[
https://issues.apache.org/jira/browse/BIGTOP-835?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13567442#comment-13567442
]
Johnny Zhang commented on BIGTOP-835:
-------------------------------------
[~hshreedharan], thanks for explaining this, your use case strongly justify the
needs. It is actually like what I met in Mahout testing: the data mining
algorithm takes long long time to generate the data. However, my test itself
just need to make sure Mahout finish the job, it won't check the results by
comparing with any golden standard.
The only concern for me is what if test trigger a MR job and it hangs? It will
hold all the other MR jobs in the queue forever until you manually kill it.
Since no error message, or other trace, it is hard to debug the test
failure/hanging. This situation will This is what exactly I met in Mahout
testing. If this is the case, how about adding a block of self cleaning in the
end? For example, check if the job hangs or doesn't fail gracefully, if so,
kill the MR job.
> The shell exec method must have variants which have timeout and can run in
> background
> -------------------------------------------------------------------------------------
>
> Key: BIGTOP-835
> URL: https://issues.apache.org/jira/browse/BIGTOP-835
> Project: Bigtop
> Issue Type: Bug
> Reporter: Hari Shreedharan
> Attachments: BIGTOP-835-1.patch, BIGTOP-835-2.patch,
> BIGTOP-835-2.patch, BIGTOP-835-3.patch
>
>
--
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
