[
https://issues.apache.org/jira/browse/HADOOP-9747?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16137504#comment-16137504
]
Daryn Sharp commented on HADOOP-9747:
-------------------------------------
Had an incident today with a RM going dead-in-the-water that this patch would
have allowed to self-heal. A re-login was triggered, and it failed. Now the
UGI was left with no private credentials. New calls to getCurrentUser() found
neither a keytab nor ticket instance, so any call to relogin was a no-op. This
patch "remembers" the ugi was logged in from a keytab so re-logins would have
been attempted.
Regarding earlier concerns, we've been running with the locking on subject's
private credentials since early Feb 2017 after experiencing DN lockups. We
just aren't running the code to remember the login conf.
> 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-9747.2.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
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
