rondagostino commented on a change in pull request #10003:
URL: https://github.com/apache/kafka/pull/10003#discussion_r568730408
##########
File path: core/src/main/scala/kafka/server/ReplicaManager.scala
##########
@@ -257,6 +264,37 @@ class ReplicaManager(val config: KafkaConfig,
}
}
+ // Changes are initially deferred when using a Raft-based metadata quorum,
and they may flip-flop to not
+ // being deferred and being deferred again thereafter as the broker
(re)acquires/loses its lease.
+ // Changes are never deferred when using ZooKeeper. When true, this
indicates that we should transition
+ // online partitions to the deferred state if we see a metadata update for
that partition.
+ @volatile private var deferringMetadataChanges: Boolean =
!config.requiresZookeeper
Review comment:
Good point. The same thread will be both managing this value and
submitting metadata changes, so it can remember if it has already started
deferring changes if it needs to do so; plus everything here related to
deferral happens within the lock anyway. I removed `@volatile`.
----------------------------------------------------------------
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:
us...@infra.apache.org
