[
https://issues.apache.org/jira/browse/KAFKA-15580?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18122134#comment-18122134
]
Tom Thornton commented on KAFKA-15580:
--------------------------------------
We (Slack) ran into an issue with our Kafka deployment where having an
Aggressive strategy would have elected the best replica automatically. I
noticed that part of this was
[implemented|https://github.com/apache/kafka/pull/20371] and later
[reverted|https://github.com/apache/kafka/pull/20371]. I published a
[PR|https://github.com/apache/kafka/pull/23680] that takes a smaller approach:
instead of pulling the replica info from brokers (new RPC) it adds in
leaderless partition info (log end offset, last epoch) to the heartbeats
already pushed to controllers regularly. Since push differs from the original
proposal, I'm happy to update the KIP or raise this in the email discussion,
let me know what you all think. Thanks!
> KIP-966: Unclean Recovery
> -------------------------
>
> Key: KAFKA-15580
> URL: https://issues.apache.org/jira/browse/KAFKA-15580
> Project: Kafka
> Issue Type: New Feature
> Reporter: Calvin Liu
> Assignee: Jonah Hooper
> Priority: Major
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
