xintongsong commented on a change in pull request #15095:
URL: https://github.com/apache/flink/pull/15095#discussion_r589181652
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/active/ActiveResourceManager.java
##########
@@ -235,6 +241,8 @@ public void
onPreviousAttemptWorkersRecovered(Collection<WorkerType> recoveredWo
for (WorkerType worker : recoveredWorkers) {
final ResourceID resourceId = worker.getResourceID();
workerNodeMap.put(resourceId, worker);
+ previousAttemptUnregisteredWorkers.add(resourceId);
Review comment:
That's a good point.
I don't think a standby RM will re-request workers, because it only
re-request workers when there are unfulfilled resource requirements, which a
standby RM should not have since it doesn't accept any resource requirement
until becoming the leader. However, it is indeed possible that the standby RM
tries to stop the worker. I'll add a check to make sure it only tries to stop
workers when it has the leadership.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
