[
https://issues.apache.org/jira/browse/HBASE-8855?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13702120#comment-13702120
]
stack commented on HBASE-8855:
------------------------------
On Apache, hadoop2 builds have actually passed on trunk and 0.95:
https://builds.apache.org/view/H-L/view/HBase/job/HBase-TRUNK-on-Hadoop-2.0.0/599/
https://builds.apache.org/view/H-L/view/HBase/job/HBase-TRUNK-on-Hadoop-2.0.0/595/
https://builds.apache.org/view/H-L/view/HBase/job/hbase-0.95-on-hadoop2/168/
https://builds.apache.org/view/H-L/view/HBase/job/hbase-0.95-on-hadoop2/166/
Apache does not seem to run into the issue we see here.
Buids have never passed on ec2 build though but look here from this morning:
http://54.241.6.143/job/HBase-0.95-Hadoop-2/581/
http://54.241.6.143/job/HBase-0.95-Hadoop-2/580/
We failed in integration tests which means we got through all the unit tests.
I'm not sure that has happened in the past (We don't do integration tests up on
apache build box). So, we almost succeed (Failing integration tests are
another issue being dealt w/ elsewhere -- it is retry and interval based)
Here is what I what I have currently in ec2 build after messing some w/
permsizes:
-Xmx4g -XX:+CMSClassUnloadingEnabled -XX:+CMSPermGenSweepingEnabled
-XX:MaxPermSize=1g
Apache build just has 3G heap set on MAVEN_OPTS. Let me do same over on ec2
build and see if that helps.
> TestTableInputFormatScan1/2 fail semi-silently with the hadoop 2 profile
> ------------------------------------------------------------------------
>
> Key: HBASE-8855
> URL: https://issues.apache.org/jira/browse/HBASE-8855
> Project: HBase
> Issue Type: Bug
> Affects Versions: 0.94.8, 0.95.1
> Reporter: Jean-Daniel Cryans
> Attachments: 8855.patch
>
>
> It looks like TestTableInputFormatScan1 and TestTableInputFormatScan2 never
> complete and surefire doesn't complain about it. Sure, you may see this:
> {noformat}
> Tests run: 6, Failures: 5, Errors: 1, Skipped: 0, Time elapsed: 269.036 sec
> <<< FAILURE!
> org.apache.maven.surefire.util.SurefireReflectionException: ...
> org.apache.maven.surefire.booter.ForkedBooter.main(ForkedBooter.java:74)
> Caused by: java.lang.OutOfMemoryError: PermGen space
> org.apache.maven.surefire.booter.SurefireBooterForkException: Error occurred
> in starting fork, check output in log
> ...
>
> testScanEmptyToBBA(org.apache.hadoop.hbase.mapreduce.TestTableInputFormatScan1)
>
> testScanEmptyToBBB(org.apache.hadoop.hbase.mapreduce.TestTableInputFormatScan1)
>
> testScanEmptyToOPP(org.apache.hadoop.hbase.mapreduce.TestTableInputFormatScan1)
>
> testScanEmptyToEmpty(org.apache.hadoop.hbase.mapreduce.TestTableInputFormatScan1)
> {noformat}
> But then:
> {noformat}
> [INFO] HBase - Server .................................... SUCCESS
> [10:59.929s]
> {noformat}
> This is on my machine. On our local jenkins it's leaking and the processes
> never die. And this is only with Hadoop 2. It also looks like other tests are
> failing with PermGen space.
--
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
