[
https://issues.apache.org/jira/browse/HADOOP-15593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16554680#comment-16554680
]
Xiao Chen commented on HADOOP-15593:
------------------------------------
It makes sense for the best-effort retries in general, so the renewal thread
doesn't abort prematurely due to intermittent failures.
But could you clarify a little more? If a tgt is destroyed, how can it be
renewed?
Looks to me KDC outage would result in relogin failure and possibly getTGT()
being null without an exception, after which the current code just does a null
check on tgt and return without retries. IMO we should be consistent with it
and just return.
I don't feel strongly that having the last try as patch 4 is a big problem, but
it's not clear under which scenario this could possibly succeed.
> UserGroupInformation TGT renewer throws NPE
> -------------------------------------------
>
> Key: HADOOP-15593
> URL: https://issues.apache.org/jira/browse/HADOOP-15593
> Project: Hadoop Common
> Issue Type: Bug
> Components: security
> Affects Versions: 3.0.0
> Reporter: Wei-Chiu Chuang
> Assignee: Gabor Bota
> Priority: Blocker
> Attachments: HADOOP-15593.001.patch, HADOOP-15593.002.patch,
> HADOOP-15593.003.patch, HADOOP-15593.004.patch
>
>
> Found the following NPE thrown in UGI tgt renewer. The NPE was thrown within
> an exception handler so the original exception was hidden, though it's likely
> caused by expired tgt.
> {noformat}
> 18/07/02 10:30:57 ERROR util.SparkUncaughtExceptionHandler: Uncaught
> exception in thread Thread[TGT Renewer for f...@example.com,5,main]
> java.lang.NullPointerException
> at
> javax.security.auth.kerberos.KerberosTicket.getEndTime(KerberosTicket.java:482)
> at
> org.apache.hadoop.security.UserGroupInformation$1.run(UserGroupInformation.java:894)
> at java.lang.Thread.run(Thread.java:748){noformat}
> Suspect it's related to [https://bugs.openjdk.java.net/browse/JDK-8154889].
> The relevant code was added in HADOOP-13590. File this jira to handle the
> exception better.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org
