[ 
https://issues.apache.org/jira/browse/STORM-2876?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16309880#comment-16309880
 ] 

Robert Joseph Evans commented on STORM-2876:
--------------------------------------------

I found the root cause of the issue.

https://issues.apache.org/jira/browse/HDFS-12984

I am not totally sure why it is not happening more frequently.  I guess it must 
have something to do with how our build boxes are configured and the data nodes 
end up taking a small amount more ram then on other systems.  Because I don't 
think this is going to be fixed any time soon I will work around it by 
launching all of the tests in separate JVMs.  I'll see what the hit is, and 
might just do it only for the HDFS tests.

> Some storm-hdfs tests fail with out of memory periodically
> ----------------------------------------------------------
>
>                 Key: STORM-2876
>                 URL: https://issues.apache.org/jira/browse/STORM-2876
>             Project: Apache Storm
>          Issue Type: Bug
>          Components: storm-hdfs
>    Affects Versions: 2.0.0
>            Reporter: Robert Joseph Evans
>            Assignee: Robert Joseph Evans
>
> In our 2.x automated testing we have noticed that every so often we will see 
> TestFileLock fail with out of memory errors.
> {code}
> java.lang.OutOfMemoryError: Java heap space
> ...
> {code}
> Which then appears to trigger other failures.  Not sure if there is a memory 
> leak involved or if the tests really need 1.5GB of memory periodically.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to