[
https://issues.apache.org/jira/browse/SPARK-4216?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14202420#comment-14202420
]
Josh Rosen edited comment on SPARK-4216 at 11/7/14 6:38 PM:
------------------------------------------------------------
A large part of the problem is that the Jenkins GHPRB plugin has a lot of
settings that are global rather than per-project. In this case, I think the
duplicate postings are being generated by the "fall back on posting comments in
case the GitHub commit status API call fails" setting. We can't use the status
API in Spark, but I guess the other AMP Lab projects used to use it and didn't
require this fallback. At some point, I think we switched the comment fallback
on because some other project needed it, leading to these duplicate updates.
As I've commented elsewhere, one solution would be to simply not use the GHPRB
plugin for Spark and instead use a parameterized build that's triggered
remotely (e.g. through spark-prs.appspot.com). I think that we could easily
build this layer on top of spark-prs; it's just a matter of finding the time to
do it (and to add the necessary features, like automatic detection of when new
commits have been pushed, listening to commands addressed to Jenkins, etc.) I
already have the triggering working manually (this runs
NewSparkPullRequestBuilder), so the only remaining piece is the automatic
triggering / ACLs.
was (Author: joshrosen):
A large part of the problem is that the Jenkins GHPRB plugin has a lot of
settings that are global rather than per-project. In this case, I think the
duplicate postings are being generated by the "fall back on posting comments in
case the GitHub commit status API call fails." We can't use the status API in
Spark, but I guess the other AMP Lab projects used to use it and didn't require
this fallback. At some point, I think we switched the comment fallback on
because some other project needed it, leading to these duplicate updates.
As I've commented elsewhere, one solution would be to simply not use the GHPRB
plugin for Spark and instead use a parameterized build that's triggered
remotely (e.g. through spark-prs.appspot.com). I think that we could easily
build this layer on top of spark-prs; it's just a matter of finding the time to
do it (and to add the necessary features, like automatic detection of when new
commits have been pushed, listening to commands addressed to Jenkins, etc.) I
already have the triggering working manually (this runs
NewSparkPullRequestBuilder), so the only remaining piece is the automatic
triggering / ACLs.
> Eliminate duplicate Jenkins GitHub posts from AMPLab
> ----------------------------------------------------
>
> Key: SPARK-4216
> URL: https://issues.apache.org/jira/browse/SPARK-4216
> Project: Spark
> Issue Type: Bug
> Components: Build, Project Infra
> Reporter: Nicholas Chammas
> Priority: Minor
>
> * [Real Jenkins |
> https://github.com/apache/spark/pull/2988#issuecomment-60873361]
> * [Imposter Jenkins |
> https://github.com/apache/spark/pull/2988#issuecomment-60873366]
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
