[
https://issues.apache.org/jira/browse/HADOOP-9747?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16367627#comment-16367627
]
Daryn Sharp commented on HADOOP-9747:
-------------------------------------
A synchronize somehow crept back into loginUserFromKeytab. Removed it.
I put back the previous invalid options for IBM_JAVA because it doesn't hurt
anything and is arguably the least risk.
Added missing @Test annotations.
Updated javadoc for getLoginUser to be a bit more informative about what it
does (actually has always done).
Don't understand the concern over a race between getLoginUser and other login
methods (which does exist and is handled) is helped by synchronization. Sure
synchronize guards against 2 racing getLoginUser calls but not concurrent calls
to the other login methods. Whereas the CAS logic ensures that if any login
call beats us to setting the default login user, that winner will be returned
instead. I unwound the code a bit (logic is identical) with better comments to
better convey what's happening.
> Reduce unnecessary UGI synchronization
> --------------------------------------
>
> Key: HADOOP-9747
> URL: https://issues.apache.org/jira/browse/HADOOP-9747
> Project: Hadoop Common
> Issue Type: Bug
> Components: security
> Affects Versions: 0.23.0, 2.0.0-alpha, 3.0.0-alpha1
> Reporter: Daryn Sharp
> Assignee: Daryn Sharp
> Priority: Critical
> Attachments: HADOOP-9747-trunk-03.patch, HADOOP-9747-trunk-04.patch,
> HADOOP-9747-trunk.01.patch, HADOOP-9747-trunk.02.patch,
> HADOOP-9747.2.branch-2.patch, HADOOP-9747.2.trunk.patch,
> HADOOP-9747.branch-2.patch, HADOOP-9747.trunk.patch
>
>
> Jstacks of heavily loaded NNs show up to dozens of threads blocking in the
> UGI.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
