[
https://issues.apache.org/jira/browse/HBASE-19742?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16321911#comment-16321911
]
Chia-Ping Tsai edited comment on HBASE-19742 at 1/11/18 9:22 AM:
-----------------------------------------------------------------
Another issue is why the test fails? MSLAB should handle the OOM so the test
should pass even if the OOM happens.
I have attached a patch to fix the bug. We should add the chunk to map after
the chunk succeed to allocate the BB.
BTW, the test eats almost 2G on my local. Perhaps we should reduce the number
of MSLAB thread to prevent the OOM when we execute tests in parallel.
was (Author: chia7712):
Another issue is why the test fails? MSLAB does handle the OOM so the test
should pass even if the OOM happens.
I have attached a patch to fix the bug. We should add the chunk to map after
the chunk succeed to allocate the BB.
BTW, the test eats almost 2G on my local. Perhaps we should reduce the number
of MSLAB thread to prevent the OOM when we execute tests in parallel.
> TestMemstoreLABWithoutPool fails on some machines (including Apache Jenkins)
> ----------------------------------------------------------------------------
>
> Key: HBASE-19742
> URL: https://issues.apache.org/jira/browse/HBASE-19742
> Project: HBase
> Issue Type: Bug
> Components: test
> Reporter: stack
> Attachments: HBASE-19742.test.patch
>
>
> On Andrew Purtell's linux he sees this when he runs
> TestMemstoreLABWithoutPool (See thread "[VOTE] The second hbase-2.0.0-beta-1
> Release Candidate is available for download"):
> {code}
> TestMemstoreLABWithoutPool.org.apache.hadoop.hbase.regionserver.TestMemstoreLABWithoutPool
> can fail with OOME if not run in isolation.
> If run in isolation, I get
> org.apache.hadoop.hbase.regionserver.TestMemstoreLABWithoutPool.testLABChunkQueueWithMultipleMSLABs(org.apache.hadoop.hbase.regionserver.TestMemstoreLABWithoutPool)
> [ERROR] Run 1:
> TestMemstoreLABWithoutPool.testLABChunkQueueWithMultipleMSLABs:143 All the
> chunks must have been cleared
> [ERROR] Run 2:
> TestMemstoreLABWithoutPool.testLABChunkQueueWithMultipleMSLABs:143 All the
> chunks must have been cleared
> [ERROR] Run 3:
> TestMemstoreLABWithoutPool.testLABChunkQueueWithMultipleMSLABs:143 All the
> chunks must have been cleared
> [ERROR] Run 4:
> TestMemstoreLABWithoutPool.testLABChunkQueueWithMultipleMSLABs:143 All the
> chunks must have been cleared
> {code}
> It runs fine for me on my linux and mac. [~ram_krish] / [~anastas] ? You see
> any of this?
> Going to disable this test in meantime.... in subissue.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
