fix(ecstore): restore decommission test imports

.config/nextest.toml

@@ -78,6 +78,12 @@ test-group = 'embedded-test-ports'
 filter = 'package(rustfs-ecstore) & test(manual_transition_page_checkpoint_persists_durable_job_progress)'
 test-group = 'ecstore-serial-flaky'
 
+# The durable ILM decommission regressions build isolated multi-pool stores and
+# deliberately take source or target disks offline while checking fencing.
+[[profile.default.overrides]]
+filter = 'package(rustfs-ecstore) & (test(decommission_migrates_and_verifies_registered_durable_ilm_records) | test(decommission_durable_ilm_target_read_error_is_not_masked_by_peer_success) | test(decommission_durable_ilm_terminal_receipt_recovers_failed_source_cleanup) | test(decommission_durable_ilm_receipt_pagination_fails_closed_on_second_page) | test(decommission_durable_ilm_recovery_keeps_multiple_active_sources))'
+test-group = 'ecstore-serial-flaky'
+
 # Serialize the bucket-incarnation / lifecycle-fence tests. They drive
 # init_bucket_metadata_sys and bucket_metadata_sys_of, i.e. process-global
 # OnceLock state that serial_test's #[serial] cannot protect across nextest's
@@ -190,6 +196,10 @@ test-group = 'embedded-test-ports'
 filter = 'package(rustfs-ecstore) & test(manual_transition_page_checkpoint_persists_durable_job_progress)'
 test-group = 'ecstore-serial-flaky'
 
+[[profile.ci.overrides]]
+filter = 'package(rustfs-ecstore) & (test(decommission_migrates_and_verifies_registered_durable_ilm_records) | test(decommission_durable_ilm_target_read_error_is_not_masked_by_peer_success) | test(decommission_durable_ilm_terminal_receipt_recovers_failed_source_cleanup) | test(decommission_durable_ilm_receipt_pagination_fails_closed_on_second_page) | test(decommission_durable_ilm_recovery_keeps_multiple_active_sources))'
+test-group = 'ecstore-serial-flaky'
+
 # Serialize the bucket-incarnation / lifecycle-fence tests under the ci profile
 # too (see the matching default-profile override near the top). No retries.
 [[profile.ci.overrides]]

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

@@ -9136,6 +9136,7 @@ mod tests {
         assert_eq!(loaded.report.scanned, 37);
         assert_eq!(loaded.report.eligible, 11);
         assert_eq!(loaded.report.enqueued, 5);
+        assert_eq!(loaded.cursor_revision, Some(1));
         assert!(loaded.lease_expires_at_unix_nanos > 0);
         let token = loaded
             .report
@@ -9154,6 +9155,30 @@ mod tests {
         assert_eq!(admission.lease_id, loaded.lease_id);
         assert_eq!(admission.lease_expires_at_unix_nanos, loaded.lease_expires_at_unix_nanos);
 
+        let mut same_marker_report = report.clone();
+        same_marker_report.scanned += 1;
+        persist_manual_transition_page_checkpoint(
+            &checkpoint_options,
+            &same_marker_report,
+            Some("logs/page-end".to_string()),
+            Some("opaque-next-version".to_string()),
+        )
+        .await
+        .expect("same-marker version checkpoint should persist through the durable progress sink");
+        let same_marker_checkpointed = load_manual_transition_job_record(ecstore.clone(), job_id)
+            .await
+            .expect("same-marker version checkpoint should reload");
+        assert_eq!(same_marker_checkpointed.cursor_revision, Some(2));
+        let (_, version_marker) = decode_manual_transition_continuation_token(
+            same_marker_checkpointed
+                .report
+                .continuation_token
+                .as_deref()
+                .expect("same-marker version checkpoint should persist a cursor"),
+        )
+        .expect("same-marker version cursor should decode");
+        assert_eq!(version_marker.as_deref(), Some("opaque-next-version"));
+
         create_test_bucket(&ecstore, &bucket).await;
         let lifecycle_xml = format!(
             r#"<?xml version="1.0" encoding="UTF-8"?>
@@ -9210,6 +9235,7 @@ mod tests {
         assert_eq!(checkpointed.report.scanned, 1000);
         assert_eq!(checkpointed.report.eligible, 1000);
         assert_eq!(checkpointed.report.dry_run_eligible, 1000);
+        assert_eq!(checkpointed.cursor_revision, Some(3));
         let token = checkpointed
             .report
             .continuation_token

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

@@ -24,6 +24,10 @@ use crate::bucket::lifecycle::bucket_lifecycle_ops::{
     ManualTransitionQueueSnapshot, ManualTransitionRunOptions, ManualTransitionRunReport,
 };
 use crate::bucket::lifecycle::config_boundary;
+use crate::bucket::lifecycle::durable_namespace::{
+    MANUAL_TRANSITION_JOB_NAMESPACE, MANUAL_TRANSITION_SCOPE_NAMESPACE, MANUAL_TRANSITION_TASK_NAMESPACE,
+    MANUAL_TRANSITION_WORKER_RESULT_NAMESPACE,
+};
 use crate::disk::RUSTFS_META_BUCKET;
 use crate::error::{Error, Result as EcstoreResult};
 use crate::object_api::ObjectOptions;
@@ -34,10 +38,10 @@ use crate::store::ECStore;
 pub const MANUAL_TRANSITION_JOB_SCHEMA: &str = "rustfs-manual-transition-job-v1";
 pub const MANUAL_TRANSITION_TASK_SCHEMA: &str = "rustfs-manual-transition-task-v1";
 pub const MANUAL_TRANSITION_WORKER_RESULT_SCHEMA: &str = "rustfs-manual-transition-worker-result-v1";
-pub const MANUAL_TRANSITION_JOB_RECORD_PREFIX: &str = "ilm/manual-transition/jobs";
-pub const MANUAL_TRANSITION_SCOPE_RECORD_PREFIX: &str = "ilm/manual-transition/scopes";
-pub const MANUAL_TRANSITION_TASK_PREFIX: &str = "ilm/manual-transition/tasks";
-pub const MANUAL_TRANSITION_WORKER_RESULT_PREFIX: &str = "ilm/manual-transition/results";
+pub const MANUAL_TRANSITION_JOB_RECORD_PREFIX: &str = MANUAL_TRANSITION_JOB_NAMESPACE.prefix;
+pub const MANUAL_TRANSITION_SCOPE_RECORD_PREFIX: &str = MANUAL_TRANSITION_SCOPE_NAMESPACE.prefix;
+pub const MANUAL_TRANSITION_TASK_PREFIX: &str = MANUAL_TRANSITION_TASK_NAMESPACE.prefix;
+pub const MANUAL_TRANSITION_WORKER_RESULT_PREFIX: &str = MANUAL_TRANSITION_WORKER_RESULT_NAMESPACE.prefix;
 pub const MAX_MANUAL_TRANSITION_JOB_RECORD_SIZE: usize = 64 * 1024;
 pub const MAX_MANUAL_TRANSITION_TASK_RECORD_SIZE: usize = 16 * 1024;
 pub const MAX_MANUAL_TRANSITION_WORKER_RESULT_RECORD_SIZE: usize = 8 * 1024;
@@ -195,6 +199,8 @@ pub struct ManualTransitionJobRecord {
     pub updated_at_unix_nanos: i128,
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub completed_at_unix_nanos: Option<i128>,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub cursor_revision: Option<u64>,
     pub report: ManualTransitionRunReport,
     pub queue_snapshot: ManualTransitionQueueSnapshot,
     #[serde(default, skip_serializing_if = "Option::is_none")]
@@ -224,6 +230,7 @@ impl ManualTransitionJobRecord {
             created_at_unix_nanos: now,
             updated_at_unix_nanos: now,
             completed_at_unix_nanos: None,
+            cursor_revision: Some(0),
             report: ManualTransitionRunReport {
                 bucket: bucket.to_string(),
                 prefix: options.prefix.clone(),
@@ -238,7 +245,7 @@ impl ManualTransitionJobRecord {
 
     pub fn complete(&mut self, report: ManualTransitionRunReport, queue_snapshot: ManualTransitionQueueSnapshot) {
         self.scan_completed = true;
-        self.report.merge_scan_report_preserving_worker(&report);
+        self.merge_scan_report(&report);
         self.queue_snapshot = queue_snapshot;
         self.error = None;
         self.mark_terminal_if_worker_drained();
@@ -316,7 +323,7 @@ impl ManualTransitionJobRecord {
             }
         }
         self.queue_snapshot = queue_snapshot;
-        self.updated_at_unix_nanos = OffsetDateTime::now_utc().unix_timestamp_nanos();
+        self.advance_updated_at();
         self.mark_terminal_if_worker_drained();
     }
 
@@ -359,14 +366,14 @@ impl ManualTransitionJobRecord {
         self.report.tier_failure = scan_tier_failure.saturating_add(transition_failed);
         self.report.tier_failure_by_reason = scan_tier_failure_by_reason;
         self.queue_snapshot = queue_snapshot;
-        self.updated_at_unix_nanos = OffsetDateTime::now_utc().unix_timestamp_nanos();
+        self.advance_updated_at();
         self.mark_terminal_if_worker_drained();
         true
     }
 
     pub fn mark_cancel_requested(&mut self) {
         self.cancel_requested = true;
-        self.updated_at_unix_nanos = OffsetDateTime::now_utc().unix_timestamp_nanos();
+        self.advance_updated_at();
     }
 
     pub fn claim_recovery_lease(&mut self, owner_id: impl Into<String>, queue_snapshot: ManualTransitionQueueSnapshot) {
@@ -380,7 +387,7 @@ impl ManualTransitionJobRecord {
     pub fn abandon_recovery_lease(&mut self, lease_id: Uuid) {
         if self.state == ManualTransitionJobState::Running && self.lease_id == lease_id {
             self.lease_expires_at_unix_nanos = 0;
-            self.updated_at_unix_nanos = OffsetDateTime::now_utc().unix_timestamp_nanos();
+            self.advance_updated_at();
         }
     }
 
@@ -398,7 +405,7 @@ impl ManualTransitionJobRecord {
 
     pub fn renew_lease(&mut self, queue_snapshot: ManualTransitionQueueSnapshot) {
         let now = OffsetDateTime::now_utc().unix_timestamp_nanos();
-        self.updated_at_unix_nanos = now;
+        self.updated_at_unix_nanos = self.updated_at_unix_nanos.saturating_add(1).max(now);
         self.lease_expires_at_unix_nanos = manual_transition_job_lease_expires_at(now);
         self.queue_snapshot = queue_snapshot;
     }
@@ -438,11 +445,18 @@ impl ManualTransitionJobRecord {
 
     pub fn update_running_progress(&mut self, report: ManualTransitionRunReport, queue_snapshot: ManualTransitionQueueSnapshot) {
         if self.state == ManualTransitionJobState::Running {
-            self.report.merge_scan_report_preserving_worker(&report);
+            self.merge_scan_report(&report);
             self.renew_lease(queue_snapshot);
         }
     }
 
+    fn merge_scan_report(&mut self, report: &ManualTransitionRunReport) {
+        if self.report.continuation_token != report.continuation_token {
+            self.cursor_revision = Some(self.cursor_revision.unwrap_or(0).saturating_add(1));
+        }
+        self.report.merge_scan_report_preserving_worker(report);
+    }
+
     pub fn mark_unknown_if_unowned(&mut self) {
         if self.state == ManualTransitionJobState::Running {
             self.state = ManualTransitionJobState::Unknown;
@@ -463,9 +477,13 @@ impl ManualTransitionJobRecord {
     }
 
     fn mark_updated_terminal(&mut self) {
+        self.advance_updated_at();
+        self.completed_at_unix_nanos = Some(self.updated_at_unix_nanos);
+    }
+
+    fn advance_updated_at(&mut self) {
         let now = OffsetDateTime::now_utc().unix_timestamp_nanos();
-        self.updated_at_unix_nanos = now;
-        self.completed_at_unix_nanos = Some(now);
+        self.updated_at_unix_nanos = self.updated_at_unix_nanos.saturating_add(1).max(now);
     }
 
     fn mark_terminal_if_worker_drained(&mut self) {
@@ -1109,7 +1127,8 @@ pub fn manual_transition_scope_record_object_name(scope_key: &str) -> Result<Str
 pub async fn save_manual_transition_job_record(api: Arc<ECStore>, job: &ManualTransitionJobRecord) -> EcstoreResult<()> {
     let object = manual_transition_job_record_object_name(job.job_id).map_err(manual_transition_job_store_error)?;
     let data = job.encode().map_err(manual_transition_job_store_error)?;
-    config_boundary::save_config(api, &object, data).await
+    config_boundary::save_config(api.clone(), &object, data.clone()).await?;
+    api.record_durable_ilm_decommission_progress(&object, &data).await
 }
 
 pub async fn load_manual_transition_job_record(api: Arc<ECStore>, job_id: Uuid) -> EcstoreResult<ManualTransitionJobRecord> {
@@ -1142,9 +1161,9 @@ pub async fn save_manual_transition_job_record_if_current(
     let object = manual_transition_job_record_object_name(job.job_id).map_err(manual_transition_job_store_error)?;
     let data = job.encode().map_err(manual_transition_job_store_error)?;
     config_boundary::save_config_with_opts_quiet(
-        api,
+        api.clone(),
         &object,
-        data,
+        data.clone(),
         &ObjectOptions {
             max_parity: true,
             http_preconditions: Some(HTTPPreconditions {
@@ -1154,7 +1173,8 @@ pub async fn save_manual_transition_job_record_if_current(
             ..Default::default()
         },
     )
-    .await
+    .await?;
+    api.record_durable_ilm_decommission_progress(&object, &data).await
 }
 
 /// Applies a job-record mutation with optimistic concurrency control.
@@ -1592,9 +1612,9 @@ pub async fn save_manual_transition_scope_admission_if_absent(
     let object = manual_transition_scope_record_object_name(&admission.scope_key).map_err(manual_transition_job_store_error)?;
     let data = serde_json::to_vec(admission).map_err(Error::other)?;
     config_boundary::save_config_with_opts(
-        api,
+        api.clone(),
         &object,
-        data,
+        data.clone(),
         &ObjectOptions {
             max_parity: true,
             http_preconditions: Some(HTTPPreconditions {
@@ -1604,7 +1624,8 @@ pub async fn save_manual_transition_scope_admission_if_absent(
             ..Default::default()
         },
     )
-    .await
+    .await?;
+    api.record_durable_ilm_decommission_progress(&object, &data).await
 }
 
 pub async fn load_manual_transition_scope_admission(
@@ -1642,9 +1663,9 @@ pub async fn save_manual_transition_scope_admission_if_current(
     let object = manual_transition_scope_record_object_name(&admission.scope_key).map_err(manual_transition_job_store_error)?;
     let data = serde_json::to_vec(admission).map_err(Error::other)?;
     match config_boundary::save_config_with_opts(
-        api,
+        api.clone(),
         &object,
-        data,
+        data.clone(),
         &ObjectOptions {
             max_parity: true,
             http_preconditions: Some(HTTPPreconditions {
@@ -1660,7 +1681,8 @@ pub async fn save_manual_transition_scope_admission_if_current(
             Err(Error::PreconditionFailed)
         }
         result => result,
-    }
+    }?;
+    api.record_durable_ilm_decommission_progress(&object, &data).await
 }
 
 pub async fn claim_manual_transition_scope_admission(
@@ -1953,13 +1975,15 @@ pub async fn delete_manual_transition_scope_admission_if_current(
     job_id: Uuid,
     lease_id: Uuid,
 ) -> EcstoreResult<bool> {
-    let etag = match load_manual_transition_scope_admission_with_etag(api.clone(), scope_key).await {
-        Ok((admission, etag)) if admission.job_id == job_id && admission.lease_id == lease_id => etag,
+    let (admission, etag) = match load_manual_transition_scope_admission_with_etag(api.clone(), scope_key).await {
+        Ok((admission, etag)) if admission.job_id == job_id && admission.lease_id == lease_id => (admission, etag),
         Ok(_) => return Ok(false),
         Err(Error::ConfigNotFound) => return Ok(true),
         Err(err) => return Err(err),
     };
     let object = manual_transition_scope_record_object_name(scope_key).map_err(manual_transition_job_store_error)?;
+    let data = serde_json::to_vec(&admission).map_err(Error::other)?;
+    api.record_durable_ilm_decommission_terminal(&object, &data).await?;
     match config_boundary::delete_config_if_match(api, &object, &etag).await {
         Ok(()) | Err(Error::ConfigNotFound) => Ok(true),
         Err(Error::PreconditionFailed) => Ok(false),
@@ -2577,6 +2601,25 @@ mod tests {
         assert!(decoded.report.tier_failure_by_reason.is_empty());
     }
 
+    #[test]
+    fn manual_transition_job_record_decodes_legacy_cursor_without_revision() {
+        let options = ManualTransitionRunOptions::default();
+        let record = ManualTransitionJobRecord::new(Uuid::new_v4(), "bucket", &options, TEST_OWNER);
+        let encoded = record.encode().expect("job record should encode");
+        let mut value: serde_json::Value = serde_json::from_slice(&encoded).expect("encoded job should be json");
+        value["job"]
+            .as_object_mut()
+            .expect("job should be object")
+            .remove("cursor_revision");
+        let record_bytes = serde_json::to_vec(&value["job"]).expect("legacy job should encode");
+        value["content_sha256"] = serde_json::Value::String(hex_sha256(&record_bytes, ToOwned::to_owned));
+        let legacy = serde_json::to_vec(&value).expect("legacy envelope should encode");
+
+        let decoded = ManualTransitionJobRecord::decode(record.job_id, &legacy).expect("legacy job should decode");
+
+        assert_eq!(decoded.cursor_revision, None);
+    }
+
     #[test]
     fn manual_transition_job_record_rejects_unknown_report_fields() {
         let options = ManualTransitionRunOptions::default();

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

@@ -16,6 +16,7 @@ pub mod bucket_lifecycle_audit;
 pub mod bucket_lifecycle_ops;
 mod config_boundary;
 pub mod core;
+mod durable_namespace;
 pub mod evaluator;
 pub mod manual_transition_job;
 mod metadata_boundary;
@@ -31,3 +32,8 @@ pub mod tier_free_version_recovery;
 pub mod tier_last_day_stats;
 pub mod tier_sweeper;
 pub mod transition_transaction;
+
+pub(crate) use durable_namespace::{
+    DurableIlmRecordCheckpoint, ILM_META_PREFIX, ValidatedDurableIlmRecord, classify_durable_ilm_record,
+    validate_durable_ilm_record,
+};

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

@@ -20,6 +20,7 @@ use tokio_util::sync::CancellationToken;
 use tracing::{debug, warn};
 
 use crate::bucket::lifecycle::config_boundary;
+use crate::bucket::lifecycle::durable_namespace::TIER_DELETE_JOURNAL_NAMESPACE;
 use crate::bucket::lifecycle::runtime_boundary;
 use crate::bucket::lifecycle::tier_sweeper::{
     Jentry, TierDeleteJournalState, TierDeleteSourceIdentity,
@@ -49,7 +50,7 @@ const TIER_DELETE_JOURNAL_VERSION: u8 = 2;
 const TIER_DELETE_JOURNAL_EXACT_VERSION: u8 = 3;
 const TIER_DELETE_JOURNAL_STATE_VERSION: u8 = 4;
 const TIER_DELETE_JOURNAL_TRANSACTION_VERSION: u8 = 5;
-pub(crate) const TIER_DELETE_JOURNAL_PREFIX: &str = "ilm/tier-delete-journal/";
+pub(crate) const TIER_DELETE_JOURNAL_PREFIX: &str = TIER_DELETE_JOURNAL_NAMESPACE.prefix;
 
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
 #[serde(deny_unknown_fields)]
@@ -432,6 +433,21 @@ async fn process_committed_tier_delete_journal_entry(api: Arc<ECStore>, je: &Jen
         )
         .await?;
     }
+    let path = tier_delete_journal_object_name(je);
+    let data = encode_tier_delete_journal_entry(je).map_err(std::io::Error::other)?;
+    let target_pool_indices = api
+        .record_durable_ilm_decommission_terminal_target_pools(&path, &data)
+        .await
+        .map_err(std::io::Error::other)?;
+    if let Some(target_pool_indices) = target_pool_indices {
+        for target_pool_idx in target_pool_indices {
+            match config_boundary::delete_config(api.pools[target_pool_idx].clone(), &path).await {
+                Ok(()) | Err(Error::ConfigNotFound) => {}
+                Err(err) => return Err(std::io::Error::other(err)),
+            }
+        }
+        return Ok(());
+    }
     remove_tier_delete_journal_entry(api, je).await
 }
 

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

@@ -21,6 +21,7 @@ use tracing::{debug, warn};
 use uuid::Uuid;
 
 use crate::bucket::lifecycle::config_boundary;
+use crate::bucket::lifecycle::durable_namespace::TRANSITION_TRANSACTION_NAMESPACE;
 use crate::bucket::lifecycle::lifecycle::TRANSITION_COMPLETE;
 use crate::bucket::lifecycle::tier_sweeper::{
     delete_confirmed_transition_candidate_exact_with_lease_idempotent,
@@ -42,7 +43,7 @@ const TRANSITION_TRANSACTION_RECOVERY_INTERVAL: Duration = Duration::from_secs(6
 const TRANSITION_TRANSACTION_RECOVERY_TIMEOUT: Duration = Duration::from_secs(300);
 pub const TRANSITION_TRANSACTION_SCHEMA: &str = "rustfs-transition-transaction-v1";
 pub const TRANSITION_TRANSACTION_PREFIX: &str = "ilm/transition-transactions";
-pub const TRANSITION_TRANSACTION_RECORD_PREFIX: &str = "ilm/transition-transactions/records";
+pub const TRANSITION_TRANSACTION_RECORD_PREFIX: &str = TRANSITION_TRANSACTION_NAMESPACE.prefix;
 pub const MAX_TRANSITION_TRANSACTION_SIZE: usize = 64 * 1024;
 
 pub type Result<T> = std::result::Result<T, TransitionTransactionError>;
@@ -584,7 +585,8 @@ pub(crate) async fn save_transition_transaction_record(
     let object =
         transition_transaction_record_object_name(transaction.transaction_id).map_err(transition_transaction_store_error)?;
     let data = transaction.encode().map_err(transition_transaction_store_error)?;
-    config_boundary::save_config(api, &object, data).await
+    config_boundary::save_config(api.clone(), &object, data.clone()).await?;
+    api.record_durable_ilm_decommission_progress(&object, &data).await
 }
 
 pub(crate) async fn load_transition_transaction_record(
@@ -596,8 +598,14 @@ pub(crate) async fn load_transition_transaction_record(
     TransitionTransaction::decode(transaction_id, &data).map_err(transition_transaction_store_error)
 }
 
-pub(crate) async fn delete_transition_transaction_record(api: Arc<ECStore>, transaction_id: Uuid) -> EcstoreResult<()> {
-    let object = transition_transaction_record_object_name(transaction_id).map_err(transition_transaction_store_error)?;
+pub(crate) async fn delete_transition_transaction_record(
+    api: Arc<ECStore>,
+    transaction: &TransitionTransaction,
+) -> EcstoreResult<()> {
+    let object =
+        transition_transaction_record_object_name(transaction.transaction_id).map_err(transition_transaction_store_error)?;
+    let data = transaction.encode().map_err(transition_transaction_store_error)?;
+    api.record_durable_ilm_decommission_terminal(&object, &data).await?;
     match config_boundary::delete_config(api, &object).await {
         Ok(()) | Err(Error::ConfigNotFound) => Ok(()),
         Err(err) => Err(err),
@@ -813,7 +821,7 @@ pub async fn finalize_missing_transition_transaction_for_operator(
     if probe != TransitionOperatorProbe::Missing {
         return Err(TransitionOperatorError::CandidateNotMissing(probe));
     }
-    delete_transition_transaction_record(api, transaction_id)
+    delete_transition_transaction_record(api, &transaction)
         .await
         .map_err(TransitionOperatorError::Store)
 }
@@ -849,22 +857,22 @@ pub async fn process_transition_transaction_record(
     match transaction.state {
         TransitionTransactionState::Uploaded => {
             delete_transition_remote_candidate(api.clone(), transaction).await?;
-            delete_transition_transaction_record(api, transaction.transaction_id).await?;
+            delete_transition_transaction_record(api, transaction).await?;
             Ok(TransitionTransactionRecoveryOutcome::RemoteCandidateDeleted)
         }
         TransitionTransactionState::CleanupPending => match local_commit_matches_transaction(api.clone(), transaction).await {
             Ok(true) => {
-                delete_transition_transaction_record(api, transaction.transaction_id).await?;
+                delete_transition_transaction_record(api, transaction).await?;
                 Ok(TransitionTransactionRecoveryOutcome::RecordDeleted)
             }
             Ok(false) => {
                 delete_transition_remote_candidate(api.clone(), transaction).await?;
-                delete_transition_transaction_record(api, transaction.transaction_id).await?;
+                delete_transition_transaction_record(api, transaction).await?;
                 Ok(TransitionTransactionRecoveryOutcome::RemoteCandidateDeleted)
             }
             Err(err) if transition_source_is_missing(&err) => {
                 delete_transition_remote_candidate(api.clone(), transaction).await?;
-                delete_transition_transaction_record(api, transaction.transaction_id).await?;
+                delete_transition_transaction_record(api, transaction).await?;
                 Ok(TransitionTransactionRecoveryOutcome::RemoteCandidateDeleted)
             }
             Err(err) => Err(err),
@@ -872,7 +880,7 @@ pub async fn process_transition_transaction_record(
         TransitionTransactionState::LocalCommitStarted => {
             match local_commit_matches_transaction(api.clone(), transaction).await {
                 Ok(true) => {
-                    delete_transition_transaction_record(api, transaction.transaction_id).await?;
+                    delete_transition_transaction_record(api, transaction).await?;
                     Ok(TransitionTransactionRecoveryOutcome::RecordDeleted)
                 }
                 Ok(false) => Ok(TransitionTransactionRecoveryOutcome::Retained),
@@ -881,7 +889,7 @@ pub async fn process_transition_transaction_record(
             }
         }
         TransitionTransactionState::AbortedNoRemote | TransitionTransactionState::Committed => {
-            delete_transition_transaction_record(api, transaction.transaction_id).await?;
+            delete_transition_transaction_record(api, transaction).await?;
             Ok(TransitionTransactionRecoveryOutcome::RecordDeleted)
         }
         TransitionTransactionState::UploadOutcomeUnknown => recover_unknown_upload_outcome(api, transaction).await,
@@ -907,7 +915,7 @@ async fn recover_unknown_upload_outcome(
         .map_err(Error::other)?
     {
         TransitionCandidateProbe::Missing => {
-            delete_transition_transaction_record(api, transaction.transaction_id).await?;
+            delete_transition_transaction_record(api, transaction).await?;
             Ok(TransitionTransactionRecoveryOutcome::RecordDeleted)
         }
         TransitionCandidateProbe::UnversionedPresent => {
@@ -925,7 +933,7 @@ async fn recover_unknown_upload_outcome(
             )
             .await
             .map_err(Error::other)?;
-            delete_transition_transaction_record(api, transaction.transaction_id).await?;
+            delete_transition_transaction_record(api, transaction).await?;
             Ok(TransitionTransactionRecoveryOutcome::RemoteCandidateDeleted)
         }
         TransitionCandidateProbe::VersionedPresent(version_id) => {
@@ -958,7 +966,7 @@ async fn cleanup_recovered_unknown_upload_candidate(
         .map_err(transition_transaction_store_error)?;
     save_transition_transaction_record(api.clone(), &cleanup).await?;
     delete_transition_remote_candidate(api.clone(), &cleanup).await?;
-    delete_transition_transaction_record(api, cleanup.transaction_id).await?;
+    delete_transition_transaction_record(api, &cleanup).await?;
     Ok(TransitionTransactionRecoveryOutcome::RemoteCandidateDeleted)
 }
 

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

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

crates/ecstore/src/config/com.rs

@@ -406,6 +406,25 @@ where
     Ok(data)
 }
 
+pub(crate) async fn read_config_limited_preserve_empty<S>(api: Arc<S>, file: &str, max_bytes: usize) -> Result<Vec<u8>>
+where
+    S: EcstoreObjectIO,
+{
+    let (data, _obj) = read_config_limited_preserve_empty_with_metadata(api, file, max_bytes).await?;
+    Ok(data)
+}
+
+pub(crate) async fn read_config_limited_preserve_empty_with_metadata<S>(
+    api: Arc<S>,
+    file: &str,
+    max_bytes: usize,
+) -> Result<(Vec<u8>, ObjectInfo)>
+where
+    S: EcstoreObjectIO,
+{
+    read_config_with_metadata_inner(api, file, &ObjectOptions::default(), true, Some(max_bytes)).await
+}
+
 /// Read an existing config object without treating an empty payload as absent.
 /// Callers that validate their own payload format need to distinguish corruption
 /// from `ConfigNotFound`.

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

@@ -37,7 +37,7 @@ use crate::bucket::lifecycle::{
     transition_transaction::{
         TransitionRemoteVersion, TransitionSourceIdentity, TransitionSourceVersionMode, TransitionTransaction,
         TransitionTransactionInit, TransitionTransactionState, delete_transition_transaction_record,
-        save_transition_transaction_record,
+        load_transition_transaction_record, save_transition_transaction_record,
     },
 };
 use crate::bucket::quota::reservation;
@@ -4246,7 +4246,12 @@ fn record_transition_uploaded_save_attempt(transaction: &TransitionTransaction,
 
 async fn delete_transition_transaction_if_available(api: Option<&Arc<ECStore>>, transaction_id: Uuid) -> Result<()> {
     if let Some(api) = api {
-        return delete_transition_transaction_record(api.clone(), transaction_id).await;
+        let transaction = match load_transition_transaction_record(api.clone(), transaction_id).await {
+            Ok(transaction) => transaction,
+            Err(Error::ConfigNotFound) => return Ok(()),
+            Err(err) => return Err(err),
+        };
+        return delete_transition_transaction_record(api.clone(), &transaction).await;
     }
     Ok(())
 }

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

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

crates/protos/src/node.proto

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

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

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