fix(site-replication): merge incoming ILM expiry documents instead of overwriting

The lc-config receiver replaced the whole local lifecycle config with
the peer's document (and deleted it wholesale on peer delete), so an
expiry-only update erased the receiver's local tier/transition rules,
and a peer's transition rules were installed across sites
(backlog#1675 P1-1).

Receiver (apply_bucket_meta_item):
- lc-config now merges via merge_incoming_lifecycle_config, mirroring
  MinIO's mergeWithCurrentLCConfig with a trust-boundary hardening:
  incoming transition fields are discarded outright; the local
  transition side of a same-id rule is authoritative. A peer delete
  merges with the empty set — pure-expiry rules go away, transition
  rules survive with their expiry side cleared, and only an empty
  result deletes the config file.
- Staleness moves to the expiry axis (config.expiry_updated_at):
  lifecycle_config_updated_at also moves on local transition-only
  edits, which shadowed newer peer expiry updates.
- Receiver-side replicateILMExpiry gate, symmetric with the sender
  hook (previously any peer could install expiry rules while the
  option was off).
- Rule order is deterministic (local order, incoming-new appended), so
  re-delivering the same document is byte-stable and does not rewrite
  bucket metadata per broadcast.

Sender:
- Both admin choke points — the bucket-meta hook and the SRInfo bucket
  entry feeding bootstrap/repair and consistency views — now emit only
  the expiry subset (transition fields stripped, non-expiry rules
  dropped). MinIO receivers install incoming rules verbatim, so
  transition rules must never leave the site. An unparseable local
  config is forwarded unfiltered rather than degraded to a delete.

Not covered here (follow-up): a two-site e2e with a real tier backend
to exercise transition-rule preservation end to end; receiver-side
validate_transition_tier for merged configs.

rustfs/src/admin/handlers/site_replication.rs

@@ -4220,13 +4220,23 @@ pub async fn site_replication_delete_bucket_hook(bucket: &str, force_delete: boo
     broadcast_site_replication_json(&path, &serde_json::json!({})).await
 }
 
-pub async fn site_replication_bucket_meta_hook(item: SRBucketMeta) -> S3Result<()> {
+pub async fn site_replication_bucket_meta_hook(mut item: SRBucketMeta) -> S3Result<()> {
     let Some(runtime) = runtime_site_replication_targets().await? else {
         return Ok(());
     };
     if item.r#type == "lc-config" && !site_replication_state_replicates_ilm_expiry(&runtime.state) {
         return Ok(());
     }
+    if item.r#type == "lc-config" {
+        // Only the expiry subset travels (MinIO peers install incoming rules
+        // verbatim, so transition rules must never leave this site). An empty
+        // subset becomes a delete, which the receiver merges with the empty
+        // set — local transition rules there survive.
+        item.expiry_lc_config = item
+            .expiry_lc_config
+            .and_then(|raw| lifecycle_expiry_subset_xml(raw.as_bytes()))
+            .map(|data| String::from_utf8_lossy(&data).into_owned());
+    }
     broadcast_site_replication_json_with_runtime(
         &runtime,
         "/rustfs/admin/v3/site-replication/peer/bucket-meta",
@@ -4319,7 +4329,12 @@ async fn build_sr_info(state: &SiteReplicationState, local_peer: &PeerInfo) -> S
             entry.sse_config = raw_config_to_base64(&metadata.encryption_config_xml);
             entry.replication_config = raw_config_to_base64(&metadata.replication_config_xml);
             entry.quota_config = raw_config_to_base64(&metadata.quota_config_json);
-            entry.expiry_lc_config = raw_config_to_base64(&metadata.lifecycle_config_xml);
+            // Expiry subset only: this entry feeds both the bootstrap/repair
+            // plan (peers must not receive transition rules) and cross-site
+            // consistency views (transition rules are site-local and would
+            // read as false mismatches).
+            entry.expiry_lc_config =
+                lifecycle_expiry_subset_xml(&metadata.lifecycle_config_xml).and_then(|data| raw_config_to_base64(&data));
             entry.cors_config = raw_config_to_base64(&metadata.cors_config_xml);
             entry.policy_updated_at = maybe_time(metadata.policy_config_updated_at);
             entry.tag_config_updated_at = maybe_time(metadata.tagging_config_updated_at);
@@ -6859,8 +6874,128 @@ fn merge_incoming_lifecycle_config(
     local: Option<s3s::dto::BucketLifecycleConfiguration>,
     updated_at: Option<OffsetDateTime>,
 ) -> Option<s3s::dto::BucketLifecycleConfiguration> {
-    let _ = (local, updated_at);
-    incoming
+    // Incoming rules reduced to their expiry side (trust boundary). Rules
+    // that carry no expiry semantics after the strip — transition-only or
+    // abort-multipart-only — are site-local and are not installed.
+    let mut incoming_by_id: HashMap<String, s3s::dto::LifecycleRule> = HashMap::new();
+    let mut incoming_order: Vec<String> = Vec::new();
+    for mut rule in incoming.into_iter().flat_map(|config| config.rules) {
+        rule.transitions = None;
+        rule.noncurrent_version_transitions = None;
+        if !lifecycle_rule_has_expiry(&rule) {
+            continue;
+        }
+        let id = rule.id.clone().unwrap_or_default();
+        if incoming_by_id.insert(id.clone(), rule).is_none() {
+            incoming_order.push(id);
+        }
+    }
+
+    // Local order first, incoming-new appended: repeated delivery of the same
+    // document is byte-stable, so bucket metadata is written once, not on
+    // every broadcast.
+    let local_expiry_updated_at = local.as_ref().and_then(|config| config.expiry_updated_at.clone());
+    let mut rules: Vec<s3s::dto::LifecycleRule> = Vec::new();
+    for mut rule in local.into_iter().flat_map(|config| config.rules) {
+        let id = rule.id.clone().unwrap_or_default();
+        if let Some(mut incoming_rule) = incoming_by_id.remove(&id) {
+            incoming_order.retain(|pending| pending != &id);
+            // The incoming expiry side wins; the local transition side is
+            // authoritative (MinIO CloneNonTransition + restore).
+            incoming_rule.transitions = rule.transitions.take();
+            incoming_rule.noncurrent_version_transitions = rule.noncurrent_version_transitions.take();
+            rules.push(incoming_rule);
+        } else if lifecycle_rule_has_expiry(&rule) {
+            // Expiry rule dropped upstream: remove a pure-expiry rule, or
+            // strip only the expiry side when a transition side remains.
+            if lifecycle_rule_has_transition(&rule) {
+                rule.expiration = None;
+                rule.noncurrent_version_expiration = None;
+                rule.del_marker_expiration = None;
+                rules.push(rule);
+            }
+        } else {
+            // No expiry semantics (transition-only / abort-mpu-only): not
+            // managed by expiry replication, keep untouched.
+            rules.push(rule);
+        }
+    }
+    for id in incoming_order {
+        if let Some(rule) = incoming_by_id.remove(&id) {
+            rules.push(rule);
+        }
+    }
+
+    if rules.is_empty() {
+        return None;
+    }
+
+    Some(s3s::dto::BucketLifecycleConfiguration {
+        rules,
+        // Stamp the expiry axis so the staleness guard compares expiry
+        // updates against expiry updates; a local transition-only edit moves
+        // only the whole-config timestamp, not this one.
+        expiry_updated_at: updated_at.map(s3s::dto::Timestamp::from).or(local_expiry_updated_at),
+    })
+}
+
+/// True when the rule carries expiry semantics that `replicateILMExpiry` is
+/// allowed to propagate.
+fn lifecycle_rule_has_expiry(rule: &s3s::dto::LifecycleRule) -> bool {
+    rule.expiration.is_some() || rule.noncurrent_version_expiration.is_some() || rule.del_marker_expiration.is_some()
+}
+
+fn lifecycle_rule_has_transition(rule: &s3s::dto::LifecycleRule) -> bool {
+    rule.transitions.as_ref().is_some_and(|transitions| !transitions.is_empty())
+        || rule
+            .noncurrent_version_transitions
+            .as_ref()
+            .is_some_and(|transitions| !transitions.is_empty())
+}
+
+/// Reduce a lifecycle XML document to the expiry subset that is allowed to
+/// travel between sites (what MinIO's sender emits): transition fields are
+/// stripped and rules left with no expiry semantics are dropped. Returns
+/// `None` when nothing remains — the receiver then merges with the empty set,
+/// which is exactly the "no expiry rules here" statement. A document that
+/// fails to parse is forwarded unfiltered (`Some(original)`): the receiver
+/// merge strips it anyway, and turning a local parse error into a `None`
+/// would delete the peers' replicated expiry rules.
+fn lifecycle_expiry_subset_xml(raw: &[u8]) -> Option<Vec<u8>> {
+    if raw.is_empty() {
+        return None;
+    }
+    let config: s3s::dto::BucketLifecycleConfiguration = match deserialize(raw) {
+        Ok(config) => config,
+        Err(err) => {
+            warn!("failed to parse local lifecycle config for expiry replication; forwarding unfiltered: {err}");
+            return Some(raw.to_vec());
+        }
+    };
+    let expiry_updated_at = config.expiry_updated_at.clone();
+    let rules: Vec<s3s::dto::LifecycleRule> = config
+        .rules
+        .into_iter()
+        .filter_map(|mut rule| {
+            rule.transitions = None;
+            rule.noncurrent_version_transitions = None;
+            lifecycle_rule_has_expiry(&rule).then_some(rule)
+        })
+        .collect();
+    if rules.is_empty() {
+        return None;
+    }
+    let subset = s3s::dto::BucketLifecycleConfiguration {
+        rules,
+        expiry_updated_at,
+    };
+    match serialize(&subset) {
+        Ok(data) => Some(data),
+        Err(err) => {
+            warn!("failed to serialize lifecycle expiry subset; forwarding unfiltered: {err}");
+            Some(raw.to_vec())
+        }
+    }
 }
 
 fn replication_rule_deployment_id(rule: &ReplicationRule) -> Option<String> {
@@ -7944,7 +8079,12 @@ async fn apply_bucket_meta_item(item: SRBucketMeta) -> S3Result<()> {
     } else {
         None
     };
-    if let Ok(bucket_meta) = metadata_sys::get(&item.bucket).await {
+    // lc-config staleness is judged on the expiry axis inside its merge block
+    // below: `lifecycle_config_updated_at` moves on local transition-only
+    // edits too, which would shadow newer peer expiry updates.
+    if item.r#type != "lc-config"
+        && let Ok(bucket_meta) = metadata_sys::get(&item.bucket).await
+    {
         let local_updated_at = bucket_meta_local_updated_at(&bucket_meta, config_file);
         if is_stale_update(local_updated_at, incoming_updated_at) {
             return Ok(());
@@ -7986,6 +8126,54 @@ async fn apply_bucket_meta_item(item: SRBucketMeta) -> S3Result<()> {
             None
         };
 
+    let merged_lifecycle_config = if item.r#type == "lc-config" {
+        // Receiver-side gate, symmetric with the sender hook: a peer must not
+        // install expiry rules here while `replicateILMExpiry` is off. When
+        // the state cannot be read, fall through and apply (pre-gate
+        // behavior) rather than silently dropping a legitimate update.
+        if let Ok(state) = load_site_replication_state().await
+            && !site_replication_state_replicates_ilm_expiry(&state)
+        {
+            return Ok(());
+        }
+
+        let incoming = item
+            .expiry_lc_config
+            .as_ref()
+            .map(|raw| {
+                let data = decode_bucket_meta_wire_value(raw);
+                deserialize::<s3s::dto::BucketLifecycleConfiguration>(&data)
+            })
+            .transpose()
+            .map_err(|e| s3_error!(InvalidRequest, "invalid lifecycle config: {e}"))?;
+        let local = match metadata_sys::get_lifecycle_config(&item.bucket).await {
+            Ok((config, _)) => Some(config),
+            Err(StorageError::ConfigNotFound) => None,
+            Err(err) => return Err(ApiError::from(err).into()),
+        };
+        let local_expiry_updated_at = local
+            .as_ref()
+            .and_then(|config| config.expiry_updated_at.clone())
+            .map(OffsetDateTime::from)
+            .unwrap_or(OffsetDateTime::UNIX_EPOCH);
+        if is_stale_update(local_expiry_updated_at, incoming_updated_at) {
+            return Ok(());
+        }
+        let local_absent = local.is_none();
+        match merge_incoming_lifecycle_config(incoming, local, incoming_updated_at) {
+            Some(config) => Some(
+                serialize(&config)
+                    .map_err(|e| S3Error::with_message(S3ErrorCode::InternalError, format!("serialize lifecycle failed: {e}")))?,
+            ),
+            None => {
+                skip_config_write = local_absent;
+                None
+            }
+        }
+    } else {
+        None
+    };
+
     let data = match item.r#type.as_str() {
         "policy" => item
             .policy
@@ -8002,7 +8190,7 @@ async fn apply_bucket_meta_item(item: SRBucketMeta) -> S3Result<()> {
         "object-lock-config" => decode_bucket_meta_wire_option(item.object_lock_config),
         "sse-config" => decode_bucket_meta_wire_option(item.sse_config),
         "replication-config" => merged_replication_config,
-        "lc-config" => decode_bucket_meta_wire_option(item.expiry_lc_config),
+        "lc-config" => merged_lifecycle_config,
         "cors-config" => decode_bucket_meta_wire_option(item.cors),
         _ => unreachable!(),
     };
@@ -14537,6 +14725,36 @@ mod tests {
         assert_eq!(OffsetDateTime::from(stamped).unix_timestamp(), updated_at.unix_timestamp());
     }
 
+    /// Sender-side filter: only the expiry subset leaves this site. MinIO
+    /// peers install incoming rules verbatim, so a full document would plant
+    /// this site's transition rules there.
+    #[test]
+    fn test_lifecycle_expiry_subset_xml_strips_transitions() {
+        let full = serialize(&lc_config(vec![lc_rule("mixed", Some(1), Some(30)), lc_rule("t-only", None, Some(7))]))
+            .expect("serialize full config");
+
+        let subset = lifecycle_expiry_subset_xml(&full).expect("expiry subset should remain");
+        let parsed: s3s::dto::BucketLifecycleConfiguration = deserialize(&subset).expect("subset should parse");
+        assert_eq!(rule_ids(&parsed), vec!["mixed"]);
+        assert!(parsed.rules[0].transitions.is_none(), "transition side must not travel");
+
+        let transition_only =
+            serialize(&lc_config(vec![lc_rule("t-only", None, Some(7))])).expect("serialize transition-only config");
+        assert!(
+            lifecycle_expiry_subset_xml(&transition_only).is_none(),
+            "a transition-only config states 'no expiry rules' (delete semantics)"
+        );
+        assert!(lifecycle_expiry_subset_xml(b"").is_none());
+    }
+
+    /// A local parse failure must forward the document unfiltered — mapping
+    /// it to `None` would delete the peers' replicated expiry rules.
+    #[test]
+    fn test_lifecycle_expiry_subset_xml_forwards_unparseable_config() {
+        let garbage = b"<LifecycleConfiguration><Rule></Broken>";
+        assert_eq!(lifecycle_expiry_subset_xml(garbage).as_deref(), Some(garbage.as_slice()));
+    }
+
     // `role` is part of the bucket's S3-visible configuration. Repairing a reverse rule must
     // drop only a sender-owned site-replication ARN, never an operator's own role — the same
     // rule the merge path applies, so both paths agree on what is ours to rewrite.