[GitHub] [kafka] cmccabe commented on a diff in pull request #12181: KAFKA-13916; Fenced replicas should not be allowed to join the ISR in KRaft (KIP-841)

[GitBox <gi...@apache.org>]

cmccabe commented on code in PR #12181:
URL: https://github.com/apache/kafka/pull/12181#discussion_r881916133


##########
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 think it's hard to make a decision about this since we don't have clear numbers (or even empirical data) about how big a problem the extra AlterPartition requests are. We also know that having a shutting down state won't totally avoid all the unecessary AlterPartition requests, just some of them.
   
   Just as a matter of general principle, I would be a lot more tempted to go with something simple like exponential backoff for AlterPartition requests. That works even if the network is messed up, even if we're using ZK mode, etc. etc.
   
   Anyway, @hachikuji , can we agree to consider adding a shutting down state as a possible future improvement, and just get in the current approach first? If you like we can open a JIRA talking about a possible shutting down state, so that we don't lose track of the idea.



-- 
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-unsubscribe@kafka.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org