[jira] [Commented] (HBASE-9657) Race condition in low replication checking and FSHLog#rollWriter()

["Ted Yu (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-9657?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13777882#comment-13777882 ] 

Ted Yu commented on HBASE-9657:
-------------------------------

Patch changes rollWriterLock to be a ReentrantLock.

The patch keeps FSHLog#rollWriter() with priority over low replication checking. Boolean logRollRunning is removed since it is not needed.

18 iterations of TestLogRolling have passed against hadoop-2. Running 82 more iterations.
                
> Race condition in low replication checking and FSHLog#rollWriter()
> ------------------------------------------------------------------
>
>                 Key: HBASE-9657
>                 URL: https://issues.apache.org/jira/browse/HBASE-9657
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Ted Yu
>            Assignee: Ted Yu
>         Attachments: 9657-v1.txt
>
>
> In FSHLog#syncer(), we have this comment:
> {code}
>       // TODO: preserving the old behavior for now, but this check is strange. It's not
>       //       protected by any locks here, so for all we know rolling locks might start
>       //       as soon as we enter the "if". Is this best-effort optimization check?
>       if (!this.logRollRunning) {
>         checkLowReplication();
> {code}
> The implication is that checkLowReplication() may be running when FSHLog#rollWriter() is also running.

--
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