CCR: Use local cluster state request (#100323) (#100520)

Auto following indices from leader asks for the cluster state waiting for the next version to be available. These requests are no longer being forwarded to the master on the leader cluster, since waiting for the state to appear on the coordinator is equally good and avoids burdening the master with serializing a potentially large cluster state.
elastic · Oct 9, 2023 · 2803d6a · 2803d6a
1 parent 090bcd8
commit 2803d6a
Show file tree

Hide file tree

Showing 2 changed files with 6 additions and 0 deletions.
diff --git a/docs/changelog/100323.yaml b/docs/changelog/100323.yaml
@@ -0,0 +1,5 @@
+pr: 100323
+summary: "CCR: Use local cluster state request"
+area: CCR
+type: bug
+issues: []
diff --git a/...ck/plugin/ccr/src/main/java/org/elasticsearch/xpack/ccr/action/AutoFollowCoordinator.java b/...ck/plugin/ccr/src/main/java/org/elasticsearch/xpack/ccr/action/AutoFollowCoordinator.java
@@ -303,6 +303,7 @@ void getRemoteClusterState(
                         new ClusterStateRequest().clear()
                             .metadata(true)
                             .routingTable(true)
+                            .local(true)
                             .waitForMetadataVersion(metadataVersion)
                             .waitForTimeout(waitForMetadataTimeOut),
                         e -> handler.accept(null, e),