[
https://issues.apache.org/jira/browse/SPARK-11653?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15000390#comment-15000390
]
Adrian Bridgett commented on SPARK-11653:
-----------------------------------------
Thanks Sean, I took a slightly different approach - basically just using an
environment variable, I didn't add any argument handling support as it's a bit
of an advanced case (so treating it just as SPARK_NICENESS or SPARK_LOG_DIR).
The discussion on SPARK-5964 seemed to move into a cleanup of the code (which
is fine, but perhaps should be a separate task). I'm happy to do that if
desired to accept multiple longargs, print errors to stderr etc.
> Would be very useful if spark-daemon.sh supported foreground operations
> -----------------------------------------------------------------------
>
> Key: SPARK-11653
> URL: https://issues.apache.org/jira/browse/SPARK-11653
> Project: Spark
> Issue Type: Improvement
> Components: Deploy
> Affects Versions: 1.5.1
> Reporter: Adrian Bridgett
> Priority: Minor
>
> See also SPARK-5964. I've raised a new ticket for a few reasons:
> - alternative more minimal patch (I've not submitted a PR as I thought any
> discussion would be better here):
> https://github.com/abridgett/spark/tree/feature/add_spark_daemon_foreground
> - couldn't re-open the original ticket and didn't know if a comment would
> suffice
> There are quite a few use cases for this, e.g. when running under Mesos the
> shuffle service needs to be running and trying to do this under Marathon is a
> bit painful without this feature.
> I see that there was a plan for these scripts to be rewritten in Python,
> however until then this would be very useful (and can feed into the rewrite
> as well).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
