Github user steveloughran commented on the pull request:
https://github.com/apache/spark/pull/11033#issuecomment-184202371
I know that you can skip credential pickup for hbase & hive, this patch
also skips trying to collect any HDFS tokens for NNs, though of course that
could be handled by setting "spark.yarn.access.namenodes"
What this does do is make the fact that env-supplied credentials are coming
in more explicit, and dumping out their values.
I do now agree, by turning the right config options you could get away
without being aware of the fact that the client is running in Oozie. In which
case I propose a leaner patch which
1. Still does a dump of those tokens passed in and when they expire. This
helps diagnose that problem where not enough tokens have been collected, and
the problem only surfacing later.
1. Document how to use spark under oozie
what do you think?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
