fix(ecstore): retry decommission entries safely on source changes

A single object's SourceChanged during decommission cleanup no longer
cancels the shared worker token and fails the whole pool operation.
Cleanup preflight and source-cleanup outcomes now retry per entry with
bounded attempts and cancellation-aware backoff, applied uniformly to
ordinary versions, delete markers, and tiered copies (removing the
try-once-only branches); every retry re-lists the entry and redoes
version multiset validation before touching the source. Only quorum
loss, unrecoverable system errors, or exceeding a pool-level
SourceChanged exhaustion threshold still fails the decommission, and
exhausted entries never delete their source versions.

Retry attempts, backoff, and deferred/exhausted reasons are logged per
entry for observability. Heavy regression tests spawn on dedicated
32MiB stacks following the existing store-test pattern.

Fixes rustfs/backlog#1913

Cargo.lock

@@ -12688,6 +12688,7 @@ dependencies = [
  "js-sys",
  "rand 0.10.2",
  "serde_core",
+ "sha1_smol",
  "wasm-bindgen",
 ]
 

crates/ecstore/src/core/sets.rs

@@ -988,14 +988,11 @@ impl crate::storage_api_contracts::multipart::MultipartOperations for Sets {
     }
 }
 
-#[async_trait::async_trait]
-impl crate::storage_api_contracts::heal::HealOperations for Sets {
-    type Error = Error;
-    type HealResultItem = HealResultItem;
-    type HealOptions = HealOpts;
-
-    #[tracing::instrument(skip(self))]
-    async fn heal_format(&self, dry_run: bool) -> Result<(HealResultItem, Option<Error>)> {
+impl Sets {
+    pub(crate) async fn heal_format_with_fence<F>(&self, dry_run: bool, fence_lost: F) -> Result<(HealResultItem, Option<Error>)>
+    where
+        F: Fn() -> bool + Send + Sync,
+    {
         let (disks, init_errs) = init_storage_disks_with_errors(
             &self.endpoints.endpoints,
             &DiskOption {
@@ -1068,6 +1065,9 @@ impl crate::storage_api_contracts::heal::HealOperations for Sets {
             // Save new formats `format.json` on unformatted disks.
             for (index, (fm, disk)) in tmp_new_formats.iter_mut().zip(disks.iter()).enumerate() {
                 if fm.is_some() && disk.is_some() {
+                    if fence_lost() {
+                        return Ok((res, Some(StorageError::SlowDown)));
+                    }
                     if let Err(err) = save_format_file(disk, fm).await {
                         if let Some(disk) = disk.as_ref() {
                             let _ = disk.close().await;
@@ -1101,6 +1101,18 @@ impl crate::storage_api_contracts::heal::HealOperations for Sets {
         }
         Ok((res, None))
     }
+}
+
+#[async_trait::async_trait]
+impl crate::storage_api_contracts::heal::HealOperations for Sets {
+    type Error = Error;
+    type HealResultItem = HealResultItem;
+    type HealOptions = HealOpts;
+
+    #[tracing::instrument(skip(self))]
+    async fn heal_format(&self, dry_run: bool) -> Result<(HealResultItem, Option<Error>)> {
+        self.heal_format_with_fence(dry_run, || false).await
+    }
     #[tracing::instrument(skip(self))]
     async fn heal_bucket(&self, bucket: &str, opts: &HealOpts) -> Result<HealResultItem> {
         let mut result = HealResultItem {

crates/ecstore/src/set_disk/mod.rs

@@ -4736,7 +4736,15 @@ impl SetDisks {
                 achieved: 0,
             });
         }
-        let parts_metadata = vec![fi.clone(); disks.len()];
+        // Rebuilt tiered metadata starts with index zero, but shuffling validates
+        // each source slot before assigning the shuffled index below.
+        let parts_metadata: Vec<FileInfo> = (0..disks.len())
+            .map(|disk_index| {
+                let mut part = fi.clone();
+                part.erasure.index = fi.erasure.distribution[disk_index];
+                part
+            })
+            .collect();
         let (shuffle_disks, parts_metadata) = Self::shuffle_disks_and_parts_metadata(&disks, &parts_metadata, &fi);
 
         let mut errs = Vec::with_capacity(shuffle_disks.len());

crates/ecstore/src/store/heal.rs

@@ -13,7 +13,12 @@
 // limitations under the License.
 
 use super::*;
+use crate::core::pools::POOL_META_NAME;
+use crate::services::rebalance::{REBAL_META_NAME, RebalStatus};
+use crate::set_disk::get_lock_acquire_timeout;
 use crate::storage_api_contracts::heal::HealOperations as _;
+use crate::storage_api_contracts::namespace::NamespaceLocking as _;
+use rustfs_lock::NamespaceLockGuard;
 use tracing::trace;
 
 const LOG_COMPONENT_ECSTORE: &str = "ecstore";
@@ -30,7 +35,119 @@ fn invalid_heal_pool_index(pool_idx: usize, pool_count: usize) -> Error {
     )
 }
 
+#[derive(Debug, Clone, Copy)]
+enum HealFormatPoolSkip {
+    Completed,
+    Retryable,
+}
+
+fn classify_heal_format_pool(
+    pool_idx: usize,
+    pool_cmd_line: &str,
+    pool_meta: &PoolMeta,
+    rebalance_meta: Option<&RebalanceMeta>,
+) -> Option<HealFormatPoolSkip> {
+    let Some(pool) = pool_meta.pools.get(pool_idx) else {
+        return Some(HealFormatPoolSkip::Retryable);
+    };
+
+    if pool.id != pool_idx || pool_cmd_line.is_empty() || pool.cmd_line.is_empty() || pool.cmd_line != pool_cmd_line {
+        return Some(HealFormatPoolSkip::Retryable);
+    }
+
+    if let Some(decommission) = pool.decommission.as_ref() {
+        if decommission.complete {
+            return Some(HealFormatPoolSkip::Completed);
+        }
+        if decommission.failed || decommission.canceled || decommission.queued || pool_meta.is_suspended(pool_idx) {
+            return Some(HealFormatPoolSkip::Retryable);
+        }
+    }
+
+    if let Some(meta) = rebalance_meta {
+        let Some(pool_stats) = meta.pool_stats.get(pool_idx) else {
+            return Some(HealFormatPoolSkip::Retryable);
+        };
+        if pool_stats.info.stopping || (pool_stats.participating && pool_stats.info.status == RebalStatus::Started) {
+            return Some(HealFormatPoolSkip::Retryable);
+        }
+    }
+
+    None
+}
+
+fn heal_format_pool_skip_error(skip: HealFormatPoolSkip) -> Error {
+    match skip {
+        HealFormatPoolSkip::Completed => StorageError::NoHealRequired,
+        HealFormatPoolSkip::Retryable => StorageError::SlowDown,
+    }
+}
+
+fn heal_format_fence_lost_error() -> Error {
+    StorageError::SlowDown
+}
+
 impl ECStore {
+    async fn acquire_heal_format_fence(
+        &self,
+    ) -> Result<(NamespaceLockGuard, NamespaceLockGuard, PoolMeta, Option<RebalanceMeta>)> {
+        let metadata_pool = self
+            .pools
+            .first()
+            .cloned()
+            .ok_or_else(|| Error::other("heal format requires at least one storage pool"))?;
+
+        // Metadata fence order is part of the decommission/rebalance protocol:
+        // pool.bin must always be acquired before rebalance.bin.
+        let pool_lock = metadata_pool.new_ns_lock(RUSTFS_META_BUCKET, POOL_META_NAME).await?;
+        let pool_guard = pool_lock.get_write_lock(get_lock_acquire_timeout()).await?;
+        let rebalance_lock = metadata_pool.new_ns_lock(RUSTFS_META_BUCKET, REBAL_META_NAME).await?;
+        let rebalance_guard = rebalance_lock.get_write_lock(get_lock_acquire_timeout()).await?;
+
+        if pool_guard.is_lock_lost() || rebalance_guard.is_lock_lost() {
+            return Err(heal_format_fence_lost_error());
+        }
+
+        let mut pool_meta = PoolMeta::default();
+        pool_meta.load_no_lock(metadata_pool.clone()).await?;
+        if pool_meta.pools.len() != self.pools.len()
+            || pool_meta.pools.iter().enumerate().any(|(pool_idx, pool)| {
+                pool.id != pool_idx || pool.cmd_line.is_empty() || pool.cmd_line != self.pools[pool_idx].endpoints.cmd_line
+            })
+        {
+            return Err(heal_format_fence_lost_error());
+        }
+
+        let mut rebalance_meta = RebalanceMeta::new();
+        let rebalance_meta = match rebalance_meta
+            .load_with_opts(
+                metadata_pool,
+                ObjectOptions {
+                    no_lock: true,
+                    ..Default::default()
+                },
+            )
+            .await
+        {
+            Ok(()) => Some(rebalance_meta),
+            Err(Error::ConfigNotFound) => None,
+            Err(err) => return Err(err),
+        };
+
+        if rebalance_meta
+            .as_ref()
+            .is_some_and(|meta| meta.pool_stats.len() != self.pools.len())
+        {
+            return Err(heal_format_fence_lost_error());
+        }
+
+        if pool_guard.is_lock_lost() || rebalance_guard.is_lock_lost() {
+            return Err(heal_format_fence_lost_error());
+        }
+
+        Ok((pool_guard, rebalance_guard, pool_meta, rebalance_meta))
+    }
+
     fn get_pools_for_heal_object(&self, opts: &HealOpts) -> Result<Vec<Arc<Sets>>> {
         match opts.pool {
             Some(pool_idx) => Ok(vec![
@@ -52,9 +169,26 @@ impl ECStore {
         };
 
         let mut count_no_heal = 0;
+        let mut count_completed = 0;
         let mut first_error = None;
-        for pool in self.pools.iter() {
-            let (mut result, err) = pool.heal_format(dry_run).await?;
+        for (pool_idx, pool) in self.pools.iter().enumerate() {
+            let (pool_guard, rebalance_guard, pool_meta, rebalance_meta) = self.acquire_heal_format_fence().await?;
+            if pool_guard.is_lock_lost() || rebalance_guard.is_lock_lost() {
+                first_error.get_or_insert(heal_format_fence_lost_error());
+                break;
+            }
+            if let Some(skip) = classify_heal_format_pool(pool_idx, &pool.endpoints.cmd_line, &pool_meta, rebalance_meta.as_ref())
+            {
+                if matches!(skip, HealFormatPoolSkip::Completed) {
+                    count_completed += 1;
+                } else {
+                    first_error.get_or_insert(heal_format_pool_skip_error(skip));
+                }
+                continue;
+            }
+
+            let fence_lost = || pool_guard.is_lock_lost() || rebalance_guard.is_lock_lost();
+            let (mut result, err) = pool.heal_format_with_fence(dry_run, fence_lost).await?;
             if let Some(err) = err {
                 match err {
                     StorageError::NoHealRequired => {
@@ -69,11 +203,18 @@ impl ECStore {
             r.set_count += result.set_count;
             r.before.drives.append(&mut result.before.drives);
             r.after.drives.append(&mut result.after.drives);
+
+            // A lease can be lost after the final write; fail closed before
+            // reporting the pool as successfully healed.
+            if pool_guard.is_lock_lost() || rebalance_guard.is_lock_lost() {
+                first_error.get_or_insert(heal_format_fence_lost_error());
+                break;
+            }
         }
         if let Some(err) = first_error {
             return Ok((r, Some(err)));
         }
-        if count_no_heal == self.pools.len() {
+        if count_no_heal + count_completed == self.pools.len() {
             info!(
                 event = EVENT_HEAL_FORMAT_COMPLETED,
                 component = LOG_COMPONENT_ECSTORE,
@@ -302,6 +443,7 @@ mod tests {
     use crate::disk::{DeleteOptions, DiskOption, format::FormatV3, new_disk};
     use crate::layout::endpoints::{EndpointServerPools, Endpoints, PoolEndpoints};
     use crate::runtime::instance::InstanceContext;
+    use crate::services::rebalance::{RebalanceInfo, RebalanceStats};
     use crate::storage_api_contracts::bucket::{BucketOperations, MakeBucketOptions};
     use crate::storage_api_contracts::object::{ObjectIO as _, ObjectOperations};
     use crate::store::init_format::{load_format_erasure, save_format_file};
@@ -353,6 +495,164 @@ mod tests {
         }
     }
 
+    fn pool_meta_with_decommission(info: PoolDecommissionInfo) -> PoolMeta {
+        PoolMeta {
+            pools: vec![PoolStatus {
+                id: 0,
+                cmd_line: "pool-0".to_string(),
+                last_update: OffsetDateTime::UNIX_EPOCH,
+                decommission: Some(info),
+            }],
+            ..Default::default()
+        }
+    }
+
+    #[test]
+    fn heal_format_pool_state_barriers_are_classified() {
+        let active = pool_meta_with_decommission(PoolDecommissionInfo {
+            start_time: Some(OffsetDateTime::UNIX_EPOCH),
+            ..Default::default()
+        });
+        assert!(matches!(
+            classify_heal_format_pool(0, "pool-0", &active, None),
+            Some(HealFormatPoolSkip::Retryable)
+        ));
+
+        for info in [
+            PoolDecommissionInfo {
+                failed: true,
+                ..Default::default()
+            },
+            PoolDecommissionInfo {
+                canceled: true,
+                ..Default::default()
+            },
+        ] {
+            assert!(matches!(
+                classify_heal_format_pool(0, "pool-0", &pool_meta_with_decommission(info), None),
+                Some(HealFormatPoolSkip::Retryable)
+            ));
+        }
+
+        let completed = pool_meta_with_decommission(PoolDecommissionInfo {
+            complete: true,
+            ..Default::default()
+        });
+        assert!(matches!(
+            classify_heal_format_pool(0, "pool-0", &completed, None),
+            Some(HealFormatPoolSkip::Completed)
+        ));
+    }
+
+    #[test]
+    fn heal_format_pool_rebalance_barriers_and_identity_are_fail_closed() {
+        let identity_meta = pool_meta_with_decommission(PoolDecommissionInfo::default());
+        let rebalance = RebalanceMeta {
+            pool_stats: vec![RebalanceStats {
+                participating: true,
+                info: RebalanceInfo {
+                    status: RebalStatus::Started,
+                    ..Default::default()
+                },
+                ..Default::default()
+            }],
+            ..Default::default()
+        };
+        assert!(matches!(
+            classify_heal_format_pool(0, "pool-0", &identity_meta, Some(&rebalance)),
+            Some(HealFormatPoolSkip::Retryable)
+        ));
+
+        let stopping = RebalanceMeta {
+            pool_stats: vec![RebalanceStats {
+                info: RebalanceInfo {
+                    stopping: true,
+                    ..Default::default()
+                },
+                ..Default::default()
+            }],
+            ..Default::default()
+        };
+        assert!(matches!(
+            classify_heal_format_pool(0, "pool-0", &identity_meta, Some(&stopping)),
+            Some(HealFormatPoolSkip::Retryable)
+        ));
+
+        let identity = pool_meta_with_decommission(PoolDecommissionInfo::default());
+        assert!(matches!(
+            classify_heal_format_pool(0, "pool-new", &identity, None),
+            Some(HealFormatPoolSkip::Retryable)
+        ));
+
+        let identity_without_decommission = PoolMeta {
+            pools: vec![PoolStatus {
+                id: 0,
+                cmd_line: "pool-0".to_string(),
+                last_update: OffsetDateTime::UNIX_EPOCH,
+                decommission: None,
+            }],
+            ..Default::default()
+        };
+        assert!(matches!(
+            classify_heal_format_pool(0, "pool-new", &identity_without_decommission, None),
+            Some(HealFormatPoolSkip::Retryable)
+        ));
+
+        assert!(matches!(
+            classify_heal_format_pool(0, "", &identity_meta, None),
+            Some(HealFormatPoolSkip::Retryable)
+        ));
+
+        assert!(matches!(
+            classify_heal_format_pool(0, "pool-0", &PoolMeta::default(), None),
+            Some(HealFormatPoolSkip::Retryable)
+        ));
+
+        let stopped = RebalanceMeta {
+            stopped_at: Some(OffsetDateTime::UNIX_EPOCH),
+            pool_stats: vec![RebalanceStats {
+                participating: true,
+                info: RebalanceInfo {
+                    status: RebalStatus::Stopped,
+                    ..Default::default()
+                },
+                ..Default::default()
+            }],
+            ..Default::default()
+        };
+        assert!(classify_heal_format_pool(0, "pool-0", &identity_meta, Some(&stopped)).is_none());
+
+        let stopping_after_stop = RebalanceMeta {
+            stopped_at: Some(OffsetDateTime::UNIX_EPOCH),
+            pool_stats: vec![RebalanceStats {
+                participating: true,
+                info: RebalanceInfo {
+                    status: RebalStatus::Started,
+                    stopping: true,
+                    ..Default::default()
+                },
+                ..Default::default()
+            }],
+            ..Default::default()
+        };
+        assert!(matches!(
+            classify_heal_format_pool(0, "pool-0", &identity_meta, Some(&stopping_after_stop)),
+            Some(HealFormatPoolSkip::Retryable)
+        ));
+    }
+
+    #[test]
+    fn skipped_heal_format_pool_is_never_reported_as_success() {
+        assert!(matches!(
+            heal_format_pool_skip_error(HealFormatPoolSkip::Retryable),
+            StorageError::SlowDown
+        ));
+        assert!(matches!(
+            heal_format_pool_skip_error(HealFormatPoolSkip::Completed),
+            StorageError::NoHealRequired
+        ));
+    }
+
     async fn multi_pool_heal_store() -> (tempfile::TempDir, Arc<ECStore>, CancellationToken) {
         let temp_dir = tempfile::tempdir().expect("multi-pool heal test directory should be created");
         let mut pool_endpoints = Vec::new();
@@ -889,6 +1189,18 @@ mod tests {
             bucket_fence_registry: std::sync::Arc::default(),
         };
 
+        let err = store
+            .handle_heal_format(false)
+            .await
+            .expect_err("missing pool metadata must fail closed before format writes");
+        assert!(matches!(err, StorageError::SlowDown));
+
+        let pool_meta = PoolMeta::new(&store.pools, &PoolMeta::default());
+        pool_meta
+            .save(store.pools.clone())
+            .await
+            .expect("pool metadata should be persisted before format heal");
+
         let (result, err) = store
             .handle_heal_format(false)
             .await
@@ -902,5 +1214,22 @@ mod tests {
             .await
             .expect("the later pool should be healed despite the first pool error");
         assert_eq!(healed.erasure.this, recoverable_format.erasure.sets[0][2]);
+
+        let mut completed_meta = PoolMeta::new(&store.pools, &PoolMeta::default());
+        for status in &mut completed_meta.pools {
+            status.decommission = Some(PoolDecommissionInfo {
+                complete: true,
+                ..Default::default()
+            });
+        }
+        completed_meta
+            .save(store.pools.clone())
+            .await
+            .expect("completed pool metadata should be persisted");
+        let (_, err) = store
+            .handle_heal_format(false)
+            .await
+            .expect("completed pools should be reported as a no-op");
+        assert!(matches!(err, Some(StorageError::NoHealRequired)));
     }
 }

crates/ecstore/src/store/init.rs

@@ -626,7 +626,7 @@ mod tests {
         io::Cursor,
         sync::{
             Arc,
-            atomic::{AtomicBool, Ordering},
+            atomic::{AtomicBool, AtomicUsize, Ordering},
         },
         time::Duration,
     };
@@ -1307,6 +1307,85 @@ mod tests {
         });
     }
 
+    const DECOMMISSION_TEST_FAULT_STAGE_DELETE_MARKER: &str = "delete_marker_copy";
+    const DECOMMISSION_TEST_FAULT_STAGE_MIGRATE_OBJECT: &str = "migrate_object";
+    #[cfg(feature = "test-util")]
+    const DECOMMISSION_TEST_FAULT_STAGE_TIERED: &str = "decommission_tiered_object";
+
+    async fn seed_decommission_source(
+        store: &Arc<crate::store::ECStore>,
+        bucket: &str,
+        object: &str,
+        body: Vec<u8>,
+        opts: &ObjectOptions,
+    ) {
+        let mut reader = PutObjReader::from_vec(body);
+        store.pools[0]
+            .put_object(bucket, object, &mut reader, opts)
+            .await
+            .expect("seed decommission source object");
+    }
+
+    async fn run_decommission_entry_retry_test(
+        store: &Arc<crate::store::ECStore>,
+        rx: CancellationToken,
+        bucket: &str,
+        object: &str,
+        expected_bucket_incarnation_id: Option<uuid::Uuid>,
+        source_changed_exhaustions: Arc<AtomicUsize>,
+    ) -> crate::error::Result<()> {
+        let source_set = store.pools[0].get_disks_by_key(object);
+        store
+            .decommission_entry_with_retry_state_for_test(
+                rx,
+                0,
+                MetaCacheEntry {
+                    name: object.to_string(),
+                    ..Default::default()
+                },
+                bucket.to_string(),
+                source_set,
+                expected_bucket_incarnation_id,
+                source_changed_exhaustions,
+            )
+            .await
+    }
+
+    async fn read_decommission_target_body(
+        store: &Arc<crate::store::ECStore>,
+        bucket: &str,
+        object: &str,
+        opts: &ObjectOptions,
+    ) -> Vec<u8> {
+        let mut reader = store.pools[1]
+            .get_object_reader(bucket, object, None, HeaderMap::new(), opts)
+            .await
+            .expect("read decommission target object");
+        let mut body = Vec::new();
+        reader
+            .stream
+            .read_to_end(&mut body)
+            .await
+            .expect("drain decommission target body");
+        body
+    }
+
+    async fn assert_decommission_source_absent(
+        store: &Arc<crate::store::ECStore>,
+        bucket: &str,
+        object: &str,
+        opts: &ObjectOptions,
+    ) {
+        let err = store.pools[0]
+            .get_object_info(bucket, object, opts)
+            .await
+            .expect_err("decommission source must be retained until target commit, then removed");
+        assert!(
+            matches!(err, StorageError::ObjectNotFound(_, _) | StorageError::VersionNotFound(_, _, _)),
+            "unexpected decommission source result: {err:?}"
+        );
+    }
+
     async fn write_decommission_test_multipart_source(
         store: &Arc<crate::store::ECStore>,
         pool_idx: usize,
@@ -3102,6 +3181,537 @@ mod tests {
         shutdown.cancel();
     }
 
+    #[test]
+    #[serial_test::serial(storage_class_env)]
+    fn decommission_entry_retries_source_changed_without_canceling_other_bucket() {
+        let handle = std::thread::Builder::new()
+            .name("decommission_entry_retries_source_changed_without_canceling_other_bucket".to_string())
+            .stack_size(32 * 1024 * 1024)
+            .spawn(|| {
+                let runtime = tokio::runtime::Builder::new_multi_thread()
+                    .enable_all()
+                    .worker_threads(2)
+                    .build()
+                    .expect("test runtime should build");
+                runtime.block_on(async {
+                    let temp_dir = tempfile::tempdir().expect("create decommission retry store dir");
+                    let (_ctx, store, shutdown) = without_storage_class_env(build_isolated_test_store(
+                        temp_dir.path(),
+                        "decommission-entry-retry",
+                        &[4, 4],
+                    ))
+                    .await;
+                    crate::bucket::metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+
+                    let changed_bucket = format!("decom-retry-a-{}", uuid::Uuid::new_v4());
+                    let other_bucket = format!("decom-retry-b-{}", uuid::Uuid::new_v4());
+                    for bucket in [&changed_bucket, &other_bucket] {
+                        store
+                            .make_bucket(bucket, &MakeBucketOptions::default())
+                            .await
+                            .expect("create decommission retry bucket");
+                    }
+
+                    let changed_object = "changed.bin";
+                    let first_version = uuid::Uuid::new_v4();
+                    let second_version = uuid::Uuid::new_v4();
+                    let base_time = OffsetDateTime::now_utc();
+                    seed_decommission_source(
+                        &store,
+                        &changed_bucket,
+                        changed_object,
+                        b"first generation".to_vec(),
+                        &ObjectOptions {
+                            versioned: true,
+                            version_id: Some(first_version.to_string()),
+                            mod_time: Some(base_time),
+                            ..Default::default()
+                        },
+                    )
+                    .await;
+                    let other_object = "other.bin";
+                    seed_decommission_source(
+                        &store,
+                        &other_bucket,
+                        other_object,
+                        b"other bucket generation".to_vec(),
+                        &ObjectOptions::default(),
+                    )
+                    .await;
+                    mark_test_pool_decommissioning(&store, 0).await;
+
+                    let mutation_calls = Arc::new(AtomicUsize::new(0));
+                    let mutation_calls_for_hook = Arc::clone(&mutation_calls);
+                    let mutation_store = Arc::clone(&store);
+                    let mutation_bucket = changed_bucket.clone();
+                    let _mutation_guard = crate::core::pools::DecommissionCleanupMutationGuard::install(Arc::new(
+                        move |bucket, object, attempt| {
+                            let is_target = bucket == mutation_bucket.as_str() && object == changed_object;
+                            let calls = Arc::clone(&mutation_calls_for_hook);
+                            let store = Arc::clone(&mutation_store);
+                            let bucket = mutation_bucket.clone();
+                            Box::pin(async move {
+                                if !is_target {
+                                    return;
+                                }
+                                calls.fetch_add(1, Ordering::SeqCst);
+                                if attempt == 1 {
+                                    seed_decommission_source(
+                                        &store,
+                                        &bucket,
+                                        changed_object,
+                                        b"second generation".to_vec(),
+                                        &ObjectOptions {
+                                            versioned: true,
+                                            version_id: Some(second_version.to_string()),
+                                            mod_time: Some(base_time + time::Duration::seconds(1)),
+                                            ..Default::default()
+                                        },
+                                    )
+                                    .await;
+                                }
+                            })
+                        },
+                    ));
+
+                    let ordinary_faults = Arc::new(AtomicUsize::new(0));
+                    let ordinary_faults_for_hook = Arc::clone(&ordinary_faults);
+                    let fault_bucket = other_bucket.clone();
+                    let _fault_guard = crate::core::pools::DecommissionTestFaultGuard::install(Arc::new(
+                        move |stage, bucket, object, attempt| {
+                            let injected = stage == DECOMMISSION_TEST_FAULT_STAGE_MIGRATE_OBJECT
+                                && bucket == fault_bucket.as_str()
+                                && object == other_object
+                                && attempt < crate::core::pools::DECOMMISSION_VERSION_COPY_ATTEMPTS;
+                            if injected {
+                                ordinary_faults_for_hook.fetch_add(1, Ordering::SeqCst);
+                            }
+                            injected
+                        },
+                    ));
+
+                    let rx = CancellationToken::new();
+                    let source_changed_exhaustions = Arc::new(AtomicUsize::new(0));
+                    let changed_incarnation = Some(
+                        store
+                            .bucket_incarnation_id(&changed_bucket)
+                            .await
+                            .expect("changed bucket incarnation"),
+                    );
+                    let other_incarnation = Some(
+                        store
+                            .bucket_incarnation_id(&other_bucket)
+                            .await
+                            .expect("other bucket incarnation"),
+                    );
+                    let (changed_result, other_result) = tokio::join!(
+                        run_decommission_entry_retry_test(
+                            &store,
+                            rx.clone(),
+                            &changed_bucket,
+                            changed_object,
+                            changed_incarnation,
+                            Arc::clone(&source_changed_exhaustions),
+                        ),
+                        run_decommission_entry_retry_test(
+                            &store,
+                            rx.clone(),
+                            &other_bucket,
+                            other_object,
+                            other_incarnation,
+                            Arc::clone(&source_changed_exhaustions),
+                        )
+                    );
+                    changed_result.expect("SourceChanged entry retry must converge");
+                    other_result.expect("other bucket entry must continue through ordinary copy retries");
+
+                    assert!(!rx.is_cancelled(), "entry-level SourceChanged must not cancel the shared worker token");
+                    assert_eq!(mutation_calls.load(Ordering::SeqCst), 2, "entry must be re-listed after SourceChanged");
+                    assert_eq!(ordinary_faults.load(Ordering::SeqCst), 2, "ordinary copy must consume the retry budget");
+                    assert_eq!(source_changed_exhaustions.load(Ordering::SeqCst), 0);
+
+                    for (version_id, expected_body) in [
+                        (first_version, b"first generation".as_slice()),
+                        (second_version, b"second generation".as_slice()),
+                    ] {
+                        let opts = ObjectOptions {
+                            versioned: true,
+                            version_id: Some(version_id.to_string()),
+                            ..Default::default()
+                        };
+                        assert_decommission_source_absent(&store, &changed_bucket, changed_object, &opts).await;
+                        assert_eq!(
+                            read_decommission_target_body(&store, &changed_bucket, changed_object, &opts).await,
+                            expected_body
+                        );
+                    }
+                    assert_decommission_source_absent(&store, &other_bucket, other_object, &ObjectOptions::default()).await;
+                    assert_eq!(
+                        read_decommission_target_body(&store, &other_bucket, other_object, &ObjectOptions::default()).await,
+                        b"other bucket generation"
+                    );
+
+                    shutdown.cancel();
+                });
+            })
+            .expect("spawn decommission retry test thread");
+        if let Err(payload) = handle.join() {
+            std::panic::resume_unwind(payload);
+        }
+    }
+
+    #[test]
+    #[serial_test::serial(storage_class_env)]
+    fn decommission_entry_exhausted_source_changed_retains_source_and_records_failure() {
+        let handle = std::thread::Builder::new()
+            .name("decommission_entry_exhausted_source_changed_retains_source_and_records_failure".to_string())
+            .stack_size(32 * 1024 * 1024)
+            .spawn(|| {
+                let runtime = tokio::runtime::Builder::new_multi_thread()
+                    .enable_all()
+                    .worker_threads(2)
+                    .build()
+                    .expect("test runtime should build");
+                runtime.block_on(async {
+                    let temp_dir = tempfile::tempdir().expect("create decommission exhaustion store dir");
+                    let (_ctx, store, shutdown) = without_storage_class_env(build_isolated_test_store(
+                        temp_dir.path(),
+                        "decommission-entry-exhaustion",
+                        &[4, 4],
+                    ))
+                    .await;
+                    crate::bucket::metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+
+                    let bucket = format!("decom-exhausted-{}", uuid::Uuid::new_v4());
+                    let object = "exhausted.bin";
+                    store
+                        .make_bucket(&bucket, &MakeBucketOptions::default())
+                        .await
+                        .expect("create decommission exhaustion bucket");
+                    let original_version = uuid::Uuid::new_v4();
+                    let base_time = OffsetDateTime::now_utc();
+                    seed_decommission_source(
+                        &store,
+                        &bucket,
+                        object,
+                        b"original generation".to_vec(),
+                        &ObjectOptions {
+                            versioned: true,
+                            version_id: Some(original_version.to_string()),
+                            mod_time: Some(base_time),
+                            ..Default::default()
+                        },
+                    )
+                    .await;
+                    mark_test_pool_decommissioning(&store, 0).await;
+
+                    let mutation_calls = Arc::new(AtomicUsize::new(0));
+                    let mutation_calls_for_hook = Arc::clone(&mutation_calls);
+                    let mutation_store = Arc::clone(&store);
+                    let mutation_bucket = bucket.clone();
+                    let _mutation_guard = crate::core::pools::DecommissionCleanupMutationGuard::install(Arc::new(
+                        move |called_bucket, called_object, _attempt| {
+                            let is_target = called_bucket == mutation_bucket.as_str() && called_object == object;
+                            let calls = Arc::clone(&mutation_calls_for_hook);
+                            let store = Arc::clone(&mutation_store);
+                            let bucket = mutation_bucket.clone();
+                            Box::pin(async move {
+                                if !is_target {
+                                    return;
+                                }
+                                let call = calls.fetch_add(1, Ordering::SeqCst) + 1;
+                                let offset = i64::try_from(call).expect("entry retry count should fit i64");
+                                seed_decommission_source(
+                                    &store,
+                                    &bucket,
+                                    object,
+                                    format!("concurrent generation {call}").into_bytes(),
+                                    &ObjectOptions {
+                                        versioned: true,
+                                        version_id: Some(uuid::Uuid::new_v4().to_string()),
+                                        mod_time: Some(base_time + time::Duration::seconds(offset)),
+                                        ..Default::default()
+                                    },
+                                )
+                                .await;
+                            })
+                        },
+                    ));
+
+                    let rx = CancellationToken::new();
+                    let source_changed_exhaustions = Arc::new(AtomicUsize::new(0));
+                    let incarnation = Some(store.bucket_incarnation_id(&bucket).await.expect("bucket incarnation"));
+                    run_decommission_entry_retry_test(
+                        &store,
+                        rx.clone(),
+                        &bucket,
+                        object,
+                        incarnation,
+                        Arc::clone(&source_changed_exhaustions),
+                    )
+                    .await
+                    .expect("entry-level exhaustion must stay local below the pool threshold");
+
+                    assert!(!rx.is_cancelled(), "one exhausted entry must not cancel other bucket workers");
+                    assert_eq!(mutation_calls.load(Ordering::SeqCst), crate::core::pools::DECOMMISSION_ENTRY_MAX_ATTEMPTS);
+                    assert_eq!(source_changed_exhaustions.load(Ordering::SeqCst), 1);
+                    store.pools[0]
+                        .get_object_info(
+                            &bucket,
+                            object,
+                            &ObjectOptions {
+                                versioned: true,
+                                version_id: Some(original_version.to_string()),
+                                ..Default::default()
+                            },
+                        )
+                        .await
+                        .expect("retry exhaustion must retain the original source version");
+                    let pool_meta = store.pool_meta.read().await;
+                    let info = pool_meta.pools[0]
+                        .decommission
+                        .as_ref()
+                        .expect("decommission progress must be initialized");
+                    assert_eq!(info.items_decommission_failed, 1, "exhausted entry must be visible as failed");
+                    drop(pool_meta);
+
+                    shutdown.cancel();
+                });
+            })
+            .expect("spawn decommission retry test thread");
+        if let Err(payload) = handle.join() {
+            std::panic::resume_unwind(payload);
+        }
+    }
+
+    #[test]
+    #[serial_test::serial(storage_class_env)]
+    fn decommission_entry_delete_marker_copy_retries_real_path() {
+        let handle = std::thread::Builder::new()
+            .name("decommission_entry_delete_marker_copy_retries_real_path".to_string())
+            .stack_size(32 * 1024 * 1024)
+            .spawn(|| {
+                let runtime = tokio::runtime::Builder::new_multi_thread()
+                    .enable_all()
+                    .worker_threads(2)
+                    .build()
+                    .expect("test runtime should build");
+                runtime.block_on(async {
+                    let temp_dir = tempfile::tempdir().expect("create delete marker retry store dir");
+                    let (_ctx, store, shutdown) = without_storage_class_env(build_isolated_test_store(
+                        temp_dir.path(),
+                        "decommission-delete-marker-retry",
+                        &[4, 4],
+                    ))
+                    .await;
+                    crate::bucket::metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+
+                    let bucket = format!("decom-marker-{}", uuid::Uuid::new_v4());
+                    let object = "marker.bin";
+                    store
+                        .make_bucket(&bucket, &MakeBucketOptions::default())
+                        .await
+                        .expect("create delete marker retry bucket");
+                    let data_version = uuid::Uuid::new_v4();
+                    let marker_version = uuid::Uuid::new_v4();
+                    let base_time = OffsetDateTime::now_utc();
+                    seed_decommission_source(
+                        &store,
+                        &bucket,
+                        object,
+                        b"delete marker data".to_vec(),
+                        &ObjectOptions {
+                            versioned: true,
+                            version_id: Some(data_version.to_string()),
+                            mod_time: Some(base_time),
+                            ..Default::default()
+                        },
+                    )
+                    .await;
+                    store.pools[0]
+                        .delete_object(
+                            &bucket,
+                            object,
+                            ObjectOptions {
+                                versioned: true,
+                                version_id: Some(marker_version.to_string()),
+                                delete_marker: true,
+                                mod_time: Some(base_time + time::Duration::seconds(1)),
+                                ..Default::default()
+                            },
+                        )
+                        .await
+                        .expect("seed source delete marker");
+                    mark_test_pool_decommissioning(&store, 0).await;
+
+                    let fault_calls = Arc::new(AtomicUsize::new(0));
+                    let fault_calls_for_hook = Arc::clone(&fault_calls);
+                    let fault_bucket = bucket.clone();
+                    let _fault_guard = crate::core::pools::DecommissionTestFaultGuard::install(Arc::new(
+                        move |stage, called_bucket, called_object, attempt| {
+                            let injected = stage == DECOMMISSION_TEST_FAULT_STAGE_DELETE_MARKER
+                                && called_bucket == fault_bucket.as_str()
+                                && called_object == object
+                                && attempt < crate::core::pools::DECOMMISSION_VERSION_COPY_ATTEMPTS;
+                            if injected {
+                                fault_calls_for_hook.fetch_add(1, Ordering::SeqCst);
+                            }
+                            injected
+                        },
+                    ));
+
+                    let incarnation = Some(store.bucket_incarnation_id(&bucket).await.expect("bucket incarnation"));
+                    run_decommission_entry_retry_test(
+                        &store,
+                        CancellationToken::new(),
+                        &bucket,
+                        object,
+                        incarnation,
+                        Arc::new(AtomicUsize::new(0)),
+                    )
+                    .await
+                    .expect("delete marker copy retries must converge");
+
+                    assert_eq!(
+                        fault_calls.load(Ordering::SeqCst),
+                        crate::core::pools::DECOMMISSION_VERSION_COPY_ATTEMPTS - 1
+                    );
+                    let marker_opts = ObjectOptions {
+                        versioned: true,
+                        version_id: Some(marker_version.to_string()),
+                        ..Default::default()
+                    };
+                    let target_marker = store.pools[1]
+                        .get_object_info(&bucket, object, &marker_opts)
+                        .await
+                        .expect("target delete marker must exist");
+                    assert!(target_marker.delete_marker);
+                    assert_decommission_source_absent(&store, &bucket, object, &marker_opts).await;
+
+                    let data_opts = ObjectOptions {
+                        versioned: true,
+                        version_id: Some(data_version.to_string()),
+                        ..Default::default()
+                    };
+                    assert_eq!(
+                        read_decommission_target_body(&store, &bucket, object, &data_opts).await,
+                        b"delete marker data"
+                    );
+                    assert_decommission_source_absent(&store, &bucket, object, &data_opts).await;
+
+                    shutdown.cancel();
+                });
+            })
+            .expect("spawn decommission retry test thread");
+        if let Err(payload) = handle.join() {
+            std::panic::resume_unwind(payload);
+        }
+    }
+
+    #[cfg(feature = "test-util")]
+    #[test]
+    #[serial_test::serial(storage_class_env)]
+    fn decommission_entry_tiered_copy_retries_real_path() {
+        let handle = std::thread::Builder::new()
+            .name("decommission_entry_tiered_copy_retries_real_path".to_string())
+            .stack_size(32 * 1024 * 1024)
+            .spawn(|| {
+                let runtime = tokio::runtime::Builder::new_multi_thread()
+                    .enable_all()
+                    .worker_threads(2)
+                    .build()
+                    .expect("test runtime should build");
+                runtime.block_on(async {
+                    let temp_dir = tempfile::tempdir().expect("create tiered retry store dir");
+                    let (ctx, store, shutdown) = without_storage_class_env(build_isolated_test_store(
+                        temp_dir.path(),
+                        "decommission-tiered-retry",
+                        &[4, 4],
+                    ))
+                    .await;
+                    crate::bucket::metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+
+                    let bucket = format!("decom-tiered-{}", uuid::Uuid::new_v4());
+                    let object = "tiered.bin";
+                    store
+                        .make_bucket(&bucket, &MakeBucketOptions::default())
+                        .await
+                        .expect("create tiered retry bucket");
+                    let mut reader = PutObjReader::from_vec(b"tiered generation".to_vec());
+                    let original = store.pools[0]
+                        .put_object(&bucket, object, &mut reader, &ObjectOptions::default())
+                        .await
+                        .expect("seed tiered source object");
+                    let tier_name = format!("DECOM{}", &uuid::Uuid::new_v4().simple().to_string()[..8]).to_uppercase();
+                    register_mock_tier(&ctx.tier_config_mgr(), &tier_name).await;
+                    store.pools[0]
+                        .transition_object(
+                            &bucket,
+                            object,
+                            &ObjectOptions {
+                                transition: TransitionOptions {
+                                    status: TRANSITION_PENDING.to_string(),
+                                    tier: tier_name,
+                                    etag: original.etag.clone().expect("tiered source ETag"),
+                                    ..Default::default()
+                                },
+                                version_id: original.version_id.map(|version_id| version_id.to_string()),
+                                mod_time: original.mod_time,
+                                ..Default::default()
+                            },
+                        )
+                        .await
+                        .expect("transition source object to mock tier");
+                    mark_test_pool_decommissioning(&store, 0).await;
+
+                    let fault_calls = Arc::new(AtomicUsize::new(0));
+                    let fault_calls_for_hook = Arc::clone(&fault_calls);
+                    let fault_bucket = bucket.clone();
+                    let _fault_guard = crate::core::pools::DecommissionTestFaultGuard::install(Arc::new(
+                        move |stage, called_bucket, called_object, attempt| {
+                            let injected = stage == DECOMMISSION_TEST_FAULT_STAGE_TIERED
+                                && called_bucket == fault_bucket.as_str()
+                                && called_object == object
+                                && attempt < crate::core::pools::DECOMMISSION_VERSION_COPY_ATTEMPTS;
+                            if injected {
+                                fault_calls_for_hook.fetch_add(1, Ordering::SeqCst);
+                            }
+                            injected
+                        },
+                    ));
+
+                    let incarnation = Some(store.bucket_incarnation_id(&bucket).await.expect("bucket incarnation"));
+                    run_decommission_entry_retry_test(
+                        &store,
+                        CancellationToken::new(),
+                        &bucket,
+                        object,
+                        incarnation,
+                        Arc::new(AtomicUsize::new(0)),
+                    )
+                    .await
+                    .expect("tiered copy retries must converge");
+
+                    assert_eq!(
+                        fault_calls.load(Ordering::SeqCst),
+                        crate::core::pools::DECOMMISSION_VERSION_COPY_ATTEMPTS - 1
+                    );
+                    let target = store.pools[1]
+                        .get_object_info(&bucket, object, &ObjectOptions::default())
+                        .await
+                        .expect("tiered target metadata must exist");
+                    assert_eq!(target.transitioned_object.status, rustfs_filemeta::TRANSITION_COMPLETE);
+                    assert_decommission_source_absent(&store, &bucket, object, &ObjectOptions::default()).await;
+
+                    shutdown.cancel();
+                });
+            })
+            .expect("spawn decommission retry test thread");
+        if let Err(payload) = handle.join() {
+            std::panic::resume_unwind(payload);
+        }
+    }
+
     #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
     #[serial_test::serial(storage_class_env)]
     async fn decommission_outer_fence_loss_blocks_target_put_commit() {

crates/heal/src/heal/task/heal_erasure_set.rs

@@ -231,6 +231,10 @@ impl HealTask {
                             "Heal erasure set format repair skipped because no format heal was required"
                         );
                     } else {
+                        let error = e;
+                        if error.is_recoverable_heal() {
+                            return Err(error);
+                        }
                         error!(
                             target: "rustfs::heal::task",
                             event = EVENT_HEAL_ERASURE_SET_RESULT,
@@ -239,7 +243,7 @@ impl HealTask {
                             task_id = %self.id,
                             set_disk_id,
                             result = "format_failed",
-                            error = %e,
+                            error = %error,
                             "Heal erasure set failed"
                         );
                         {
@@ -247,7 +251,7 @@ impl HealTask {
                             progress.update_progress(4, 4, 0, 0);
                         }
                         return Err(Error::TaskExecutionFailed {
-                            message: format!("Failed to heal disk format for {set_disk_id}: {e}"),
+                            message: format!("Failed to heal disk format for {set_disk_id}: {error}"),
                         });
                     }
                 } else {
@@ -284,6 +288,9 @@ impl HealTask {
             Err(Error::TaskCancelled) => return Err(Error::TaskCancelled),
             Err(Error::TaskTimeout) => return Err(Error::TaskTimeout),
             Err(e) => {
+                if e.is_recoverable_heal() {
+                    return Err(e);
+                }
                 error!(
                     target: "rustfs::heal::task",
                     event = EVENT_HEAL_ERASURE_SET_RESULT,

crates/heal/src/heal/task/tests.rs

@@ -547,6 +547,7 @@ struct MockStorage {
     heal_object_outcome: Mutex<Option<MockHealObjectOutcome>>,
     heal_object_outcomes: Mutex<HashMap<String, VecDeque<MockHealObjectOutcome>>>,
     format_no_heal_required: Mutex<bool>,
+    format_error: Mutex<Option<Error>>,
     global_format_calls: Mutex<u32>,
     replacement_format_calls: Mutex<Vec<(usize, usize, Vec<String>)>>,
     replacement_targets_ready: Mutex<bool>,
@@ -867,6 +868,9 @@ impl HealStorageAPI for MockStorage {
 
     async fn heal_format(&self, _dry_run: bool) -> Result<(HealResultItem, Option<Error>)> {
         *self.global_format_calls.lock().unwrap() += 1;
+        if let Some(error) = self.format_error.lock().unwrap().take() {
+            return Err(error);
+        }
         let no_heal_required = *self.format_no_heal_required.lock().unwrap();
         if no_heal_required {
             Ok((HealResultItem::default(), Some(Error::Storage(EcstoreError::NoHealRequired))))
@@ -2052,6 +2056,30 @@ async fn test_erasure_set_heal_continues_after_format_no_heal_required() {
     );
 }
 
+#[tokio::test]
+async fn erasure_set_format_slowdown_is_propagated() {
+    let storage = Arc::new(MockStorage {
+        format_error: Mutex::new(Some(Error::Storage(EcstoreError::SlowDown))),
+        ..Default::default()
+    });
+    let request = HealRequest::new(
+        HealType::ErasureSet {
+            buckets: Vec::new(),
+            set_disk_id: "pool_0_set_0".to_string(),
+        },
+        HealOptions::default(),
+        HealPriority::Normal,
+    );
+    let task = HealTask::from_request(request, storage);
+
+    let error = task
+        .execute()
+        .await
+        .expect_err("format SlowDown must remain recoverable for the task manager");
+
+    assert!(matches!(error, Error::Storage(EcstoreError::SlowDown)));
+}
+
 #[tokio::test]
 async fn erasure_set_bucket_prepass_failure_stops_before_object_heal() {
     let temp = TempDir::new().expect("temporary directory should be created");

crates/test-utils/src/lib.rs

@@ -245,6 +245,18 @@ impl TestECStoreEnvBuilder {
             .await
             .expect("build test ECStore");
 
+        // The production bootstrap only persists pool.bin from the elected
+        // first cluster node.  Test stores intentionally have no cluster
+        // election, but heal-format still requires that durable fence before
+        // it can write any disk format.  Materialize the validated topology
+        // here so the shared fixture models a ready single-node store.
+        let mut pool_meta = ecstore.pool_meta.read().await.clone();
+        pool_meta.dont_save = false;
+        pool_meta
+            .save(ecstore.pools.clone())
+            .await
+            .expect("persist test pool metadata");
+
         if self.init_bucket_metadata {
             let buckets_list = ecstore
                 .list_bucket(&BucketOptions {

rustfs/Cargo.toml

@@ -322,7 +322,7 @@ thiserror = { workspace = true }
 tracing.workspace = true
 url = { workspace = true }
 urlencoding = { workspace = true }
-uuid = { workspace = true, features = ["v4", "fast-rng", "macro-diagnostics"] }
+uuid = { workspace = true, features = ["v4", "v5", "fast-rng", "macro-diagnostics"] }
 zip = { workspace = true }
 libc = { workspace = true }
 rand = { workspace = true, features = ["serde"] }
@@ -345,7 +345,7 @@ libsystemd.workspace = true
 libmimalloc-sys.workspace = true
 
 [dev-dependencies]
-uuid = { workspace = true, features = ["v4", "fast-rng", "macro-diagnostics"] }
+uuid = { workspace = true, features = ["v4", "v5", "fast-rng", "macro-diagnostics"] }
 serial_test = { workspace = true }
 tempfile = { workspace = true }
 aws-config = { workspace = true }

rustfs/src/admin/handlers/site_replication.rs

@@ -41,7 +41,7 @@ use crate::admin::storage_api::config::save_admin_config;
 use crate::admin::storage_api::contract::bucket::{
     BucketOperations, BucketOptions, DeleteBucketOptions, MakeBucketOptions, SRBucketDeleteOp,
 };
-use crate::admin::storage_api::error::Error as StorageError;
+use crate::admin::storage_api::error::{Error as StorageError, is_err_bucket_not_found};
 use crate::admin::storage_api::runtime::ECStore;
 use crate::admin::utils::{encode_compatible_admin_payload, read_compatible_admin_body};
 use crate::auth::constant_time_eq;
@@ -55,6 +55,7 @@ use crate::storage::storage_api::{
 use base64::Engine;
 use base64::engine::general_purpose::STANDARD as BASE64_STANDARD;
 use base64::engine::general_purpose::URL_SAFE_NO_PAD;
+use futures::StreamExt;
 use hmac::{Hmac, Mac};
 use http::header::{CONTENT_TYPE, HOST};
 use http::{HeaderMap, HeaderValue, Uri};
@@ -2096,6 +2097,18 @@ async fn remote_add_preflight_info(site: &PeerSite) -> S3Result<SiteReplicationA
             format!("invalid site replication metainfo from `{}`: {e}", site.endpoint),
         )
     })?;
+    if info.deployment_id.is_empty() {
+        // The peer will be tracked under a locally derived fallback ID
+        // (deployment_id_for_endpoint) instead of its real deployment ID.
+        warn!(
+            event = EVENT_ADMIN_SITE_REPLICATION_STATE,
+            component = LOG_COMPONENT_ADMIN,
+            subsystem = LOG_SUBSYSTEM_SITE_REPLICATION,
+            result = "peer_deployment_id_missing",
+            peer_endpoint = %site.endpoint,
+            "admin site replication state"
+        );
+    }
 
     let idp_body = send_peer_admin_get_request_with_client(
         &client,
@@ -2206,20 +2219,30 @@ fn site_replication_bootstrap_token(uri: &Uri) -> Option<String> {
     query_pairs(uri).get("bootstrapToken").cloned()
 }
 
-fn bootstrap_bucket_make_op_path(bucket: &SRBucketInfo) -> String {
+/// Query for a peer `make-with-versioning` bucket op. `versioningEnabled`
+/// always travels so the outbound query matches MinIO's site-replication
+/// make-bucket wire contract: MinIO's own create-bucket hook sends
+/// `versioningEnabled=true` on this op. RustFS's inbound handler
+/// force-enables versioning either way.
+fn make_with_versioning_bucket_op_path(bucket: &str, created_at: Option<&str>, lock_enabled: bool) -> String {
     let mut query = form_urlencoded::Serializer::new(String::new());
-    query.append_pair("bucket", &bucket.bucket);
-    query.append_pair("operation", "make-with-versioning");
-    if let Some(created_at) = bucket
-        .created_at
-        .and_then(|value| value.format(&time::format_description::well_known::Rfc3339).ok())
-    {
-        query.append_pair("createdAt", &created_at);
+    query.append_pair("bucket", bucket);
+    query.append_pair("operation", SITE_REPLICATION_BUCKET_OP_MAKE_WITH_VERSIONING);
+    query.append_pair("versioningEnabled", "true");
+    if let Some(created_at) = created_at {
+        query.append_pair("createdAt", created_at);
     }
-    if bucket.object_lock_config.is_some() {
+    if lock_enabled {
         query.append_pair("lockEnabled", "true");
     }
-    format!("/rustfs/admin/v3/site-replication/peer/bucket-ops?{}", query.finish())
+    format!("{SITE_REPLICATION_PEER_BUCKET_OPS_PATH}?{}", query.finish())
+}
+
+fn bootstrap_bucket_make_op_path(bucket: &SRBucketInfo) -> String {
+    let created_at = bucket
+        .created_at
+        .and_then(|value| value.format(&time::format_description::well_known::Rfc3339).ok());
+    make_with_versioning_bucket_op_path(&bucket.bucket, created_at.as_deref(), bucket.object_lock_config.is_some())
 }
 
 fn bootstrap_bucket_meta_item(bucket: &SRBucketInfo, item_type: &str, updated_at: Option<OffsetDateTime>) -> SRBucketMeta {
@@ -4246,16 +4269,7 @@ async fn broadcast_site_replication_make_bucket(
         .format(&time::format_description::well_known::Rfc3339)
         .unwrap_or_default();
 
-    let path = {
-        let mut query = form_urlencoded::Serializer::new(String::new());
-        query.append_pair("bucket", bucket);
-        query.append_pair("operation", "make-with-versioning");
-        query.append_pair("createdAt", &created_at);
-        if lock_enabled {
-            query.append_pair("lockEnabled", "true");
-        }
-        format!("/rustfs/admin/v3/site-replication/peer/bucket-ops?{}", query.finish())
-    };
+    let path = make_with_versioning_bucket_op_path(bucket, Some(&created_at), lock_enabled);
     let path = if let Some(token) = bootstrap_token {
         with_site_replication_bootstrap_token(&path, token)
     } else {
@@ -10206,13 +10220,25 @@ impl Operation for SiteReplicationStatusHandler {
     }
 }
 
+/// `POST /v3/site-replication/devnull` — peer link-check upload drain.
+/// MinIO streams multi-megabyte probe bodies here during site netperf link
+/// checks and expects an unbounded discard (its handler copies to io.Discard);
+/// buffering through the 1MB admin body cap turned any larger probe into a
+/// 400 and a false link failure. Stream and discard instead — no size cap.
+async fn drain_site_replication_devnull(mut input: Body) -> S3Result<()> {
+    while let Some(chunk) = input.next().await {
+        chunk.map_err(|e| s3_error!(InvalidRequest, "failed to read devnull stream: {}", e))?;
+    }
+    Ok(())
+}
+
 pub struct SiteReplicationDevNullHandler {}
 
 #[async_trait::async_trait]
 impl Operation for SiteReplicationDevNullHandler {
     async fn call(&self, req: S3Request<Body>, _params: Params<'_, '_>) -> S3Result<S3Response<(StatusCode, Body)>> {
         validate_site_replication_admin_request(&req, AdminAction::SiteReplicationOperationAction).await?;
-        let _ = read_plain_admin_body(req.input).await?;
+        drain_site_replication_devnull(req.input).await?;
         Ok(empty_response(StatusCode::NO_CONTENT))
     }
 }
@@ -10471,6 +10497,19 @@ impl Operation for SRPeerJoinHandler {
     }
 }
 
+/// Outcome of a peer-driven `purge-deleted-bucket` replay. A bucket that is
+/// already gone means the purge raced an earlier replay or a local delete —
+/// that is success — but any other failure must reach the sender like the
+/// sibling delete branches do: swallowing it answered 200 while the bucket
+/// survived on this site.
+fn purge_deleted_bucket_result(result: Result<(), StorageError>) -> S3Result<()> {
+    match result {
+        Ok(()) => Ok(()),
+        Err(err) if is_err_bucket_not_found(&err) => Ok(()),
+        Err(err) => Err(ApiError::from(err).into()),
+    }
+}
+
 pub struct SRPeerBucketOpsHandler {}
 
 #[async_trait::async_trait]
@@ -10570,16 +10609,18 @@ impl Operation for SRPeerBucketOpsHandler {
                     .map_err(ApiError::from)?;
             }
             "purge-deleted-bucket" => {
-                let _ = store
-                    .delete_bucket(
-                        &bucket,
-                        &DeleteBucketOptions {
-                            force: true,
-                            srdelete_op: SRBucketDeleteOp::Purge,
-                            ..Default::default()
-                        },
-                    )
-                    .await;
+                purge_deleted_bucket_result(
+                    store
+                        .delete_bucket(
+                            &bucket,
+                            &DeleteBucketOptions {
+                                force: true,
+                                srdelete_op: SRBucketDeleteOp::Purge,
+                                ..Default::default()
+                            },
+                        )
+                        .await,
+                )?;
             }
             _ => return Err(s3_error!(InvalidRequest, "unsupported site replication bucket operation")),
         }
@@ -13925,6 +13966,54 @@ mod tests {
         assert!(!query_flag(&uri, "missing"));
     }
 
+    /// A5 red-light: a `purge-deleted-bucket` replay must report success when
+    /// the bucket is already gone, and must propagate every other failure —
+    /// the swallowed error answered 200 while the bucket survived.
+    #[test]
+    fn test_purge_deleted_bucket_result_tolerates_only_missing_bucket() {
+        assert!(purge_deleted_bucket_result(Ok(())).is_ok());
+        assert!(purge_deleted_bucket_result(Err(StorageError::BucketNotFound("photos".to_string()))).is_ok());
+        assert!(purge_deleted_bucket_result(Err(StorageError::VolumeNotFound)).is_ok());
+        let err = purge_deleted_bucket_result(Err(StorageError::StorageFull))
+            .expect_err("non-not-found delete failures must propagate");
+        assert_ne!(*err.code(), S3ErrorCode::NoSuchBucket);
+    }
+
+    /// C5 red-light: the site-replication devnull drain must accept bodies
+    /// beyond the 1MB admin body cap — MinIO's link check streams large
+    /// probe bodies and treats a 400 as a broken link.
+    #[tokio::test]
+    async fn test_site_replication_devnull_drains_body_beyond_admin_cap() {
+        let body = Body::from(vec![0u8; MAX_ADMIN_REQUEST_BODY_SIZE + 1]);
+        drain_site_replication_devnull(body)
+            .await
+            .expect("devnull must drain bodies larger than the admin body cap");
+    }
+
+    /// A3 red-light: `versioningEnabled` must travel on every outbound
+    /// make-with-versioning bucket op so the query matches MinIO's
+    /// site-replication make-bucket wire contract (MinIO's own hook sends
+    /// `versioningEnabled=true` on this op).
+    #[test]
+    fn test_make_with_versioning_op_paths_send_versioning_enabled() {
+        let bucket = SRBucketInfo {
+            bucket: "photos".to_string(),
+            created_at: Some(OffsetDateTime::UNIX_EPOCH),
+            object_lock_config: Some(BASE64_STANDARD.encode("<ObjectLockConfiguration/>")),
+            ..Default::default()
+        };
+        let bootstrap = bootstrap_bucket_make_op_path(&bucket);
+        assert!(bootstrap.contains("operation=make-with-versioning"), "{bootstrap}");
+        assert!(bootstrap.contains("versioningEnabled=true"), "{bootstrap}");
+        assert!(bootstrap.contains("createdAt="), "{bootstrap}");
+        assert!(bootstrap.contains("lockEnabled=true"), "{bootstrap}");
+
+        // The broadcast path (create-bucket hook) shares the same builder.
+        let broadcast = make_with_versioning_bucket_op_path("photos", Some("1970-01-01T00:00:00Z"), false);
+        assert!(broadcast.contains("versioningEnabled=true"), "{broadcast}");
+        assert!(!broadcast.contains("lockEnabled"), "{broadcast}");
+    }
+
     #[tokio::test]
     #[serial]
     async fn test_add_bootstrap_scope_only_allows_expected_bucket_setup_until_guard_drops() {

rustfs/src/admin/site_replication_identity.rs

@@ -13,9 +13,9 @@
 // limitations under the License.
 
 use rustfs_madmin::{PeerInfo, SyncStatus};
-use std::collections::{BTreeMap, hash_map::DefaultHasher};
-use std::hash::{Hash, Hasher};
+use std::collections::BTreeMap;
 use url::Url;
+use uuid::Uuid;
 
 fn has_http_scheme(endpoint: &str) -> bool {
     endpoint.get(..7).is_some_and(|prefix| prefix.eq_ignore_ascii_case("http://"))
@@ -66,10 +66,12 @@ pub fn site_identity_key(endpoint: &str) -> String {
         .unwrap_or_else(|| trimmed.to_ascii_lowercase())
 }
 
+/// Fallback deployment ID for a peer that reported none. UUIDv5 over the
+/// canonical endpoint: the ID is persisted in site-replication state and
+/// broadcast to peers, so it must be identical across Rust toolchains
+/// (`DefaultHasher` is not) and across spellings of the same endpoint.
 pub fn deployment_id_for_endpoint(endpoint: &str) -> String {
-    let mut hasher = DefaultHasher::new();
-    endpoint.hash(&mut hasher);
-    format!("{:016x}", hasher.finish())
+    Uuid::new_v5(&Uuid::NAMESPACE_URL, canonical_endpoint(endpoint).as_bytes()).to_string()
 }
 
 pub fn same_identity_endpoint(left: &str, right: &str) -> bool {
@@ -174,6 +176,23 @@ mod tests {
         }
     }
 
+    /// B8 red-light: the fallback deployment ID must be a toolchain-stable
+    /// UUIDv5 over the canonical endpoint — `DefaultHasher` output is not
+    /// guaranteed stable across Rust releases, yet the ID is persisted in
+    /// site-replication state and broadcast to peers.
+    #[test]
+    fn deployment_id_for_endpoint_is_stable_uuid_v5_over_canonical_endpoint() {
+        let endpoint = "https://node-a.example.com:9000";
+        let id = deployment_id_for_endpoint(endpoint);
+        let parsed = uuid::Uuid::parse_str(&id).expect("fallback deployment ID must be a UUID");
+        assert_eq!(parsed.get_version_num(), 5, "fallback deployment ID must be UUIDv5");
+        // Deterministic for the same endpoint and for spelling variants that
+        // share a canonical form; distinct endpoints stay distinct.
+        assert_eq!(id, deployment_id_for_endpoint(endpoint));
+        assert_eq!(id, deployment_id_for_endpoint(" HTTPS://Node-A.Example.Com:9000/ "));
+        assert_ne!(id, deployment_id_for_endpoint("https://node-b.example.com:9000"));
+    }
+
     #[test]
     fn canonical_endpoint_accepts_case_insensitive_scheme() {
         assert_eq!(

rustfs/src/admin/storage_api.rs

@@ -51,7 +51,7 @@ mod ecstore_disk {
 }
 
 mod ecstore_error {
-    pub(crate) use crate::storage::storage_api::ecstore_error::StorageError;
+    pub(crate) use crate::storage::storage_api::ecstore_error::{StorageError, is_err_bucket_not_found};
 }
 
 #[allow(unused_imports)]
@@ -919,6 +919,7 @@ pub(crate) mod contract {
 }
 
 pub(crate) mod error {
+    pub(crate) use super::ecstore_error::is_err_bucket_not_found;
     pub(crate) use super::{Error, StorageError};
 }