[
https://issues.apache.org/jira/browse/HADOOP-2558?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HADOOP-2558:
--------------------------
Attachment: 2558-v2.patch
TestTableMapReduce failed in patch build #1525 and in #1521 because
verification of the mapreduce job turned up empty cells. As has been
speculated before, it looks like updates have not all come in when scanner is
opened particularly as the two tests fail at different locations (scanners see
state of data at time of their opening). Patch v2 adds retry of verify if we
find empty cells with sleep in between.
TestSplit.testSplitRegionIsDeleted failed in patch build #1525 because of a NPE
(Scanner was null at end of count inside makeMultiRegionTable). Added test for
null scanner in v2 (though it shouldn't be null at close time) and removed this
test altogether since what its testing is being done when TTMR and
TestTableIndex run.
Test of shell failed in #1521 w/ a NPE but whats really odd is that it just did
exact same operation two times just previous w/o problem.
TestScanner2 fails in a distinctive mode almost immediately after startup of
any of its multiple tests: only thing that differs from 'normal' runs is
absence of the '... Unable to load native-hadoop library for your platform...
using builtin-java classes where applicable' log message (i.e. sequence files
haven't been attempted).
> [hbase] fixes for build up on hudson
> ------------------------------------
>
> Key: HADOOP-2558
> URL: https://issues.apache.org/jira/browse/HADOOP-2558
> Project: Hadoop
> Issue Type: Bug
> Components: contrib/hbase
> Reporter: stack
> Attachments: 2558-v2.patch, 2558.patch
>
>
> Fixes for hbase breakage up on hudson. There seem to be many reasons for the
> failings.
> One is that the .META. region of a sudden decides its 'no good' and it gets
> deployed elsewhere. Tests don't have the tolerance for this kinda churn. A
> previous commit adding in logging of why .META. is 'no good'. Hopefully that
> will help.
> Found also a case where TestTableMapReduce would fail because no sleep
> between retries when getting new scanners.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
