|| std::sync::Mutex::new(None))
+            .lock()
+            .expect("decommission mutation fence loss hooks should not poison");
+        assert!(slot.is_none(), "decommission mutation fence loss hook must be unique");
+        *slot = Some(Arc::clone(&state));
+        Self { state }
+    }
+
+    pub(crate) fn mark_lost(&self) {
+        self.state.loss_handle.store(true, Ordering::Release);
+    }
+}
+
+#[cfg(test)]
+impl Drop for DecommissionMutationFenceLossHook {
+    fn drop(&mut self) {
+        let mut slot = DECOMMISSION_MUTATION_FENCE_LOSS_HOOK
+            .get_or_init(|| std::sync::Mutex::new(None))
+            .lock()
+            .expect("decommission mutation fence loss hooks should not poison");
+        if slot.as_ref().is_some_and(|hook| Arc::ptr_eq(hook, &self.state)) {
+            *slot = None;
+        }
+    }
+}
+
+#[cfg(test)]
+fn decommission_mutation_fence_for_test(
+    bucket: &str,
+    object: &str,
+    phase: DecommissionMutationFenceTestPhase,
+) -> Option<NamespaceLockFence> {
+    DECOMMISSION_MUTATION_FENCE_LOSS_HOOK
+        .get_or_init(|| std::sync::Mutex::new(None))
+        .lock()
+        .expect("decommission mutation fence loss hooks should not poison")
+        .as_ref()
+        .filter(|hook| hook.bucket == bucket && hook.object == object && hook.phase == phase)
+        .map(|hook| hook.fence.clone())
+}
+
+pub(crate) struct SourceCleanupMutationFence {
+    guard: ObjectLockDiagGuard,
+    source_lock_covered: bool,
+}
+
+impl SourceCleanupMutationFence {
+    pub(crate) fn source_lock_covered(&self) -> bool {
+        self.source_lock_covered
+    }
+
+    pub(crate) fn is_lock_lost(&self) -> bool {
+        self.guard.is_lock_lost()
+    }
+
+    pub(crate) fn add_namespace_lock_fence(&self, opts: &mut ObjectOptions) {
+        self.guard.add_namespace_lock_fence(opts);
+    }
 }
 
 /// Opaque write-lock guard for the RestoreObject accept path; see
@@ -410,10 +524,7 @@ impl RestoreAcceptGuard {
     }
 
     pub fn add_namespace_lock_fence(&self, opts: &mut ObjectOptions) {
-        opts.ensure_namespace_lock_fence();
-        if let Some(signal) = self.0.lock_lost_signal() {
-            opts.add_namespace_lock_lost_signal(signal);
-        }
+        self.0.add_namespace_lock_fence(opts);
     }
 }
 
@@ -690,16 +801,6 @@ impl SelectObjectSnapshotLockLossWake {
     }
 }
 
-// LockRegistry clones its canonical client Arc for each endpoint host, so an
-// exact Arc set identifies one distributed namespace-lock quorum domain.
-fn same_distributed_lock_domain(left: &[Arc<dyn rustfs_lock::LockClient>], right: &[Arc<dyn rustfs_lock::LockClient>]) -> bool {
-    left.iter()
-        .all(|left_client| right.iter().any(|right_client| Arc::ptr_eq(left_client, right_client)))
-        && right
-            .iter()
-            .all(|right_client| left.iter().any(|left_client| Arc::ptr_eq(left_client, right_client)))
-}
-
 impl AsyncRead for SelectObjectSnapshotReader {
     fn poll_read(mut self: Pin<&mut Self>, cx: &mut Context<'_>, buf: &mut ReadBuf<'_>) -> Poll<std::io::Result<()>> {
         if self.lock_loss_wake.poll_lost(cx) || self.lease.is_lost() {
@@ -805,7 +906,7 @@ fn resolve_latest_object_access(
 }
 
 fn should_create_delete_marker_for_missing_object(opts: &ObjectOptions) -> bool {
-    opts.versioned && opts.version_id.is_none() && !opts.delete_marker && !opts.data_movement
+    (opts.versioned || opts.version_suspended) && opts.version_id.is_none() && !opts.delete_marker && !opts.data_movement
 }
 
 #[cfg(test)]
@@ -813,6 +914,8 @@ struct DeleteAfterObjectLockSnapshotBarrierState {
     bucket: String,
     arrived: tokio::sync::Notify,
     release: tokio::sync::Notify,
+    namespace_pending: tokio::sync::Notify,
+    namespace_acquired: AtomicBool,
 }
 
 #[cfg(test)]
@@ -832,6 +935,8 @@ impl DeleteAfterObjectLockSnapshotBarrier {
             bucket: bucket.to_string(),
             arrived: tokio::sync::Notify::new(),
             release: tokio::sync::Notify::new(),
+            namespace_pending: tokio::sync::Notify::new(),
+            namespace_acquired: AtomicBool::new(false),
         });
         let mut slot = DELETE_AFTER_OBJECT_LOCK_SNAPSHOT_BARRIER
             .get_or_init(|| std::sync::Mutex::new(None))
@@ -849,6 +954,18 @@ impl DeleteAfterObjectLockSnapshotBarrier {
     pub(crate) fn release(&self) {
         self.state.release.notify_one();
     }
+
+    pub(crate) async fn release_and_wait_until_namespace_pending(&self) {
+        let namespace_pending = self.state.namespace_pending.notified();
+        self.release();
+        tokio::time::timeout(Duration::from_secs(5), namespace_pending)
+            .await
+            .expect("delete should proceed to its namespace lock after leaving the snapshot barrier");
+    }
+
+    pub(crate) fn namespace_acquired(&self) -> bool {
+        self.state.namespace_acquired.load(Ordering::Acquire)
+    }
 }
 
 #[cfg(test)]
@@ -873,6 +990,97 @@ async fn pause_delete_after_object_lock_snapshot(bucket: &str) {
         .as_ref()
         .filter(|state| state.bucket == bucket)
         .cloned();
+    if let Some(state) = state {
+        state.arrived.notify_one();
+        state.release.notified().await;
+        state.namespace_pending.notify_one();
+    }
+}
+
+#[cfg(test)]
+fn notify_delete_namespace_acquired(bucket: &str) {
+    let state = DELETE_AFTER_OBJECT_LOCK_SNAPSHOT_BARRIER
+        .get_or_init(|| std::sync::Mutex::new(None))
+        .lock()
+        .expect("delete snapshot barrier mutex should not poison")
+        .as_ref()
+        .filter(|state| state.bucket == bucket)
+        .cloned();
+    if let Some(state) = state {
+        state.namespace_acquired.store(true, Ordering::Release);
+    }
+}
+
+#[cfg(test)]
+struct VersionedDeleteMarkerCommitBarrierState {
+    bucket: String,
+    object: String,
+    arrived: tokio::sync::Notify,
+    release: tokio::sync::Notify,
+}
+
+#[cfg(test)]
+pub(crate) struct VersionedDeleteMarkerCommitBarrier {
+    state: Arc<VersionedDeleteMarkerCommitBarrierState>,
+}
+
+#[cfg(test)]
+static VERSIONED_DELETE_MARKER_COMMIT_BARRIER: std::sync::OnceLock<
+    std::sync::Mutex<Option<Arc<VersionedDeleteMarkerCommitBarrierState>>>,
+> = std::sync::OnceLock::new();
+
+#[cfg(test)]
+impl VersionedDeleteMarkerCommitBarrier {
+    pub(crate) fn install(bucket: &str, object: &str) -> Self {
+        let state = Arc::new(VersionedDeleteMarkerCommitBarrierState {
+            bucket: bucket.to_string(),
+            object: object.to_string(),
+            arrived: tokio::sync::Notify::new(),
+            release: tokio::sync::Notify::new(),
+        });
+        let mut slot = VERSIONED_DELETE_MARKER_COMMIT_BARRIER
+            .get_or_init(|| std::sync::Mutex::new(None))
+            .lock()
+            .expect("versioned delete-marker commit barrier mutex should not poison");
+        assert!(slot.is_none(), "versioned delete-marker commit barrier must be unique");
+        *slot = Some(Arc::clone(&state));
+        Self { state }
+    }
+
+    pub(crate) async fn wait_until_paused(&self) {
+        tokio::time::timeout(Duration::from_secs(30), self.state.arrived.notified())
+            .await
+            .expect("versioned DELETE should reach the post-marker-commit barrier");
+    }
+
+    pub(crate) fn release(&self) {
+        self.state.release.notify_one();
+    }
+}
+
+#[cfg(test)]
+impl Drop for VersionedDeleteMarkerCommitBarrier {
+    fn drop(&mut self) {
+        self.state.release.notify_one();
+        let mut slot = VERSIONED_DELETE_MARKER_COMMIT_BARRIER
+            .get_or_init(|| std::sync::Mutex::new(None))
+            .lock()
+            .expect("versioned delete-marker commit barrier mutex should not poison");
+        if slot.as_ref().is_some_and(|state| Arc::ptr_eq(state, &self.state)) {
+            *slot = None;
+        }
+    }
+}
+
+#[cfg(test)]
+async fn pause_versioned_delete_marker_after_commit(bucket: &str, object: &str) {
+    let state = VERSIONED_DELETE_MARKER_COMMIT_BARRIER
+        .get_or_init(|| std::sync::Mutex::new(None))
+        .lock()
+        .expect("versioned delete-marker commit barrier mutex should not poison")
+        .as_ref()
+        .filter(|state| state.bucket == bucket && state.object == object)
+        .cloned();
     if let Some(state) = state {
         state.arrived.notify_one();
         state.release.notified().await;
@@ -913,6 +1121,160 @@ fn writer_pool_lookup_opts(opts: &ObjectOptions, no_lock: bool) -> ObjectOptions
     lookup_opts
 }
 
+fn delete_pool_lookup_opts(opts: &ObjectOptions, no_lock: bool) -> ObjectOptions {
+    let mut lookup_opts = writer_pool_lookup_opts(opts, no_lock);
+    lookup_opts.skip_decommissioned = opts.data_movement;
+    lookup_opts
+}
+
+fn should_delete_from_all_pools(opts: &ObjectOptions, pool_count: usize) -> bool {
+    pool_count > 0 && (!opts.versioned && !opts.version_suspended || opts.version_id.is_some())
+}
+
+fn batch_delete_creates_latest_marker(object: &ObjectToDelete, delete_config_snapshot: &DeleteReplicationConfigSnapshot) -> bool {
+    if object.version_id.is_some() {
+        return false;
+    }
+
+    let object_name = decode_dir_object(&object.object_name);
+    let (versioned, version_suspended) = delete_config_snapshot.versioning_config().delete_state(&object_name);
+    versioned || version_suspended
+}
+
+fn batch_delete_targets_pool(creates_latest_marker: bool, marker_target_pool_idx: Option<usize>, pool_idx: usize) -> bool {
+    !creates_latest_marker || marker_target_pool_idx == Some(pool_idx)
+}
+
+#[cfg(test)]
+struct BatchDeletePoolErrorInjectionState {
+    bucket: String,
+    pool_idx: usize,
+    errors: std::collections::HashMap<String, Error>,
+    observed: std::sync::atomic::AtomicUsize,
+}
+
+#[cfg(test)]
+pub(crate) struct BatchDeletePoolErrorInjection {
+    state: Arc<BatchDeletePoolErrorInjectionState>,
+}
+
+#[cfg(test)]
+static BATCH_DELETE_POOL_ERROR_INJECTION: std::sync::OnceLock<std::sync::Mutex<Option<Arc<BatchDeletePoolErrorInjectionState>>>> =
+    std::sync::OnceLock::new();
+
+#[cfg(test)]
+impl BatchDeletePoolErrorInjection {
+    pub(crate) fn install(bucket: &str, pool_idx: usize, errors: Vec<(String, Error)>) -> Self {
+        let state = Arc::new(BatchDeletePoolErrorInjectionState {
+            bucket: bucket.to_string(),
+            pool_idx,
+            errors: errors.into_iter().collect(),
+            observed: std::sync::atomic::AtomicUsize::new(0),
+        });
+        let mut slot = BATCH_DELETE_POOL_ERROR_INJECTION
+            .get_or_init(|| std::sync::Mutex::new(None))
+            .lock()
+            .expect("batch delete pool error injection mutex should not poison");
+        assert!(slot.is_none(), "batch delete pool error injection must be unique");
+        *slot = Some(Arc::clone(&state));
+        Self { state }
+    }
+
+    pub(crate) fn observed(&self) -> usize {
+        self.state.observed.load(Ordering::Acquire)
+    }
+}
+
+#[cfg(test)]
+impl Drop for BatchDeletePoolErrorInjection {
+    fn drop(&mut self) {
+        let mut slot = BATCH_DELETE_POOL_ERROR_INJECTION
+            .get_or_init(|| std::sync::Mutex::new(None))
+            .lock()
+            .expect("batch delete pool error injection mutex should not poison");
+        if slot.as_ref().is_some_and(|state| Arc::ptr_eq(state, &self.state)) {
+            *slot = None;
+        }
+    }
+}
+
+#[cfg(test)]
+fn inject_batch_delete_pool_errors(
+    bucket: &str,
+    pool_idx: usize,
+    object_names: &[String],
+    result: &mut (Vec<DeletedObject>, Vec<Option<Error>>),
+) {
+    let state = BATCH_DELETE_POOL_ERROR_INJECTION
+        .get_or_init(|| std::sync::Mutex::new(None))
+        .lock()
+        .expect("batch delete pool error injection mutex should not poison")
+        .as_ref()
+        .filter(|state| state.bucket == bucket && state.pool_idx == pool_idx)
+        .cloned();
+    let Some(state) = state else {
+        return;
+    };
+
+    for (idx, object_name) in object_names.iter().enumerate() {
+        let Some(error) = state.errors.get(object_name) else {
+            continue;
+        };
+        if result.1[idx].is_none() && result.0[idx].found {
+            result.1[idx] = Some(error.clone());
+            state.observed.fetch_add(1, Ordering::AcqRel);
+        }
+    }
+}
+
+fn resolve_batch_delete_pool_results<'a>(
+    initial_error: Option<Error>,
+    pool_results: impl IntoIterator<Item = (&'a DeletedObject, &'a Option<Error>)>,
+) -> (Option<DeletedObject>, Option<Error>, bool) {
+    let mut failure = initial_error.map(|err| (None, err));
+    let mut deleted = None;
+    let mut fallback: Option<(DeletedObject, Option<Error>)> = None;
+    let mut attempted = false;
+
+    for (pool_delete, pool_error) in pool_results {
+        attempted = true;
+        match pool_error {
+            Some(err) if is_err_object_not_found(err) || is_err_version_not_found(err) => {
+                if fallback.as_ref().is_none_or(|(_, error)| error.is_none()) {
+                    fallback = Some(((*pool_delete).clone(), Some(err.clone())));
+                }
+            }
+            Some(err) => {
+                if failure.is_none() {
+                    failure = Some((Some((*pool_delete).clone()), err.clone()));
+                }
+            }
+            None if pool_delete.found => {
+                if deleted.is_none() {
+                    deleted = Some((*pool_delete).clone());
+                }
+            }
+            None => {
+                if fallback.is_none() {
+                    fallback = Some(((*pool_delete).clone(), None));
+                }
+            }
+        }
+    }
+
+    if let Some((failed_delete, err)) = failure {
+        return (failed_delete, Some(err), attempted);
+    }
+    if let Some(deleted) = deleted {
+        return (Some(deleted), None, attempted);
+    }
+    if let Some((deleted, err)) = fallback {
+        return (Some(deleted), err, attempted);
+    }
+
+    (None, None, attempted)
+}
+
 fn transition_restore_pool_opts(opts: &ObjectOptions) -> ObjectOptions {
     let mut lookup_opts = opts.clone();
     lookup_opts.skip_decommissioned = true;
@@ -1533,6 +1895,89 @@ impl ECStore {
         )))
     }
 
+    pub(crate) async fn acquire_decommission_object_mutation_fence(
+        &self,
+        bucket: &str,
+        object: &str,
+    ) -> Result<ObjectLockDiagGuard> {
+        if self.ctx.lock_manager().is_disabled() {
+            return Err(Error::other("decommission object migration requires namespace locking"));
+        }
+
+        #[cfg(test)]
+        let test_namespace_lock_fence =
+            decommission_mutation_fence_for_test(bucket, object, DecommissionMutationFenceTestPhase::Migration);
+        let object = encode_dir_object(object);
+        let mut opts = ObjectOptions::default();
+        let guard = self
+            .acquire_object_read_lock_if_needed("decommission_object", bucket, &object, &mut opts)
+            .await?
+            .ok_or_else(|| Error::other("decommission object migration failed to acquire its namespace fence"))?;
+        #[cfg(test)]
+        let guard = {
+            let mut guard = guard;
+            guard.test_namespace_lock_fence = test_namespace_lock_fence;
+            guard
+        };
+        Ok(guard)
+    }
+
+    pub(super) async fn apply_decommission_target_mutation_fence(
+        &self,
+        target_pool_idx: usize,
+        object: &str,
+        opts: &mut ObjectOptions,
+        mutation_fence: Option<&ObjectLockDiagGuard>,
+    ) {
+        let Some(mutation_fence) = mutation_fence else {
+            return;
+        };
+
+        mutation_fence.add_namespace_lock_fence(opts);
+        let fixed_set = self.pools.first().and_then(|pool| pool.disk_set.first());
+        let target_set = self.pools.get(target_pool_idx).map(|pool| pool.get_disks_by_key(object));
+        opts.no_lock = match (fixed_set, target_set) {
+            (Some(fixed), Some(target)) => fixed.shares_namespace_lock_domain(&target).await,
+            _ => false,
+        };
+    }
+
+    pub(crate) async fn acquire_decommission_source_cleanup_fence(
+        &self,
+        bucket: &str,
+        object: &str,
+        source_set: &SetDisks,
+    ) -> Result<SourceCleanupMutationFence> {
+        if self.ctx.lock_manager().is_disabled() {
+            return Err(Error::other("decommission source cleanup requires namespace locking"));
+        }
+
+        #[cfg(test)]
+        crate::data_movement::notify_source_cleanup_mutation_fence_pending(bucket, object);
+        #[cfg(test)]
+        let test_namespace_lock_fence =
+            decommission_mutation_fence_for_test(bucket, object, DecommissionMutationFenceTestPhase::SourceCleanup);
+        let object = encode_dir_object(object);
+        let fixed_set = Arc::clone(&self.pools[0].disk_set[0]);
+        let source_lock_covered = fixed_set.shares_namespace_lock_domain(source_set).await;
+        // Lock order: fixed store mutation domain first; source cleanup takes its
+        // hashed source-domain lock second only when this guard does not cover it.
+        let guard = self
+            .acquire_object_write_lock("decommission_source_cleanup", bucket, &object)
+            .await?;
+        #[cfg(test)]
+        let guard = {
+            let mut guard = guard;
+            guard.test_namespace_lock_fence = test_namespace_lock_fence;
+            guard
+        };
+
+        Ok(SourceCleanupMutationFence {
+            guard,
+            source_lock_covered,
+        })
+    }
+
     pub(crate) async fn acquire_all_object_read_locks(
         &self,
         op: &'static str,
@@ -1986,14 +2431,17 @@ impl ECStore {
         object: &str,
         data: &mut PutObjReader,
         opts: &ObjectOptions,
+        mutation_fence: Option<&ObjectLockDiagGuard>,
     ) -> Result<(usize, Result<ObjectInfo>)> {
         if !opts.data_movement {
             return Err(Error::other("data movement PUT requires data_movement options"));
         }
-        let (object, opts) = self.prepare_put_object(bucket, object, opts).await?;
+        let (object, mut opts) = self.prepare_put_object(bucket, object, opts).await?;
         let idx = self
             .select_put_object_pool_idx(bucket, object.as_str(), data.size(), &opts)
             .await?;
+        self.apply_decommission_target_mutation_fence(idx, object.as_str(), &mut opts, mutation_fence)
+            .await;
         let result = self.pools[idx]
             .put_object_with_old_current_size(bucket, &object, data, &opts)
             .await
@@ -2446,6 +2894,10 @@ impl ECStore {
         } else {
             None
         };
+        #[cfg(test)]
+        if _object_lock_guard.is_some() {
+            notify_delete_namespace_acquired(bucket);
+        }
         if let Some(trigger) = opts.lifecycle_delete_all.as_ref() {
             let configs = delete_all_configs.as_ref().ok_or(StorageError::PreconditionFailed)?;
             let expected_bucket_incarnation_id = opts.expected_bucket_incarnation_id.ok_or(StorageError::PreconditionFailed)?;
@@ -2479,7 +2931,7 @@ impl ECStore {
             return Ok(ObjectInfo::default());
         }
 
-        let gopts = writer_pool_lookup_opts(&opts, true);
+        let gopts = delete_pool_lookup_opts(&opts, true);
 
         if opts.data_movement {
             let existing_pool_info = self.get_pool_info_existing_with_opts(bucket, object, &gopts).await;
@@ -2584,6 +3036,8 @@ impl ECStore {
             Err(err) if is_err_object_not_found(&err) && should_create_delete_marker_for_missing_object(&opts) => {
                 let target_pool_idx = self.get_pool_idx_no_lock(bucket, object, 0).await?;
                 let mut obj = self.pools[target_pool_idx].delete_object(bucket, object, opts).await?;
+                #[cfg(test)]
+                pause_versioned_delete_marker_after_commit(bucket, object).await;
                 obj.name = decode_dir_object(object);
                 return Ok(obj);
             }
@@ -2622,7 +3076,7 @@ impl ECStore {
             None
         };
 
-        if !errs.is_empty() && !opts.versioned && !opts.version_suspended {
+        if should_delete_from_all_pools(&opts, errs.len()) {
             let mut obj = match self.delete_object_from_all_pools(bucket, object, &opts, errs).await {
                 Ok(obj) => obj,
                 Err(err) => {
@@ -2646,6 +3100,8 @@ impl ECStore {
 
             match pool.delete_object(bucket, object, opts.clone()).await {
                 Ok(res) => {
+                    #[cfg(test)]
+                    pause_versioned_delete_marker_after_commit(bucket, object).await;
                     if let (Some(api), Some(je)) = (tier_journal_api.as_ref(), journal_entry.as_ref()) {
                         commit_prepared_tier_delete_journal_entry(api, je).await;
                     }
@@ -2776,30 +3232,104 @@ impl ECStore {
             Ok(guards) => guards,
             Err(err) => return return_batch_delete_lock_error(objects.as_slice(), err),
         };
+        #[cfg(test)]
+        if !_object_lock_guards.is_empty() {
+            notify_delete_namespace_acquired(bucket);
+        }
+
+        let delete_config_snapshot = opts
+            .delete_replication_config_snapshot
+            .as_deref()
+            .expect("batch delete replication config snapshot should be loaded");
+        let latest_marker_objects = objects
+            .iter()
+            .map(|object| batch_delete_creates_latest_marker(object, delete_config_snapshot))
+            .collect::<Vec<_>>();
+        let marker_target_results = join_all(objects.iter().zip(&latest_marker_objects).map(
+            |(object, creates_marker)| async move {
+                if *creates_marker {
+                    Some(self.get_pool_idx_no_lock(bucket, &object.object_name, 0).await)
+                } else {
+                    None
+                }
+            },
+        ))
+        .await;
+        let mut marker_target_pool_indices = Vec::with_capacity(objects.len());
+        for (idx, target_result) in marker_target_results.into_iter().enumerate() {
+            match target_result {
+                Some(Ok(pool_idx)) => marker_target_pool_indices.push(Some(pool_idx)),
+                Some(Err(err)) => {
+                    del_errs[idx] = Some(err);
+                    marker_target_pool_indices.push(None);
+                }
+                None => marker_target_pool_indices.push(None),
+            }
+        }
 
         let mut futures = Vec::with_capacity(self.pools.len());
-
         for pool in self.pools.iter() {
             if self.is_pool_rebalancing(pool.pool_idx).await {
                 continue;
             }
-            futures.push(pool.delete_objects(bucket, objects.clone(), opts.clone()));
+
+            let (object_indices, pool_objects): (Vec<_>, Vec<_>) = objects
+                .iter()
+                .enumerate()
+                .filter(|(idx, _)| {
+                    batch_delete_targets_pool(latest_marker_objects[*idx], marker_target_pool_indices[*idx], pool.pool_idx)
+                })
+                .map(|(idx, object)| (idx, object.clone()))
+                .unzip();
+            if pool_objects.is_empty() {
+                continue;
+            }
+
+            let pool_opts = opts.clone();
+            futures.push(async move {
+                #[cfg(test)]
+                let pool_object_names = pool_objects
+                    .iter()
+                    .map(|object| object.object_name.clone())
+                    .collect::<Vec<_>>();
+                let result = pool.delete_objects(bucket, pool_objects, pool_opts).await;
+                #[cfg(test)]
+                let result = {
+                    let mut result = result;
+                    inject_batch_delete_pool_errors(bucket, pool.pool_idx, &pool_object_names, &mut result);
+                    result
+                };
+                (object_indices, result)
+            });
         }
 
         let results = join_all(futures).await;
 
         for idx in 0..del_objects.len() {
-            for (dels, errs) in results.iter() {
-                if errs[idx].is_none() && dels[idx].found {
-                    del_errs[idx] = None;
-                    del_objects[idx] = dels[idx].clone();
-                    break;
-                }
+            let pool_results = results.iter().filter_map(|(object_indices, (dels, errs))| {
+                let pool_object_idx = object_indices.binary_search(&idx).ok()?;
+                Some((&dels[pool_object_idx], &errs[pool_object_idx]))
+            });
+            let (deleted, error, attempted) = resolve_batch_delete_pool_results(del_errs[idx].take(), pool_results);
+            if let Some(deleted) = deleted {
+                del_objects[idx] = deleted;
+            }
+            del_errs[idx] = error;
 
-                if del_errs[idx].is_none() {
-                    del_errs[idx] = errs[idx].clone();
-                    del_objects[idx] = dels[idx].clone();
-                }
+            if !attempted && del_errs[idx].is_none() && latest_marker_objects[idx] {
+                del_objects[idx] = DeletedObject {
+                    object_name: objects[idx].object_name.clone(),
+                    version_id: objects[idx].version_id,
+                    ..Default::default()
+                };
+                del_errs[idx] = Some(StorageError::ObjectNotFound(bucket.to_owned(), objects[idx].object_name.clone()));
+            }
+        }
+
+        #[cfg(test)]
+        for (idx, object) in objects.iter().enumerate() {
+            if del_errs[idx].is_none() && del_objects[idx].delete_marker {
+                pause_versioned_delete_marker_after_commit(bucket, &object.object_name).await;
             }
         }
 
@@ -3374,6 +3904,80 @@ mod tests {
         assert!(!same_distributed_lock_domain(&[first, second], &[other]));
     }
 
+    #[tokio::test]
+    async fn decommission_fence_covers_dist_sets_with_same_clients_despite_different_namespaces() {
+        let ctx = Arc::new(crate::runtime::instance::InstanceContext::new());
+        let (_dirs, original_sets) = make_local_two_set_sets_with_ctx(Arc::clone(&ctx)).await;
+        let mut second_set = (*original_sets.disk_set[1]).clone();
+        second_set.lockers = original_sets.disk_set[0].lockers.clone();
+        let mut sets = (*original_sets).clone();
+        sets.disk_set[1] = Arc::new(second_set);
+        let sets = Arc::new(sets);
+        ctx.update_erasure_type(SetupType::DistErasure).await;
+
+        assert!(
+            sets.disk_set[0]
+                .lockers
+                .iter()
+                .zip(&sets.disk_set[1].lockers)
+                .all(|(fixed, hashed)| Arc::ptr_eq(fixed, hashed)),
+            "the regression requires identical distributed lock clients"
+        );
+        assert_ne!(sets.disk_set[0].set_index, sets.disk_set[1].set_index);
+
+        let pool_config = sets.endpoints.clone();
+        let store = new_prepared_reader_test_store_from_pools(vec![Arc::clone(&sets)], vec![pool_config], ctx);
+        let object = (0..1_000)
+            .map(|index| format!("decommission-dist-domain-{index}.bin"))
+            .find(|candidate| Arc::ptr_eq(&sets.get_disks_by_key(candidate), &sets.disk_set[1]))
+            .expect("a key should hash to the second set namespace");
+        let mutation_fence = store
+            .acquire_decommission_object_mutation_fence("bucket", &object)
+            .await
+            .expect("the fixed distributed mutation fence should be acquired");
+        let target_lock = sets.disk_set[1]
+            .new_ns_lock("bucket", &object)
+            .await
+            .expect("the hashed-set namespace lock should be created");
+        let target_err = target_lock
+            .get_write_lock(Duration::from_millis(50))
+            .await
+            .expect_err("the fixed read fence must conflict through the shared clients");
+        assert!(matches!(target_err, rustfs_lock::LockError::Timeout { .. }));
+
+        let mut put_opts = ObjectOptions::default();
+        store
+            .apply_decommission_target_mutation_fence(0, &object, &mut put_opts, Some(&mutation_fence))
+            .await;
+        assert!(put_opts.no_lock, "migration target PUT must reuse the covering fixed fence");
+
+        let mut multipart_opts = ObjectOptions::default();
+        store
+            .apply_decommission_target_mutation_fence(0, &object, &mut multipart_opts, Some(&mutation_fence))
+            .await;
+        assert!(multipart_opts.no_lock, "migration target multipart must reuse the covering fixed fence");
+        drop(mutation_fence);
+
+        let cleanup_object = (0..1_000)
+            .map(|index| format!("decommission-dist-cleanup-{index}.bin"))
+            .find(|candidate| Arc::ptr_eq(&sets.get_disks_by_key(candidate), &sets.disk_set[1]))
+            .expect("a cleanup key should hash to the second set namespace");
+        let source_fence = store
+            .acquire_decommission_source_cleanup_fence("bucket", &cleanup_object, sets.disk_set[1].as_ref())
+            .await
+            .expect("the fixed distributed cleanup fence should be acquired");
+        assert!(source_fence.source_lock_covered(), "source cleanup must reuse the covering fixed fence");
+        let source_lock = sets.disk_set[1]
+            .new_ns_lock("bucket", &cleanup_object)
+            .await
+            .expect("the source-set namespace lock should be created");
+        let source_err = source_lock
+            .get_read_lock(Duration::from_millis(50))
+            .await
+            .expect_err("the fixed write fence must conflict through the shared clients");
+        assert!(matches!(source_err, rustfs_lock::LockError::Timeout { .. }));
+    }
+
     #[test]
     fn select_snapshot_version_matching_normalizes_null_and_uuid_forms() {
         let nil = Uuid::nil();
@@ -4433,6 +5037,159 @@ mod tests {
         assert_eq!(lookup_opts.version_id.as_deref(), Some("vid-1"));
     }
 
+    #[test]
+    fn ordinary_delete_lookup_includes_decommission_source_and_skips_rebalance_source() {
+        let lookup_opts = delete_pool_lookup_opts(&ObjectOptions::default(), true);
+
+        assert!(lookup_opts.no_lock);
+        assert!(!lookup_opts.skip_decommissioned);
+        assert!(lookup_opts.skip_rebalancing);
+
+        let explicit_version = delete_pool_lookup_opts(
+            &ObjectOptions {
+                versioned: true,
+                version_id: Some(uuid::Uuid::new_v4().to_string()),
+                ..Default::default()
+            },
+            true,
+        );
+        assert!(!explicit_version.skip_decommissioned);
+    }
+
+    #[test]
+    fn delete_fans_out_for_unversioned_and_explicit_version_mutations() {
+        assert!(should_delete_from_all_pools(&ObjectOptions::default(), 1));
+        assert!(should_delete_from_all_pools(
+            &ObjectOptions {
+                versioned: true,
+                version_id: Some(uuid::Uuid::new_v4().to_string()),
+                ..Default::default()
+            },
+            2,
+        ));
+        assert!(!should_delete_from_all_pools(
+            &ObjectOptions {
+                versioned: true,
+                ..Default::default()
+            },
+            1,
+        ));
+        assert!(!should_delete_from_all_pools(&ObjectOptions::default(), 0));
+    }
+
+    #[test]
+    fn batch_delete_identifies_only_latest_versioned_markers() {
+        let versioned = DeleteReplicationConfigSnapshot::from_configs_for_test(
+            s3s::dto::VersioningConfiguration {
+                status: Some(s3s::dto::BucketVersioningStatus::from_static(s3s::dto::BucketVersioningStatus::ENABLED)),
+                ..Default::default()
+            },
+            None,
+        );
+        let latest = ObjectToDelete {
+            object_name: "latest".to_string(),
+            ..Default::default()
+        };
+        assert!(batch_delete_creates_latest_marker(&latest, &versioned));
+        assert!(!batch_delete_targets_pool(true, Some(1), 0));
+        assert!(batch_delete_targets_pool(true, Some(1), 1));
+        assert!(!batch_delete_targets_pool(true, Some(1), 2));
+
+        let explicit = ObjectToDelete {
+            object_name: "explicit".to_string(),
+            version_id: Some(uuid::Uuid::new_v4()),
+            ..Default::default()
+        };
+        assert!(!batch_delete_creates_latest_marker(&explicit, &versioned));
+        assert!(batch_delete_targets_pool(false, Some(1), 0));
+
+        let unversioned = DeleteReplicationConfigSnapshot::default();
+        assert!(!batch_delete_creates_latest_marker(&latest, &unversioned));
+        assert!(batch_delete_targets_pool(false, None, 0));
+    }
+
+    #[test]
+    fn batch_delete_pool_failures_override_success_in_any_pool_order() {
+        let success = DeletedObject {
+            object_name: "object".to_string(),
+            found: true,
+            ..Default::default()
+        };
+        let source_errors = [
+            StorageError::ErasureWriteQuorum,
+            StorageError::NamespaceLockQuorumUnavailable {
+                mode: "delete_objects_commit",
+                bucket: "bucket".to_string(),
+                object: "object".to_string(),
+                required: 1,
+                achieved: 0,
+            },
+        ];
+
+        for source_error in source_errors {
+            for source_first in [true, false] {
+                let failed = (DeletedObject::default(), Some(source_error.clone()));
+                let succeeded = (success.clone(), None);
+                let pool_results = if source_first {
+                    vec![failed, succeeded]
+                } else {
+                    vec![succeeded, failed]
+                };
+
+                let (_, error, attempted) =
+                    resolve_batch_delete_pool_results(None, pool_results.iter().map(|(deleted, error)| (deleted, error)));
+
+                assert!(attempted);
+                assert_eq!(error, Some(source_error.clone()));
+            }
+        }
+    }
+
+    #[test]
+    fn batch_delete_ignores_missing_pool_only_after_another_pool_succeeds() {
+        let success = DeletedObject {
+            object_name: "object".to_string(),
+            found: true,
+            ..Default::default()
+        };
+        let missing_errors = [
+            StorageError::ObjectNotFound("bucket".to_string(), "object".to_string()),
+            StorageError::VersionNotFound("bucket".to_string(), "object".to_string(), "version".to_string()),
+        ];
+
+        for missing_error in missing_errors {
+            let missing = (DeletedObject::default(), Some(missing_error.clone()));
+            for missing_first in [true, false] {
+                let succeeded = (success.clone(), None);
+                let pool_results = if missing_first {
+                    vec![missing.clone(), succeeded]
+                } else {
+                    vec![succeeded, missing.clone()]
+                };
+                let (deleted, error, attempted) =
+                    resolve_batch_delete_pool_results(None, pool_results.iter().map(|(deleted, error)| (deleted, error)));
+
+                assert!(attempted);
+                let deleted = deleted.expect("successful pool result should be retained");
+                assert!(deleted.found);
+                assert_eq!(deleted.object_name, success.object_name.as_str());
+                assert!(error.is_none());
+            }
+
+            let missing_only = [missing];
+            let (_, error, attempted) =
+                resolve_batch_delete_pool_results(None, missing_only.iter().map(|(deleted, error)| (deleted, error)));
+            assert!(attempted);
+            assert_eq!(error, Some(missing_error));
+        }
+
+        let silent_missing = [(DeletedObject::default(), None)];
+        let (_, error, attempted) =
+            resolve_batch_delete_pool_results(None, silent_missing.iter().map(|(deleted, error)| (deleted, error)));
+        assert!(attempted);
+        assert!(error.is_none());
+    }
+
     #[test]
     fn data_movement_pool_lookup_opts_keeps_no_lock_for_tiered_moves() {
         let lookup_opts = data_movement_pool_lookup_opts(

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/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/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

scripts/check_test_wiring.py

@@ -206,6 +206,13 @@ def check_runner_selection(root: Path) -> list[str]:
     return errors
 
 
+def check_s3_tests_runner(root: Path) -> list[str]:
+    runner = (root / "scripts/s3-tests/run.sh").read_text()
+    if "--showlocals" in runner:
+        return ["scripts/s3-tests/run.sh: pytest failure diagnostics must not dump local values"]
+    return []
+
+
 def profile_selection(root: Path, profile: str) -> str:
     if not re.fullmatch(r"e2e-[a-z0-9-]+", profile):
         raise ValueError(f"invalid e2e profile name: {profile}")
@@ -272,6 +279,7 @@ def validate(root: Path) -> list[str]:
     errors.extend(check_e2e_modules(root))
     errors.extend(check_fuzz_targets(root))
     errors.extend(check_runner_selection(root))
+    errors.extend(check_s3_tests_runner(root))
     errors.extend(check_profile_definitions(root))
     return errors
 
@@ -341,6 +349,23 @@ class SelfTests(unittest.TestCase):
             )
             self.assertEqual(len(check_fuzz_targets(root)), 1)
 
+    def test_s3_runner_rejects_unbounded_failure_locals(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            runner = root / "scripts/s3-tests/run.sh"
+            runner.parent.mkdir(parents=True)
+            runner.write_text("tox -- -vv -ra --tb=long\n")
+            self.assertEqual(check_s3_tests_runner(root), [])
+            runner.write_text("tox -- -vv -ra --showlocals --tb=long\n")
+            self.assertEqual(len(check_s3_tests_runner(root)), 1)
+            with (
+                mock.patch(__name__ + ".check_e2e_modules", return_value=[]),
+                mock.patch(__name__ + ".check_fuzz_targets", return_value=[]),
+                mock.patch(__name__ + ".check_runner_selection", return_value=[]),
+                mock.patch(__name__ + ".check_profile_definitions", return_value=[]),
+            ):
+                self.assertEqual(len(validate(root)), 1)
+
     def test_profile_listing_enforces_selection(self) -> None:
         with tempfile.TemporaryDirectory() as tmp:
             root = Path(tmp)
@@ -411,7 +436,7 @@ def main() -> int:
         for error in errors:
             print(f"ERROR: {error}", file=sys.stderr)
         return 1
-    print("OK: e2e modules, runner selection, fuzz matrices, and profile guards are wired")
+    print("OK: e2e modules, runner selection, fuzz matrices, profiles, and bounded diagnostics are wired")
     return 0
 
 

scripts/s3-tests/run.sh

@@ -1028,10 +1028,11 @@ else
 fi
 
 # Run tests from s3tests/functional
+# Failure locals can contain multi-MiB request bodies; keep tracebacks without expanding local values.
 set +e
 S3TEST_CONF="${CONF_OUTPUT_PATH}" \
     tox -- \
-    -vv -ra --showlocals --tb=long \
+    -vv -ra --tb=long \
     --maxfail="${MAXFAIL}" \
     --timeout="${TEST_TIMEOUT}" \
     --junitxml="${ARTIFACTS_DIR}/junit.xml" \