[
https://issues.apache.org/jira/browse/KAFKA-800?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13600276#comment-13600276
]
Neha Narkhede commented on KAFKA-800:
-------------------------------------
Thanks for the patch, a few questions/comments -
ReplicaManager
1. Like Jun said, we don't need to synchronize on leaderPartitionsLock inside
getOrCreatePartition and getPartition()
2. I don't think we need to synchronize on leaderPartitionsLock inside
checkpointHighWatermarks too, that lock is protecting the leaderPartitions data
structure.
3. In stopReplica(), we need to synchronize only around leaderPartitions, not
allPartitions and the info statement following that.
> inSyncReplica in Partition needs some tweaks
> --------------------------------------------
>
> Key: KAFKA-800
> URL: https://issues.apache.org/jira/browse/KAFKA-800
> Project: Kafka
> Issue Type: Bug
> Affects Versions: 0.8
> Reporter: Sriram Subramanian
> Assignee: Sriram Subramanian
> Priority: Blocker
> Labels: kafka-0.8, p1
> Fix For: 0.8
>
> Attachments: KAFKA-800.patch, KAFKA-800-v2.patch
>
>
> 1. isUnderReplicated method in Partition.scala needs to take the
> leaderIsrUpdateLock before reading the inSyncReplica structure. We suspect
> that this could cause the gauge to read stale values from the cache.
> 2. toString api should take the leaderIsrUpdateLock to prevent anything from
> blowing up when it iterates the data structures
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
