Technoboy- opened a new pull request #14644: URL: https://github.com/apache/pulsar/pull/14644
Cherry-pick https://github.com/apache/pulsar/pull/14503 ### Motivation The current exception handler does not clean all states, which will cause some wrong behaviour. **E.g:** #### case 1: https://github.com/apache/pulsar/blob/e20f34bd2a6b4bd7ce72390a593931a855f8e250/pulsar-broker/src/main/java/org/apache/pulsar/broker/service/SystemTopicBasedTopicPoliciesService.java#L249-L264 When line 259 closes the reader with an exception, it doesn't remove namespace from ``policyCacheInitMap``, so next time this method is called, they will do nothing because ``policyCacheInitMap.putIfAbsent(namespace, false)`` is not empty. This will cause `getTopicPolicies` throw TopicPoliciesCacheNotInitException. https://github.com/apache/pulsar/blob/e20f34bd2a6b4bd7ce72390a593931a855f8e250/pulsar-broker/src/main/java/org/apache/pulsar/broker/service/SystemTopicBasedTopicPoliciesService.java#L193-L205 #### case 2: When line 384 gets ``AlreadyClosedException``, it doesn't remove namespace from ``policyCacheInitMap``, but the current namespace value in ``policyCacheInitMap`` is true. Therefore, the next time ``prepareInitPoliciesCache `` is called, a new reader is never created. This will cause `getTopicPolicies` get NULL value. https://github.com/apache/pulsar/blob/e20f34bd2a6b4bd7ce72390a593931a855f8e250/pulsar-broker/src/main/java/org/apache/pulsar/broker/service/SystemTopicBasedTopicPoliciesService.java#L376-L393 ### Modifications - Clean state when the reader gets an error. - Don't clean ``ownedBundlesCountPerNamespace`` when reader get ``AlreadyClosedException``. ### Verifying this change - [x] Make sure that the change passes the CI checks. ### Documentation - [x] `no-need-doc` -- 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]
