You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kafka.apache.org by "ASF GitHub Bot (JIRA)" <ji...@apache.org> on 2017/05/22 00:53:04 UTC
[jira] [Commented] (KAFKA-5186) Avoid expensive initialization of
producer state when upgrading
[ https://issues.apache.org/jira/browse/KAFKA-5186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16019054#comment-16019054 ]
ASF GitHub Bot commented on KAFKA-5186:
---------------------------------------
GitHub user hachikuji opened a pull request:
https://github.com/apache/kafka/pull/3113
KAFKA-5186: Avoid expensive log scan to build producer state when upgrading
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/hachikuji/kafka KAFKA-5186
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/kafka/pull/3113.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #3113
----
commit 90b4051e7984b5611fa8e2395fa3d5643b3c1f0b
Author: Jason Gustafson <ja...@confluent.io>
Date: 2017-05-22T00:51:04Z
KAFKA-5186: Avoid expensive log scan to build producer state when upgrading
----
> Avoid expensive initialization of producer state when upgrading
> ---------------------------------------------------------------
>
> Key: KAFKA-5186
> URL: https://issues.apache.org/jira/browse/KAFKA-5186
> Project: Kafka
> Issue Type: Sub-task
> Components: clients, core, producer
> Reporter: Jason Gustafson
> Assignee: Jason Gustafson
> Priority: Blocker
> Labels: exactly-once
> Fix For: 0.11.0.0
>
>
> Currently the producer state is always loaded upon broker initialization. If we don't find a snapshot file to load from, then we scan the log segments from the beginning to rebuild the state. Of course, when users upgrade to the new version, there will be no snapshot file, so the upgrade could be quite intensive. It would be nice to avoid this by assuming instead that the absence of a snapshot file means that the producer state should start clean and we can avoid the expensive scanning.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)