junrao commented on a change in pull request #11255:
URL: https://github.com/apache/kafka/pull/11255#discussion_r696035745
##########
File path: core/src/main/scala/kafka/controller/KafkaController.scala
##########
@@ -630,7 +630,7 @@ class KafkaController(val config: KafkaConfig,
// If replica failure did not require leader re-election, inform brokers
of the offline brokers
// Note that during leader re-election, brokers update their metadata
- if (partitionsWithOfflineLeader.isEmpty) {
+ if (newOfflineReplicas.isEmpty) {
Review comment:
Good catch. There are two cases where we need to send
UpdateMetadataRequest.
1. If the broker doesn't host any replicas. This is the new case that you
identified since we won't send any UpdateMetadataRequest in
PartitionStateMachine and we want to propagate the offline broker to other
brokers.
2. If partitionsWithOfflineLeader is not empty,
partitionStateMachine.triggerOnlinePartitionStateChange() would send
UpdateMetadataRequest as part of the transitioning of those partitions to
online. So, we don't need to send UpdateMetadataRequest again. If
partitionsWithOfflineLeader is empty, we do want to send UpdateMetadataRequest
to propagate the offline broker.
So, it seem that the check should be if (newOfflineReplicas.isEmpty ||
partitionsWithOfflineLeader.isEmpty). It would be useful to adjust the comment
above accordingly.
--
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: jira-unsubscr...@kafka.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
