kafka-jira mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Prasanna Gautam (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (KAFKA-5473) handle ZK session expiration properly when a new session can't be established
Date Fri, 23 Jun 2017 22:51:00 GMT

    [ https://issues.apache.org/jira/browse/KAFKA-5473?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16061610#comment-16061610
] 

Prasanna Gautam commented on KAFKA-5473:
----------------------------------------

[~junrao] Why not do an exponential backoff (with jitter) with an upper bound? If you're temporarily
disconnected, it should recover within a few seconds, otherwise an upper bound before the
broker dies feels like a more sensible solution. This way ZK nodes being network-partitioned
from kafka wouldn't immediately bring down all brokers if its a recoverable issue.
Also, if there's a cleaner way to exit that allows all writes to be synced to disk, that seems
more preferable than System.exit() too.

> handle ZK session expiration properly when a new session can't be established
> -----------------------------------------------------------------------------
>
>                 Key: KAFKA-5473
>                 URL: https://issues.apache.org/jira/browse/KAFKA-5473
>             Project: Kafka
>          Issue Type: Sub-task
>    Affects Versions: 0.9.0.0
>            Reporter: Jun Rao
>            Assignee: Prasanna Gautam
>
> In https://issues.apache.org/jira/browse/KAFKA-2405, we change the logic in handling
ZK session expiration a bit. If a new ZK session can't be established after session expiration,
we just log an error and continue. However, this can leave the broker in a bad state since
it's up, but not registered from the controller's perspective. Replicas on this broker may
never to be in sync.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message