hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "stack (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HBASE-14392) [tests] TestLogRollingNoCluster fails on master from time to time
Date Thu, 10 Sep 2015 04:08:45 GMT

     [ https://issues.apache.org/jira/browse/HBASE-14392?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

stack updated HBASE-14392:
    Attachment: 14392v2.txt

A better patch.

Allow that two threads can be operating on the stream at the one time, one closing the stream
while another is trying to sync it: e.g. log roll and an errant sync. Then, make it so can't
get an NPE any more. We might get a stream already-closed exception but will deal w/ that
if we start seeing them (trying to untangle what stream is when in hdfs, what flushs and what
hflushes is a puzzle for another time).

> [tests] TestLogRollingNoCluster fails on master from time to time
> -----------------------------------------------------------------
>                 Key: HBASE-14392
>                 URL: https://issues.apache.org/jira/browse/HBASE-14392
>             Project: HBase
>          Issue Type: Bug
>          Components: test
>            Reporter: stack
>            Assignee: stack
>         Attachments: 14392.txt, 14392.txt, 14392v2.txt
> TestLogRollingNoCluster fails from time to time on a rig I have running here.
> Java HotSpot(TM) 64-Bit Server VM warning: ignoring option MaxPermSize=256m; support
was removed in 8.0
> Running org.apache.hadoop.hbase.regionserver.wal.TestLogRollingNoCluster
> Tests run: 1, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 3.155 sec - in org.apache.hadoop.hbase.regionserver.wal.TestLogRollingNoCluster
> Results :
> Tests run: 1, Failures: 0, Errors: 0, Skipped: 0
> The test itself is a bit odd. We apparently have seen NPEs around close of a WAL while
trying to sync... which I suppose makes sense if two threads involved.
> Attached patch just fails the sync silently if stream is null... lets presume it a close.
Adds a sync on write of trailer too... 
> This patch seems to have gotten rid of the odd failure seen on a particular box here
if I keep cycling the test.

This message was sent by Atlassian JIRA

View raw message