hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Siddharth Seth (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-357) App submission should not be synchronized
Date Tue, 05 Feb 2013 16:19:17 GMT

    [ https://issues.apache.org/jira/browse/YARN-357?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13571423#comment-13571423

Siddharth Seth commented on YARN-357:

bq. Thanks for the review! I struggled quite a bit with writing these tests. I am indeed expecting
the barrier to cause the test to timeout. How do you suggest I add an explicit error? Log
the assert since there's not much else I can (easily) do? 
A flag to track the error in the thread should work ?

I couldn't get chaining to work with void return methods. The mock hangs on a poll for repeated
calls to a method with a doAnswer if there aren't more doAnswers registered with it, but since
I couldn't get the chaining to work up-front, I added the chain after the method is hit the
first time.
bq. Chaining seemed to work ok, as long as the previous invocation had completed. Even without
chaining, the mock does not seem to handle multiple parallel invocations. Struggled a bit
with this as well - which is what makes me a little wary of using mockito for this case (http://code.google.com/p/mockito/wiki/FAQ).

bq. How would a custom event handler work better in this case? I think it'd just mimic the
current behavior? Could we maybe have the separate jira, for removing other syncs, improve
the tests in general? Or could you help me out by updating the tests since I'm unclear what
to do?
It'd behave the same, except would not use mockito for the EventHandler mock. Sure, I'll try
updating the unit test.
> App submission should not be synchronized
> -----------------------------------------
>                 Key: YARN-357
>                 URL: https://issues.apache.org/jira/browse/YARN-357
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: resourcemanager
>    Affects Versions: 0.23.3, 3.0.0, 2.0.0-alpha
>            Reporter: Daryn Sharp
>            Assignee: Daryn Sharp
>         Attachments: YARN-357.patch, YARN-357.patch
> MAPREDUCE-2953 fixed a race condition with querying of app status by making {{RMClientService#submitApplication}}
synchronously invoke {{RMAppManager#submitApplication}}. However, the {{synchronized}} keyword
was also added to {{RMAppManager#submitApplication}} with the comment:
> bq. I made the submitApplication synchronized to keep it consistent with the other routines
in RMAppManager although I do not believe it needs it since the rmapp datastructure is already
a concurrentMap and I don't see anything else that would be an issue.
> It's been observed that app submission latency is being unnecessarily impacted.

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

View raw message