[ 
https://issues.apache.org/jira/browse/HDFS-11131?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15954451#comment-15954451
 ] 

Arpit Agarwal edited comment on HDFS-11131 at 4/4/17 1:38 AM:
--------------------------------------------------------------

{{testContextIsPassed}} hasn't failed lately however {{testCancellation}} 
continues to fail in test runs occasionally as Kihwal pointed out above. I ran 
~500 iterations locally and got a couple of failures. I think 
{{ScheduledThreadPoolExecutor#shutdownNow}} occasionally fails to interrupt the 
executing worker thread as I'd expect although I haven't been able to figure 
out the issue from looking at its implementation.

Uploading a patch that:
# Replaces testCancellation with {{testExceptionIsPropagated}}. It 
deterministically tests the useful part of testCancellation i.e. exceptions are 
correctly propagated to Future#get().
# Removes the part of testConcurrentChecks already covered by another test case.
# Simplifies {{ThrottledAsyncChecker#shutdown}}, since we don't need to wait 
for graceful completion of disk checks during DataNode shutdown.

Ran a few hundred more iterations locally and they passed.


was (Author: arpitagarwal):
{{testContextIsPassed}} hasn't failed lately however {{testCancellation}} 
continues to fail in test runs occasionally as Kihwal pointed out above. I ran 
~500 iterations locally and got a couple of failures. I think 
{{ScheduledExecutorService#shutdownNow}} occasionally fails to interrupt the 
executing worker thread as I'd expect although I haven't been able to figure 
out the issue from looking at its implementation.

Uploading a patch that:
# Replaces testCancellation with {{testExceptionIsPropagated}}. It 
deterministically tests the useful part of testCancellation i.e. exceptions are 
correctly propagated to Future#get().
# Removes the part of testConcurrentChecks already covered by another test case.
# Simplifies {{ThrottledAsyncChecker#shutdown}}, since we don't need to wait 
for graceful completion of disk checks during DataNode shutdown.

Ran a few hundred more iterations locally and they passed.

> TestThrottledAsyncChecker#testCancellation is flaky
> ---------------------------------------------------
>
>                 Key: HDFS-11131
>                 URL: https://issues.apache.org/jira/browse/HDFS-11131
>             Project: Hadoop HDFS
>          Issue Type: Bug
>    Affects Versions: 3.0.0-alpha2
>            Reporter: Arpit Agarwal
>            Assignee: Arpit Agarwal
>         Attachments: HDFS-11131.01.patch
>
>
> This test failed in a few precommit runs. e.g.
> https://builds.apache.org/job/PreCommit-HDFS-Build/18952/testReport/org.apache.hadoop.hdfs.server.datanode.checker/TestThrottledAsyncChecker/testCancellation/



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to