[
https://issues.apache.org/jira/browse/HBASE-16725?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15530547#comment-15530547
]
Appy commented on HBASE-16725:
------------------------------
Maybe put other thread joins also in finally?
{noformat}
putThread.join();
putThread.checkNoError();
{noformat}
Also, we should check for interrupted exception from join() otherwise we'll
fail to close region and wal.
As a side question, how accurate are "potentially hanging thread"
notifications? [~stack] your thoughts please.
> Don't let flushThread hang in TestHRegion
> -----------------------------------------
>
> Key: HBASE-16725
> URL: https://issues.apache.org/jira/browse/HBASE-16725
> Project: HBase
> Issue Type: Test
> Reporter: Ted Yu
> Priority: Minor
> Attachments: 16725.branch-1.v1.txt
>
>
> I was running TestHRegion locally and observed the following in test output:
> {code}
> 2016-09-28 16:29:36,836 INFO [main] hbase.ResourceChecker(171): after:
> regionserver.TestHRegion#testFlushCacheWhileScanning Thread=50 (was 49)
> Potentially hanging thread: FlushThread
> java.lang.Object.wait(Native Method)
> java.lang.Object.wait(Object.java:502)
>
> org.apache.hadoop.hbase.regionserver.TestHRegion$FlushThread.run(TestHRegion.java:3834)
> {code}
> Call to flushThread.done() etc should be placed in the finally block.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
