[
https://issues.apache.org/jira/browse/HDFS-6229?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13966308#comment-13966308
]
Suresh Srinivas commented on HDFS-6229:
---------------------------------------
+1 for the patch. Thanks Jing for working on this.
> Race condition in failover can cause RetryCache fail to work
> ------------------------------------------------------------
>
> Key: HDFS-6229
> URL: https://issues.apache.org/jira/browse/HDFS-6229
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: ha
> Affects Versions: 2.1.0-beta
> Reporter: Jing Zhao
> Assignee: Jing Zhao
> Attachments: HDFS-6229.000.patch, retrycache-race.patch
>
>
> Currently when NN failover happens, the old SBN first sets its state to
> active, then starts the active services (including tailing all the remaining
> editlog and building a complete retry cache based on the editlog). If a retry
> request, which has already succeeded in the old ANN (but the client fails to
> receive the response), comes in between, this retry may still get served by
> the new ANN but miss the retry cache.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
