Kalvin2077 commented on code in PR #3858:
URL: https://github.com/apache/celeborn/pull/3858#discussion_r4488969840
##########
master/src/main/java/org/apache/celeborn/service/deploy/master/clustermeta/AbstractMetaManager.java:
##########
@@ -73,6 +75,8 @@ public abstract class AbstractMetaManager implements
IMetadataHandler {
JavaUtils.newConcurrentHashMap();
public final Set<String> hostnameSet = ConcurrentHashMap.newKeySet();
public final Map<String, WorkerInfo> workersMap =
JavaUtils.newConcurrentHashMap();
+ public final Map<WorkerInfo, AtomicReference<WorkerStats>> workerStats =
+ JavaUtils.newConcurrentHashMap();
Review Comment:
Thanks for the reproduction. I've prepared a fix that replaces each worker's
observation reference on leadership loss and acquisition, while preserving
registration. Heartbeats capture their update reference before the leader
check, so requests from an earlier epoch cannot repopulate the current cache.
The A → B → A regression keeps the worker registered and covers both RPC
endpoints, including a path where no redirect heartbeat reaches A.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
