fix(admin): gate, bound, and null-map the mrf stream

Second review round:

- Authorization: the default stream enumerates object names and version
  ids, which a metrics-only principal must not see — it now requires
  admin:ReplicationDiff (MinIO parity, route policy updated);
  ?aggregate=true carries no object identities and keeps
  admin:GetReplicationMetrics.
- The nil UUID is RustFS's in-memory null-version sentinel and now
  leaves as the S3 wire token 'null' instead of a zero UUID (a
  pre-versioning object scanned after versioning + existing-object
  replication can persist it into the ledger).
- The durable ledger is not bounded by the in-memory pending cap and
  the body is buffered before send; the stream now stops at 10,000
  documents and signals truncation via
  x-rustfs-replication-mrf-truncated (mirroring the diff endpoint)
  plus a warn event, instead of staging an unbounded body.

.github/workflows/docker.yml

@@ -94,6 +94,7 @@ jobs:
       short_sha: ${{ steps.check.outputs.short_sha }}
       is_prerelease: ${{ steps.check.outputs.is_prerelease }}
       create_latest: ${{ steps.check.outputs.create_latest }}
+      source_ref: ${{ steps.check.outputs.source_ref }}
     steps:
       - name: Checkout repository
         uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
@@ -118,6 +119,7 @@ jobs:
           short_sha=""
           is_prerelease=false
           create_latest=false
+          source_ref="$GITHUB_SHA"
 
           if [[ "${{ github.event_name }}" == "workflow_run" ]]; then
             # Triggered by build workflow completion
@@ -137,6 +139,7 @@ jobs:
             # Extract version info from commit message or use commit SHA
             # Use Git to generate consistent short SHA (ensures uniqueness like build.yml)
             short_sha=$(git rev-parse --short "$HEAD_SHA")
+            source_ref="$HEAD_SHA"
 
             # Determine build type based on triggering workflow event and ref
             triggering_event="$TRIGGERING_EVENT"
@@ -261,6 +264,23 @@ jobs:
                 echo "⚠️  Only release versions (latest, v1.0.0, 1.0.0) and prereleases (v1.0.0-alpha1, 1.0.0-beta2) are supported"
                 ;;
             esac
+
+            if [[ "$should_build" == true && "$input_version" != "latest" ]]; then
+              tag_ref="refs/tags/$input_version"
+              if ! git ls-remote --exit-code origin "$tag_ref" >/dev/null 2>&1; then
+                if [[ "$input_version" == v* ]]; then
+                  tag_ref="refs/tags/${input_version#v}"
+                else
+                  tag_ref="refs/tags/v$input_version"
+                fi
+              fi
+
+              if ! git ls-remote --exit-code origin "$tag_ref" >/dev/null 2>&1; then
+                echo "❌ Release tag not found for Docker build: $input_version"
+                exit 1
+              fi
+              source_ref="$tag_ref"
+            fi
           fi
 
           {
@@ -271,6 +291,7 @@ jobs:
             echo "short_sha=$short_sha"
             echo "is_prerelease=$is_prerelease"
             echo "create_latest=$create_latest"
+            echo "source_ref=$source_ref"
           } >> "$GITHUB_OUTPUT"
 
           echo "🐳 Docker Build Summary:"
@@ -281,6 +302,7 @@ jobs:
           echo "  - Short SHA: $short_sha"
           echo "  - Is prerelease: $is_prerelease"
           echo "  - Create latest: $create_latest"
+          echo "  - Source ref: $source_ref"
 
   # Build multi-arch Docker images
   # Strategy: Build images using pre-built binaries from dl.rustfs.com
@@ -308,6 +330,7 @@ jobs:
         uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7
         with:
           persist-credentials: false
+          ref: ${{ needs.build-check.outputs.source_ref }}
 
       - name: Login to Docker Hub
         uses: docker/login-action@c94ce9fb468520275223c153574b00df6fe4bcc9 # v3
@@ -397,7 +420,8 @@ jobs:
           LABELS="org.opencontainers.image.title=RustFS"
           LABELS="$LABELS,org.opencontainers.image.description=RustFS distributed object storage system"
           LABELS="$LABELS,org.opencontainers.image.version=$VERSION"
-          LABELS="$LABELS,org.opencontainers.image.revision=${{ github.sha }}"
+          SOURCE_REVISION="$(git rev-parse HEAD)"
+          LABELS="$LABELS,org.opencontainers.image.revision=$SOURCE_REVISION"
           LABELS="$LABELS,org.opencontainers.image.source=${{ github.server_url }}/${{ github.repository }}"
           LABELS="$LABELS,org.opencontainers.image.created=$(date -u +'%Y-%m-%dT%H:%M:%SZ')"
           LABELS="$LABELS,org.opencontainers.image.build-type=$BUILD_TYPE"

rustfs/src/admin/handlers/replication.rs

@@ -1153,7 +1153,7 @@ struct MrfResponse {
 fn build_mrf_response(
     bucket: String,
     bucket_stats: &BucketStats,
-    durable: crate::admin::storage_api::replication::DurableMrfBacklog,
+    durable: &crate::admin::storage_api::replication::DurableMrfBacklog,
 ) -> MrfResponse {
     let observation_scope = if bucket_stats.replication_stats.cluster_complete {
         "cluster_aggregated"
@@ -1223,7 +1223,10 @@ fn build_mrf_response(
         total_failed_size,
         queued_count: queued.count,
         queued_size: queued.bytes,
-        per_object_entries_available: false,
+        // The default (non-aggregate) response mode streams the durable
+        // backlog per object, so the enumerable API exists whenever the
+        // backlog is readable.
+        per_object_entries_available: durable.available,
         runtime_stats_available: bucket_stats.replication_stats.provider_available,
         cluster_complete: bucket_stats.replication_stats.cluster_complete,
         observed_node_count: bucket_stats.replication_stats.observed_node_count,
@@ -1235,23 +1238,155 @@ fn build_mrf_response(
     }
 }
 
+/// One durable MRF backlog entry rendered for the default (madmin-compatible)
+/// stream. Field names are the exact json tags of madmin-go `ReplicationMRF`
+/// (replication-api.go), which `mc replicate backlog` decodes one JSON
+/// document at a time. `Size` and `TargetARNs` are RustFS extension keys with
+/// no madmin counterpart; Go decoders ignore unknown keys.
+#[derive(Debug, Serialize)]
+struct MrfEntryDocument {
+    /// The durable backlog is a cluster-shared ledger with no per-node
+    /// attribution, so the madmin `nodeName` tag is always empty.
+    #[serde(rename = "nodeName")]
+    node_name: String,
+    #[serde(rename = "bucket")]
+    bucket: String,
+    #[serde(rename = "object")]
+    object: String,
+    #[serde(rename = "versionId")]
+    version_id: String,
+    #[serde(rename = "retryCount")]
+    retry_count: i32,
+    #[serde(rename = "Size")]
+    size: i64,
+    #[serde(rename = "TargetARNs", skip_serializing_if = "Vec::is_empty")]
+    target_arns: Vec<String>,
+}
+
+/// Upper bound on the number of documents one stream response emits. The
+/// durable ledger is not bounded by the in-memory pending cap (recovery can
+/// persist far larger generations), and the body is buffered before send, so
+/// an unbounded read could stage hundreds of MB per request. A truncated
+/// stream is signalled via `x-rustfs-replication-mrf-truncated`.
+const REPLICATION_MRF_MAX_STREAM_ENTRIES: usize = 10_000;
+
+/// Project the durable backlog into madmin `ReplicationMRF` documents,
+/// scoped to `bucket` when it is non-empty (madmin allows an empty bucket to
+/// mean "across all buckets"), bounded by
+/// [`REPLICATION_MRF_MAX_STREAM_ENTRIES`]. Returns the documents and whether
+/// the backlog was truncated.
+fn mrf_entry_documents(
+    bucket: &str,
+    durable: &crate::admin::storage_api::replication::DurableMrfBacklog,
+) -> (Vec<MrfEntryDocument>, bool) {
+    let mut documents = Vec::new();
+    let mut truncated = false;
+    for entry in durable
+        .entries
+        .iter()
+        .filter(|entry| bucket.is_empty() || entry.bucket == bucket)
+    {
+        if documents.len() >= REPLICATION_MRF_MAX_STREAM_ENTRIES {
+            truncated = true;
+            break;
+        }
+        documents.push(MrfEntryDocument {
+            node_name: String::new(),
+            bucket: entry.bucket.clone(),
+            object: entry.object.clone(),
+            // Delete-marker purge entries track the marker version separately;
+            // fall back to it so those rows still carry a version identity.
+            // The nil UUID is RustFS's in-memory null-version sentinel and
+            // must leave as the S3 wire token, not a zero UUID.
+            version_id: entry
+                .version_id
+                .or(entry.delete_marker_version_id)
+                .map(|v| {
+                    if v.is_nil() {
+                        rustfs_filemeta::NULL_VERSION_ID.to_string()
+                    } else {
+                        v.to_string()
+                    }
+                })
+                .unwrap_or_default(),
+            retry_count: entry.retry_count,
+            size: entry.size,
+            target_arns: entry.target_arns.clone(),
+        });
+    }
+    (documents, truncated)
+}
+
+/// Render the MRF backlog as a response body.
+///
+/// Default (madmin-compatible) mode emits one `ReplicationMRF` JSON document
+/// per line with no envelope — madmin's `BucketReplicationMRF` reads the body
+/// with a `json.Decoder` loop, so an envelope object would decode as a single
+/// entry whose `"Bucket"` key case-insensitively matches
+/// `ReplicationMRF.Bucket` (a phantom row in `mc replicate backlog`), and an
+/// empty backlog must render an empty body so the loop ends on io.EOF with
+/// zero rows.
+///
+/// `aggregate=true` (RustFS extension) keeps the enveloped counter shape;
+/// backlog-source health (`RuntimeStatsAvailable`/`DurableBacklogAvailable`)
+/// is only representable there — an unreadable ledger fails the stream
+/// request outright in the handler (madmin only decodes the body of a 200,
+/// so an empty stream would read as a healthy zero-row backlog).
+fn render_mrf_backlog(
+    response: &MrfResponse,
+    durable: &crate::admin::storage_api::replication::DurableMrfBacklog,
+    aggregate: bool,
+) -> Result<(Vec<u8>, bool), serde_json::Error> {
+    if aggregate {
+        return Ok((serde_json::to_vec(response)?, false));
+    }
+
+    let (documents, truncated) = mrf_entry_documents(&response.bucket, durable);
+    let mut data = Vec::new();
+    for entry in documents {
+        serde_json::to_writer(&mut data, &entry)?;
+        data.push(b'\n');
+    }
+    Ok((data, truncated))
+}
+
 /// `GET /v3/replication/mrf`
 ///
 /// Reports the failed-replication backlog (MinIO's MRF concept) for a bucket.
 ///
-/// Compatibility note: MinIO returns a stream of individual MRF entries. RustFS
-/// deliberately returns aggregate runtime and durable counters instead.
-/// `PerObjectEntriesAvailable` remains false until an enumerable API exists.
-/// `PerTargetDurableEntriesAvailable` is false when the durable backlog includes
-/// older entries that cannot be attributed to a target.
+/// The default response is a madmin-compatible stream of `ReplicationMRF`
+/// documents built from the durable backlog ledger (in-memory failures that
+/// have not been flushed yet — the persister runs every few seconds — are not
+/// visible). `?aggregate=true` (RustFS extension) returns the enveloped
+/// runtime + durable counter shape instead; `PerTargetDurableEntriesAvailable`
+/// is false there when the durable backlog includes older entries that cannot
+/// be attributed to a target.
+///
+/// The madmin `node` parameter is accepted but has no filtering effect: the
+/// durable ledger is cluster-shared with no per-node attribution, so every
+/// node serves the same (complete) backlog.
+///
+/// Authorization: the stream requires `admin:ReplicationDiff` (it enumerates
+/// object names and version ids, MinIO parity); `?aggregate=true` carries no
+/// object identities and requires only `admin:GetReplicationMetrics`.
 pub struct ReplicationMrfHandler {}
 
 #[async_trait::async_trait]
 impl Operation for ReplicationMrfHandler {
     async fn call(&self, req: S3Request<Body>, _params: Params<'_, '_>) -> S3Result<S3Response<(StatusCode, Body)>> {
-        validate_replication_admin_request(&req, AdminAction::GetReplicationMetricsAction).await?;
-
         let queries = extract_query_params(&req.uri);
+        let aggregate = queries.get("aggregate").map(String::as_str) == Some("true");
+        // The default stream enumerates object names and version ids, which
+        // a metrics-only principal must not see; gate it on the same action
+        // MinIO uses for this endpoint. The aggregate counters carry no
+        // object identities and keep the metrics action.
+        let action = if aggregate {
+            AdminAction::GetReplicationMetricsAction
+        } else {
+            AdminAction::ReplicationDiff
+        };
+        validate_replication_admin_request(&req, action).await?;
+
         let Some(bucket) = queries.get("bucket").filter(|b| !b.is_empty()).cloned() else {
             return Err(s3_error!(InvalidRequest, "bucket is required"));
         };
@@ -1275,14 +1410,47 @@ impl Operation for ReplicationMrfHandler {
             return Err(ApiError::from(err).into());
         }
 
+        if let Some(node) = queries.get("node").filter(|node| !node.is_empty() && node.as_str() != "all") {
+            // The durable backlog ledger is cluster-shared with no per-node
+            // attribution, so a node-scoped request still sees the complete
+            // (superset) backlog.
+            debug!(node = %node, "replication mrf node filter has no effect on the cluster-shared backlog");
+        }
+
         let durable = crate::admin::storage_api::replication::read_durable_mrf_backlog(store).await;
         let bucket_stats = cluster_replication_stats(&bucket, app_context_from_req(&req)).await;
-        let response = build_mrf_response(bucket, &bucket_stats, durable);
+        let response = build_mrf_response(bucket, &bucket_stats, &durable);
 
-        let data = serde_json::to_vec(&response)
+        if !durable.available && !aggregate {
+            // The madmin stream has no envelope to carry source health, and
+            // madmin only decodes the body of a 200 — an empty stream would
+            // read as a clean, healthy zero-row backlog. Fail loudly instead;
+            // aggregate mode still reports the availability fields.
+            tracing::warn!(
+                bucket = %response.bucket,
+                "durable MRF backlog is unreadable; failing the stream request — use aggregate=true to see source health"
+            );
+            return Err(S3Error::with_message(
+                S3ErrorCode::ServiceUnavailable,
+                "durable MRF backlog is unreadable; retry, or use aggregate=true for source health".to_string(),
+            ));
+        }
+
+        let (data, truncated) = render_mrf_backlog(&response, &durable, aggregate)
             .map_err(|e| S3Error::with_message(S3ErrorCode::InternalError, format!("serialize failed: {e}")))?;
         let mut headers = HeaderMap::new();
         headers.insert(CONTENT_TYPE, HeaderValue::from_static("application/json"));
+        if truncated {
+            // The madmin stream has no envelope to carry truncation; signal
+            // it out-of-band (madmin/mc ignore unknown headers), mirroring
+            // x-rustfs-replication-diff-truncated.
+            tracing::warn!(
+                bucket = %response.bucket,
+                max_entries = REPLICATION_MRF_MAX_STREAM_ENTRIES,
+                "replication mrf stream truncated; narrow with ?bucket= or drain the backlog"
+            );
+            headers.insert("x-rustfs-replication-mrf-truncated", HeaderValue::from_static("true"));
+        }
         Ok(S3Response::with_headers((StatusCode::OK, Body::from(data)), headers))
     }
 }
@@ -1292,7 +1460,8 @@ mod tests {
     use super::{
         REMOTE_TARGET_UNSUPPORTED_FIELDS, REMOTE_TARGET_WRITABLE_FIELDS, RemoteTargetCredentialsRequest, RemoteTargetRequest,
         ReplicationDiffEntry, SUPPORTED_REMOTE_TARGET_API, TargetUpdateOp, build_mrf_response, extract_query_params,
-        parse_remote_target_update_ops, render_replication_diff, unique_replication_peers, validate_remote_target_tls_settings,
+        parse_remote_target_update_ops, render_mrf_backlog, render_replication_diff, unique_replication_peers,
+        validate_remote_target_tls_settings,
     };
     use crate::admin::storage_api::bucket::target::{BucketTarget, LatencyStat};
     use crate::admin::storage_api::replication::{BucketStats, DurableMrfBacklog, MrfOpKind, MrfReplicateEntry};
@@ -1510,7 +1679,7 @@ mod tests {
             ],
         };
 
-        let response = build_mrf_response("bucket-a".to_string(), &stats, durable);
+        let response = build_mrf_response("bucket-a".to_string(), &stats, &durable);
         let json = serde_json::to_value(response).expect("MRF response should serialize");
 
         assert_eq!(json["TotalFailedCount"], 3);
@@ -1522,7 +1691,9 @@ mod tests {
         assert_eq!(json["RuntimeStatsAvailable"], true);
         assert_eq!(json["ClusterComplete"], false);
         assert_eq!(json["Targets"][0]["ObservationScope"], "partial_cluster");
-        assert_eq!(json["PerObjectEntriesAvailable"], false);
+        // The bare stream enumerates the durable backlog per object, so a
+        // readable backlog advertises the enumerable API.
+        assert_eq!(json["PerObjectEntriesAvailable"], true);
         assert_eq!(json["PerTargetDurableEntriesAvailable"], true);
 
         let targets = json["Targets"].as_array().expect("targets should serialize as an array");
@@ -1569,7 +1740,7 @@ mod tests {
             }],
         };
 
-        let response = build_mrf_response("bucket-a".to_string(), &stats, durable);
+        let response = build_mrf_response("bucket-a".to_string(), &stats, &durable);
         let json = serde_json::to_value(response).expect("MRF response should serialize");
 
         assert_eq!(json["DurableBacklogAvailable"], true);
@@ -1587,7 +1758,7 @@ mod tests {
 
     #[test]
     fn mrf_response_distinguishes_unavailable_sources_from_valid_zero() {
-        let unavailable = build_mrf_response("bucket-a".to_string(), &BucketStats::default(), DurableMrfBacklog::default());
+        let unavailable = build_mrf_response("bucket-a".to_string(), &BucketStats::default(), &DurableMrfBacklog::default());
         let unavailable_json = serde_json::to_value(unavailable).expect("unavailable response should serialize");
         assert_eq!(unavailable_json["RuntimeStatsAvailable"], false);
         assert_eq!(unavailable_json["DurableBacklogAvailable"], false);
@@ -1600,7 +1771,7 @@ mod tests {
         let valid_empty = build_mrf_response(
             "bucket-a".to_string(),
             &valid_empty_stats,
-            DurableMrfBacklog {
+            &DurableMrfBacklog {
                 available: true,
                 entries: Vec::new(),
             },
@@ -1613,6 +1784,151 @@ mod tests {
         assert_eq!(valid_empty_json["PerTargetDurableEntriesAvailable"], true);
     }
 
+    fn sample_durable_backlog() -> DurableMrfBacklog {
+        DurableMrfBacklog {
+            available: true,
+            entries: vec![
+                MrfReplicateEntry {
+                    bucket: "bucket-a".to_string(),
+                    object: "object-a".to_string(),
+                    version_id: Some(uuid::Uuid::from_u128(7)),
+                    retry_count: 2,
+                    size: 250,
+                    op: MrfOpKind::Object,
+                    target_arns: vec!["arn-a".to_string()],
+                    ..Default::default()
+                },
+                MrfReplicateEntry {
+                    bucket: "other-bucket".to_string(),
+                    object: "object-b".to_string(),
+                    version_id: None,
+                    retry_count: 0,
+                    size: 999,
+                    op: MrfOpKind::Object,
+                    target_arns: Vec::new(),
+                    ..Default::default()
+                },
+            ],
+        }
+    }
+
+    /// madmin's `BucketReplicationMRF` decodes the body one `ReplicationMRF`
+    /// JSON document at a time; the default response must therefore be a bare
+    /// document stream with madmin's exact json tags, not an envelope.
+    #[test]
+    fn mrf_stream_renders_bare_madmin_documents() {
+        let durable = sample_durable_backlog();
+        let response = build_mrf_response("bucket-a".to_string(), &BucketStats::default(), &durable);
+
+        let (body, _) = render_mrf_backlog(&response, &durable, false).expect("stream body should serialize");
+        let text = String::from_utf8(body).expect("body should be utf-8");
+        let lines: Vec<&str> = text.lines().filter(|line| !line.trim().is_empty()).collect();
+
+        // Only the entry matching the requested bucket is streamed.
+        assert_eq!(lines.len(), 1, "expected one MRF document, got: {text}");
+        let doc: serde_json::Value = serde_json::from_str(lines[0]).expect("each line should be a JSON document");
+        assert_eq!(doc["bucket"], "bucket-a");
+        assert_eq!(doc["object"], "object-a");
+        assert_eq!(doc["versionId"], uuid::Uuid::from_u128(7).to_string());
+        assert_eq!(doc["retryCount"], 2);
+        // madmin `ReplicationMRF` has a `nodeName` tag; the durable backlog is
+        // cluster-shared, so RustFS reports an empty node name.
+        assert_eq!(doc["nodeName"], "");
+        // The envelope keys must not leak into the stream: a `"Bucket"` key
+        // would case-insensitively populate `ReplicationMRF.Bucket` and render
+        // a phantom row in `mc replicate backlog`.
+        assert!(doc.get("Bucket").is_none());
+        assert!(doc.get("Targets").is_none());
+    }
+
+    /// An empty backlog must produce an empty body: madmin's decoder loop then
+    /// terminates on io.EOF with zero rows instead of one phantom row.
+    #[test]
+    fn mrf_stream_renders_empty_body_for_no_entries() {
+        let durable = DurableMrfBacklog {
+            available: true,
+            entries: Vec::new(),
+        };
+        let response = build_mrf_response("bucket-a".to_string(), &BucketStats::default(), &durable);
+
+        let (body, _) = render_mrf_backlog(&response, &durable, false).expect("stream body should serialize");
+        assert!(
+            body.is_empty(),
+            "empty backlog must serialize to an empty body, got: {}",
+            String::from_utf8_lossy(&body)
+        );
+    }
+
+    /// `?aggregate=true` (RustFS extension) keeps the enveloped counter shape.
+    #[test]
+    fn mrf_aggregate_envelope_retains_counters() {
+        let durable = sample_durable_backlog();
+        let response = build_mrf_response("bucket-a".to_string(), &BucketStats::default(), &durable);
+
+        let (body, _) = render_mrf_backlog(&response, &durable, true).expect("aggregate body should serialize");
+        let json: serde_json::Value = serde_json::from_slice(&body).expect("aggregate body should be one JSON object");
+        assert_eq!(json["Bucket"], "bucket-a");
+        assert_eq!(json["DurableCount"], 1);
+        assert_eq!(json["DurableBacklogAvailable"], true);
+        // The bare stream is an enumerable per-object API, so the aggregate
+        // shell now truthfully advertises it whenever the backlog is readable.
+        assert_eq!(json["PerObjectEntriesAvailable"], true);
+    }
+
+    /// The nil UUID is RustFS's in-memory null-version sentinel; the wire
+    /// token is `null`, never the zero UUID (second review round).
+    #[test]
+    fn mrf_stream_maps_nil_version_to_null_token() {
+        let durable = DurableMrfBacklog {
+            available: true,
+            entries: vec![MrfReplicateEntry {
+                bucket: "bucket-a".to_string(),
+                object: "null-version-object".to_string(),
+                version_id: Some(uuid::Uuid::nil()),
+                retry_count: 1,
+                size: 10,
+                op: MrfOpKind::Object,
+                ..Default::default()
+            }],
+        };
+        let response = build_mrf_response("bucket-a".to_string(), &BucketStats::default(), &durable);
+
+        let (body, truncated) = render_mrf_backlog(&response, &durable, false).expect("stream body should serialize");
+        assert!(!truncated);
+        let doc: serde_json::Value =
+            serde_json::from_str(String::from_utf8(body).expect("utf-8").lines().next().expect("one line"))
+                .expect("line should be a JSON document");
+        assert_eq!(doc["versionId"], "null");
+    }
+
+    /// The durable ledger is not bounded by the in-memory pending cap; the
+    /// stream must stop at the documented bound and signal truncation
+    /// (second review round).
+    #[test]
+    fn mrf_stream_truncates_at_the_documented_bound() {
+        let entries = (0..super::REPLICATION_MRF_MAX_STREAM_ENTRIES + 1)
+            .map(|index| MrfReplicateEntry {
+                bucket: "bucket-a".to_string(),
+                object: format!("object-{index}"),
+                retry_count: 1,
+                op: MrfOpKind::Object,
+                ..Default::default()
+            })
+            .collect();
+        let durable = DurableMrfBacklog {
+            available: true,
+            entries,
+        };
+        let response = build_mrf_response("bucket-a".to_string(), &BucketStats::default(), &durable);
+
+        let (body, truncated) = render_mrf_backlog(&response, &durable, false).expect("stream body should serialize");
+        assert!(truncated, "one entry past the bound must signal truncation");
+        assert_eq!(
+            String::from_utf8(body).expect("utf-8").lines().count(),
+            super::REPLICATION_MRF_MAX_STREAM_ENTRIES
+        );
+    }
+
     #[test]
     fn test_extract_query_params_decodes_percent_encoded_values() {
         let uri: Uri = "/rustfs/admin/v3/list-remote-targets?bucket=foo%2Fbar&flag=a+b"

rustfs/src/admin/handlers/site_replication.rs

@@ -1067,9 +1067,13 @@ fn parse_site_replication_state(data: &[u8]) -> S3Result<SiteReplicationState> {
     state.peers = normalize_peer_map_by_identity(state.peers);
     // A peer-edit high-water mark only fences a CURRENT peer. A site that
     // leaves drops below two peers, which clears its own state object and
-    // restarts its generation counter at zero — a mark left over from the
-    // previous membership would then reject every edit it sends after it
-    // rejoins. Dropping departed origins on load also keeps the map bounded.
+    // restarts its generation counter — a mark left over from the previous
+    // membership must not reject the edits it sends after it rejoins. This
+    // pruning covers departures THIS site observed; an origin removed
+    // unilaterally elsewhere stays in this peer map with its mark, and the
+    // wall-clock floor in `next_peer_edit_generation` is what lifts its
+    // restarted counter over that mark. Dropping departed origins on load
+    // also keeps the map bounded.
     state
         .applied_edit_generations
         .retain(|origin, _| state.peers.contains_key(origin));
@@ -5935,11 +5939,51 @@ fn summarize_peer_error_detail(detail: &str) -> String {
     summary
 }
 
-/// Allocate the next peer-edit generation. Called inside the state
-/// transaction, so the counter is handed out under the distributed
-/// state-object lock and two nodes of this site can never take the same one.
+/// The wall clock in unix nanoseconds, clamped into u64. A pre-1970 (or
+/// post-2554) clock yields 0, which makes the hybrid allocation below
+/// degrade to the plain `previous + 1` counter — monotone, never panicking.
+fn edit_generation_wall_clock() -> u64 {
+    u64::try_from(OffsetDateTime::now_utc().unix_timestamp_nanos()).unwrap_or(0)
+}
+
+/// Allocate the next peer-edit generation as a hybrid logical clock:
+/// `max(wall clock in unix nanoseconds, previous + 1)`. Called inside the
+/// state transaction, so the value is handed out under the distributed
+/// state-object lock and two nodes of this site can never take the same one
+/// (`previous + 1` keeps the sequence strictly increasing even when two
+/// allocations land in one clock tick, and keeps it monotone on a node
+/// whose clock stepped backwards mid-lifetime).
+///
+/// The wall-clock floor is what survives the counter's death. A site
+/// removed while unreachable — the receiver never dropped it from its peer
+/// map, so the load-time mark pruning in `parse_site_replication_state`
+/// never fired — that later rejoins recreates its state object with the
+/// counter back at zero. A plain counter would then hand out generations
+/// below the receiver's stale high-water mark and every delivery would be
+/// silently fenced until the counter caught up. Jumping to wall time clears
+/// that mark: every value the deleted lifetime handed out was capped by the
+/// wall clock at its own allocation (or by a prior lifetime's cap, applied
+/// inductively), so the recreated lifetime's first allocation exceeds them
+/// all — while a pre-removal delivery still in flight stays below the new
+/// floor and remains correctly fenced. Marks recorded by pre-hybrid
+/// receivers (small plain-counter values) sit far below any wall-clock
+/// value, so a restarted origin passes those too — the fix needs only the
+/// sender upgraded, nothing on the wire or in the receiver changed.
+///
+/// A wall clock that regresses across a delete/recreate (the recreating
+/// node's clock behind the clock that fed the previous lifetime) mints
+/// below the stale mark and the origin stays fenced — but only until real
+/// time passes the previous lifetime's last allocation, because every later
+/// allocation takes the wall-clock floor again. Bounded by the skew,
+/// self-healing, and no rollback window beyond the plain counter's: a
+/// delivery applies only at or above the receiver's mark, so the one
+/// cross-lifetime interleaving that can apply stale content — a
+/// pre-removal delivery whose generation lands above everything the
+/// regressed new lifetime has minted — required the same straggler landing
+/// above the mark under the plain counter, where the recreated counter's
+/// low restart made it strictly easier to hit.
 fn next_peer_edit_generation(state: &mut SiteReplicationState) -> u64 {
-    state.edit_generation = state.edit_generation.saturating_add(1);
+    state.edit_generation = edit_generation_wall_clock().max(state.edit_generation.saturating_add(1));
     state.edit_generation
 }
 
@@ -13244,6 +13288,104 @@ mod tests {
         assert!(!peer_edit_delivery_is_stale(&reloaded, "origin-site", 1));
     }
 
+    /// The unilateral-removal rejoin gap the hybrid clock closes. The origin
+    /// was removed while unreachable, but THIS site never dropped it from
+    /// its peer map, so the load-time mark pruning never fired and the mark
+    /// from the previous membership survives. The origin's recreated state
+    /// object restarts its counter, and with a plain `previous + 1` counter
+    /// every delivery it sent — generations 1, 2, … below the stale mark —
+    /// would be silently acked-and-dropped until the counter caught up. The
+    /// wall-clock floor in `next_peer_edit_generation` lifts the restarted
+    /// counter over every value the deleted lifetime handed out. Reverting
+    /// the allocation to the plain counter (dropping the wall-clock max)
+    /// turns the not-stale assertion red.
+    #[test]
+    fn hybrid_generation_unfences_a_rejoined_origin_whose_counter_restarted() {
+        // First lifetime of the origin's state object: two allocations, both
+        // capped by the wall clock at their own allocation.
+        let mut first_life = SiteReplicationState::default();
+        let straggler = next_peer_edit_generation(&mut first_life);
+        let last_applied = next_peer_edit_generation(&mut first_life);
+        assert!(last_applied > straggler, "allocations must be strictly increasing");
+
+        // The receiver applied up to `last_applied` and keeps the origin in
+        // its peer map across the unilateral removal — reloading must keep
+        // the mark, which is exactly why pruning cannot cover this case.
+        let mut receiver = SiteReplicationState::default();
+        receiver.peers.insert(
+            "origin-site".to_string(),
+            PeerInfo {
+                deployment_id: "origin-site".to_string(),
+                ..peer("origin", "https://origin.example:9000")
+            },
+        );
+        record_applied_peer_edit_generation(&mut receiver, "origin-site", last_applied);
+        let mut receiver = parse_site_replication_state(&serde_json::to_vec(&receiver).expect("serialize")).expect("reload");
+        assert_eq!(receiver.applied_edit_generations.get("origin-site"), Some(&last_applied));
+
+        // The origin rejoins with a RECREATED state object: counter back at
+        // zero. The wall-clock floor must lift its first allocation over the
+        // previous lifetime's mark…
+        let mut second_life = SiteReplicationState::default();
+        let restarted = next_peer_edit_generation(&mut second_life);
+        assert!(
+            !peer_edit_delivery_is_stale(&receiver, "origin-site", restarted),
+            "the recreated lifetime's first allocation ({restarted}) must not be fenced by the previous lifetime's mark ({last_applied})"
+        );
+        record_applied_peer_edit_generation(&mut receiver, "origin-site", restarted);
+
+        // …while a pre-removal delivery still in flight stays below the new
+        // floor and remains correctly fenced — the rollback the fence exists
+        // to reject.
+        assert!(
+            peer_edit_delivery_is_stale(&receiver, "origin-site", straggler),
+            "a pre-removal in-flight delivery ({straggler}) must stay fenced after the rejoin"
+        );
+    }
+
+    /// Marks recorded before the hybrid clock existed are small plain-counter
+    /// values, far below any wall-clock allocation: a restarted origin passes
+    /// them as soon as the SENDER runs the hybrid clock — nothing changes on
+    /// the wire or in the receiver, so pre-hybrid receivers get the fix too.
+    /// The other direction is unchanged: among plain-counter values the
+    /// generation order still fences the delivery that lost the race.
+    #[test]
+    fn hybrid_generation_passes_marks_recorded_by_plain_counter_receivers() {
+        let mut receiver = SiteReplicationState::default();
+        record_applied_peer_edit_generation(&mut receiver, "origin-site", 57);
+        assert!(peer_edit_delivery_is_stale(&receiver, "origin-site", 56));
+        assert!(!peer_edit_delivery_is_stale(&receiver, "origin-site", 57));
+
+        let mut rejoined = SiteReplicationState::default();
+        let restarted = next_peer_edit_generation(&mut rejoined);
+        assert!(
+            !peer_edit_delivery_is_stale(&receiver, "origin-site", restarted),
+            "a wall-clock allocation ({restarted}) must clear a plain-counter mark (57)"
+        );
+    }
+
+    /// The `previous + 1` half of the hybrid clock: allocations stay strictly
+    /// increasing even when the wall clock cannot move them forward — two
+    /// allocations inside one clock tick, or a clock that stepped backwards
+    /// mid-lifetime (a counter already ahead of the wall clock advances by
+    /// exactly one per allocation instead of jumping back). Dropping the
+    /// `previous + 1` half (allocating bare wall time) turns this red.
+    #[test]
+    fn hybrid_generation_is_strictly_increasing_when_the_clock_stalls() {
+        let mut state = SiteReplicationState {
+            // A counter far ahead of any wall clock this test will see.
+            edit_generation: u64::MAX / 2,
+            ..Default::default()
+        };
+        assert_eq!(next_peer_edit_generation(&mut state), u64::MAX / 2 + 1);
+        assert_eq!(next_peer_edit_generation(&mut state), u64::MAX / 2 + 2);
+        // Saturation pins at the ceiling instead of wrapping; the equal-value
+        // escape (`applied > generation` is false for equal) keeps deliveries
+        // applying rather than fencing the origin out.
+        state.edit_generation = u64::MAX;
+        assert_eq!(next_peer_edit_generation(&mut state), u64::MAX);
+    }
+
     #[test]
     fn test_retry_stats_for_state_counts_pending_and_failed() {
         let state = SiteReplicationState {
@@ -16044,10 +16186,77 @@ mod tests {
             generations.len(),
             "two nodes took the same edit generation, so their deliveries cannot be ordered: {generations:?}"
         );
+        // The hybrid clock allocates `max(wall nanos, previous + 1)` — the
+        // persisted counter is the largest allocation, and the `+ 1` half
+        // keeps allocations distinct even inside one clock tick.
+        assert_eq!(
+            Some(&load_site_replication_state().await.expect("reload").edit_generation),
+            unique.last(),
+            "the persisted counter must be the largest allocation handed out"
+        );
+    }
+
+    /// The unilateral-removal rejoin, end to end across the state object's
+    /// real lifecycle: dropping below two peers clears the object (the
+    /// counter dies with it), and the recreated object's first allocation —
+    /// raced by two nodes — must clear the previous lifetime's values via
+    /// the wall-clock floor, so a receiver still holding the old mark
+    /// accepts the restarted counter instead of fencing it.
+    #[tokio::test(flavor = "multi_thread", worker_threads = 4)]
+    #[serial]
+    async fn test_recreated_state_object_allocates_over_the_previous_lifetimes_mark() {
+        publish_ready_iam_context().await;
+        let seed = || SiteReplicationState {
+            peers: ["site-a", "site-b"]
+                .into_iter()
+                .map(|name| (name.to_string(), peer(name, &format!("https://{name}.example:9000"))))
+                .collect(),
+            ..Default::default()
+        };
+
+        save_site_replication_state(&seed()).await.expect("seed state");
+        let straggler = update_site_replication_state(|state| Ok(next_peer_edit_generation(state)))
+            .await
+            .expect("first-life allocation");
+        let last_applied = update_site_replication_state(|state| Ok(next_peer_edit_generation(state)))
+            .await
+            .expect("first-life allocation");
+        // A receiver that never dropped this site from its peer map holds
+        // this mark across the removal.
+        let mut receiver = SiteReplicationState::default();
+        record_applied_peer_edit_generation(&mut receiver, "origin-site", last_applied);
+
+        // Unilateral removal: the site drops below two peers, which clears
+        // its state object and the counter with it.
+        let mut departed = seed();
+        departed.peers.remove("site-b");
+        save_site_replication_state(&departed).await.expect("clear state");
         assert_eq!(
             load_site_replication_state().await.expect("reload").edit_generation,
-            generations.len() as u64,
-            "the persisted counter must account for every allocation"
+            0,
+            "clearing the state object must take the counter with it"
+        );
+
+        // Rejoin recreates the state object; two nodes race the first
+        // allocation of the new life.
+        save_site_replication_state(&seed()).await.expect("recreate state");
+        let node_a = tokio::spawn(update_site_replication_state(|state| Ok(next_peer_edit_generation(state))));
+        let node_b = tokio::spawn(update_site_replication_state(|state| Ok(next_peer_edit_generation(state))));
+        let generation_a = node_a.await.expect("node a task").expect("node a allocation");
+        let generation_b = node_b.await.expect("node b task").expect("node b allocation");
+        assert_ne!(generation_a, generation_b, "racing allocations must stay distinct");
+
+        // The receiver's stale mark must not fence the restarted counter…
+        let restarted = generation_a.min(generation_b);
+        assert!(
+            !peer_edit_delivery_is_stale(&receiver, "origin-site", restarted),
+            "the recreated life's first allocation ({restarted}) must clear the previous life's mark ({last_applied})"
+        );
+        record_applied_peer_edit_generation(&mut receiver, "origin-site", restarted);
+        // …while the cleared life's in-flight leftovers stay fenced.
+        assert!(
+            peer_edit_delivery_is_stale(&receiver, "origin-site", straggler),
+            "a pre-removal in-flight delivery ({straggler}) must stay fenced after the rejoin"
         );
     }
 

rustfs/src/admin/route_policy.rs

@@ -1459,10 +1459,13 @@ pub const ADMIN_ROUTE_POLICY_SPECS: &[AdminRouteSpec] = &[
         REPLICATION_DIFF,
         RouteRiskLevel::Sensitive,
     ),
+    // The default stream enumerates object names/version ids and requires
+    // ReplicationDiff (MinIO parity); only ?aggregate=true relaxes to
+    // GetReplicationMetrics in the handler.
     admin(
         HttpMethod::Get,
         "/rustfs/admin/v3/replication/mrf",
-        GET_REPLICATION_METRICS,
+        REPLICATION_DIFF,
         RouteRiskLevel::Sensitive,
     ),
 ];

scripts/security/check_preview_release_workflow.sh

@@ -195,6 +195,13 @@ IFS= read -r -d '' expected_docker_automatic_guard <<'EOF' || true
 EOF
 expected_docker_automatic_guard=${expected_docker_automatic_guard%$'\n'}
 require_job_if "$docker_workflow" "build-check" "$expected_docker_automatic_guard"
+require_line "$docker_workflow" '      source_ref: ${{ steps.check.outputs.source_ref }}' "Docker source ref output"
+require_line "$docker_workflow" '            source_ref="$HEAD_SHA"' "automatic Docker source ref"
+require_line "$docker_workflow" '              source_ref="$tag_ref"' "manual Docker source ref"
+require_line "$docker_workflow" '          ref: ${{ needs.build-check.outputs.source_ref }}' "Docker release source checkout"
+require_line "$docker_workflow" '          SOURCE_REVISION="$(git rev-parse HEAD)"' "Docker source revision resolution"
+require_line "$docker_workflow" '          LABELS="$LABELS,org.opencontainers.image.revision=$SOURCE_REVISION"' "Docker revision label"
+require_absent "$docker_workflow" 'org.opencontainers.image.revision=${{ github.sha }}' "Docker revision must not use the workflow branch SHA"
 
 docker_manual_guard=$(awk '
   $0 == "              *-preview*)" { in_preview = 1 }