[ https://issues.apache.org/jira/browse/LIVY-392 ]
Larry McCay deleted comment on LIVY-392:
----------------------------------
was (Author: lmccay):
Due to the pending 2.0.0 release this JIRA has been pushed out to 2.1.0 as part
of a bulk update. If there is a specific reason to pull this back into the
2.0.0 release and you intend to provide a PR in the next few days please
provide justification and reset the Fix Version to 2.0.0.
> Kerberos exception: "Failed to find any Kerberos tgt" when creating a session
> -----------------------------------------------------------------------------
>
> Key: LIVY-392
> URL: https://issues.apache.org/jira/browse/LIVY-392
> Project: Livy
> Issue Type: Bug
> Components: RSC, Server
> Affects Versions: 0.3
> Environment: Livy deployed within a Kerberized Grid.
> Reporter: Pranay Hasan Yerra
> Priority: Major
> Fix For: 0.9.0
>
> Attachments: opensource-stacktrace.out, stacktrace.txt
>
>
> Livy isn't able to authenticate to RM at times between two kinit calls which
> refreshes the livy service principal ticket cache. The specific stacktrace is
> attached as stacktrace.txt.
> I also attached whole application stacktrace as opensource-stacktrace.out,
> where kinit happens at 17/08/04 09:59:09 which updates the ticket cache in
> /tmp. It happened that all the session creation requests give the TGT not
> found exception until the next kinit occurred which happened at 17/08/04
> 10:59:09
> Is there a reason why Livy isn't able to use the TGT during this period?
> Does, using UGI methods avoid this exception? Does it happen when too many
> context not found exceptions happen?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
