[
https://issues.apache.org/jira/browse/BIGTOP-1259?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13946988#comment-13946988
]
Mark Grover commented on BIGTOP-1259:
-------------------------------------
+1. FWIW, we renamed the spark package to be called spark-core so the puppet
manifests need to be updated to reflect. If you could please create a JIRA for
that and target it for 0.8, that'd be much appreciated!
> spark-worker service should be started after spark-master if ran on the same
> node
> ---------------------------------------------------------------------------------
>
> Key: BIGTOP-1259
> URL: https://issues.apache.org/jira/browse/BIGTOP-1259
> Project: Bigtop
> Issue Type: Bug
> Components: Deployment
> Affects Versions: 0.7.0
> Reporter: Konstantin Boudnik
> Assignee: Konstantin Boudnik
> Fix For: 0.8.0
>
> Attachments: BIGTOP-1259.patch
>
>
> When master and worker components are deployed to the same node it is
> important to guarantee the sequence of startup events: e.g worker will fail
> if master isn't present. Currently, this is a limitation of the Spark
> implementation where a worker doesn't wait for master to start up and fails
> immediately.
> Also, the master service only has to be started on the master node.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
