[
https://issues.apache.org/jira/browse/HBASE-12607?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14238272#comment-14238272
]
Ted Yu commented on HBASE-12607:
--------------------------------
One possible solution is for HBaseFsck to mark state as waiting to acquire lock
before entering checkAndMarkRunningHbck().
A getter method would be added to HBaseFsck so that we can tell whether the
running HBaseFsck instance is in waiting state.
In the test, having two instances of HBaseFsck is fine as long as one of them
is in waiting state.
[~busbey]: what do you think ?
> TestHBaseFsck#testParallelHbck fails running against hadoop 2.6.0
> -----------------------------------------------------------------
>
> Key: HBASE-12607
> URL: https://issues.apache.org/jira/browse/HBASE-12607
> Project: HBase
> Issue Type: Test
> Reporter: Ted Yu
> Assignee: Ted Yu
> Priority: Minor
> Attachments: 12607-v1.txt,
> org.apache.hadoop.hbase.util.TestHBaseFsck-output.txt
>
>
> TestHBaseFsck#testParallelHbck consistently fails on both Linux and Mac.
> I added more logging as shown below:
> {code}
> HBaseFsck h1 = hbck1.get();
> HBaseFsck h2 = hbck2.get();
> LOG.warn("h1: " + h1);
> LOG.warn("h2: " + h2);
> {code}
> I saw the following in test output:
> {code}
> 2014-12-01 14:58:44,728 WARN [main] util.TestHBaseFsck(576): h1:
> org.apache.hadoop.hbase.util.HBaseFsck@72aa738e
> 2014-12-01 14:58:44,729 WARN [main] util.TestHBaseFsck(577): h2:
> org.apache.hadoop.hbase.util.HBaseFsck@186e2ce6
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
