dajac commented on code in PR #12181:
URL: https://github.com/apache/kafka/pull/12181#discussion_r884847649
##########
metadata/src/main/java/org/apache/kafka/controller/BrokerHeartbeatManager.java:
##########
@@ -498,6 +498,23 @@ public UsableBroker next() {
}
}
+ /**
+ * Check if the given broker is active. A broker is active if it is
unfenced,
+ * and not shutting down.
Review Comment:
I am a bit concerned by the controlled shutdown case because there could be
a few seconds before the fenced record in written. The fenced record is written
when the shutting-down broker heartbeats if all the other brokers in the
cluster have caught up to the shutting down offset. I have looked into a few
clusters. It typically takes 2-4 seconds before the controller grants the
controlled shutdown and grants the shutdown. I have seen cases where it took 6
seconds and I suppose that it could take more time depending on the number of
partitions and brokers in the cluster. That is a pretty large time window that
is work optimising for in my opinion.
--
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]
