heesung-sn commented on code in PR #21332:
URL: https://github.com/apache/pulsar/pull/21332#discussion_r1351215592
##########
pulsar-broker/src/main/java/org/apache/pulsar/broker/loadbalance/extensions/channel/ServiceUnitStateChannelImpl.java:
##########
@@ -503,13 +503,22 @@ public CompletableFuture<Optional<String>>
getOwnerAsync(String serviceUnit) {
case Splitting -> {
return
CompletableFuture.completedFuture(Optional.of(data.sourceBroker()));
}
- case Assigning, Releasing -> {
+ case Assigning -> {
return deferGetOwnerRequest(serviceUnit).whenComplete((__, e)
-> {
if (e != null) {
ownerLookUpCounters.get(state).getFailure().incrementAndGet();
}
- }).thenApply(
- broker -> broker == null ? Optional.empty() :
Optional.of(broker));
+ }).thenApply(broker -> broker == null ? Optional.empty() :
Optional.of(broker));
+ }
+ case Releasing -> {
+ if (isTransferCommand(data)) {
+ return deferGetOwnerRequest(serviceUnit).whenComplete((__,
e) -> {
+ if (e != null) {
+
ownerLookUpCounters.get(state).getFailure().incrementAndGet();
+ }
+ }).thenApply(broker -> broker == null ? Optional.empty() :
Optional.of(broker));
+ }
+ return CompletableFuture.completedFuture(Optional.empty());
Review Comment:
Then, isn't your change in UnloadManager.handleEvent sufficient to return
the admin api?
In the test env, I feel like we can configure the monitor frequency
small(like 1 sec) to fix the stuck state quicker.
--
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]
