[
https://issues.apache.org/jira/browse/KAFKA-2406?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14658922#comment-14658922
]
Jun Rao commented on KAFKA-2406:
--------------------------------
[~becket_qin], are you suggesting not using sequential ZK node, but just having
an isrChangePath per broker and keeping updating it? Not sure how this
coordinates with the controller.
Also, not sure having the controller periodically read from the IsrChange path
is better. After the cluster is stable, there shouldn't be any ISR change and
the periodic ZK read is just pure overhead. If we batch enough, there shouldn't
be that many ZK watchers fired. Another optimization that we could do is that
on controller failover, the controller can actually delete all existing
sequential ZK nodes in isr_change before initialization. The initialization
will read the latest leader and ISR for all partitions from ZK anyway.
> ISR propagation should be throttled to avoid overwhelming controller.
> ---------------------------------------------------------------------
>
> Key: KAFKA-2406
> URL: https://issues.apache.org/jira/browse/KAFKA-2406
> Project: Kafka
> Issue Type: Bug
> Reporter: Jiangjie Qin
> Assignee: Jiangjie Qin
> Priority: Blocker
>
> This is a follow up patch for KAFKA-1367.
> We need to throttle the ISR propagation rate to avoid flooding in controller
> to broker traffic. This might significantly increase time of controlled
> shutdown or cluster startup.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
