zookeeper-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Han <h...@apache.org>
Subject Re: About Jenkins pre-commit job
Date Fri, 21 Sep 2018 17:11:16 GMT
Thanks for the work, Enrico! Looks like it's working?

>> if the description of the PR contains 'skip ci' validation won't be run
from jenkins

Do we need this option?

>> doc/wiki to update and if there is a common place for storing CI Jobs
configurations

jenkins job config is stored in the job itself. We don't version control it
separately.
With this update the "Jenkins" section is out of date:
https://cwiki.apache.org/confluence/display/ZOOKEEPER/HowToContribute
Would you please update the page to reflect latest state of the jenkins -
github integration?

>> It seems that the first line with 'Jenkins' never goes away.

Seems ok to me to keep two lines. The second line does not cover things
other than test (like find bug checks, release audit checks, etc).

On Fri, Sep 21, 2018 at 5:51 AM, Enrico Olivelli <eolivelli@gmail.com>
wrote:

> Too early...
> It seems that the first line with 'Jenkins' never goes away.
>
> Patrick, do you have some hint ?
> Otherwise I can ask to infra or builds@apache.org
>
> Enrico
>
> Il giorno ven 21 set 2018 alle ore 14:45 Enrico Olivelli
> <eolivelli@gmail.com> ha scritto:
> >
> > Okay
> > on github you see two lines, but there is only one job on Jenkins
> >
> > see
> > https://github.com/apache/zookeeper/pull/638
> >
> > @Michael Han please double check
> >
> > If all it is okay I will send a new clean summary in another email thread
> > The short version is:
> > - an user sends a PR (against master, branch-3.5 and branch-3.4)> CI
> kicks in an validates, on GitHub you see the status and the results
> > - an "Apache Committer" (from any project I guess, as I am not a ZK
> committer but it works, it is important that he is registered in Apache
> Gihub 'organization') can add a comment with only this phrase 'retest this
> please' and the PR will be re-validated from CI
> > - if the description of the PR contains 'skip ci' validation won't be
> run from jenkins
> >
> > configuration is here:
> > https://builds.apache.org/job/PreCommit-ZOOKEEPER-github-pr-build/
> >
> >
> > Please tell me if there is any doc/wiki to update and if there is a
> common place for storing CI Jobs configurations
> >
> >
> > Enrico
> >
> > Il giorno ven 21 set 2018 alle ore 11:56 Enrico Olivelli <
> eolivelli@gmail.com> ha scritto:
> >>
> >>
> >>
> >> Il giorno ven 21 set 2018 alle ore 11:27 Enrico Olivelli <
> eolivelli@gmail.com> ha scritto:
> >>>
> >>> I have set up the 'retest this please' trigger.
> >>> But now we have two jobs running for each pull request.
> >>>
> >>> I can't find the flag which makes the validation named 'Jenkins' start
> >>> https://builds.apache.org/job/PreCommit-ZOOKEEPER-github-pr-
> build/configure
> >>>
> >>> the name of the new one is 'default', and it maps to the same Job
> >>
> >>
> >> Actually the result for the new 'default' validation looks better
> >> see it here
> >> https://github.com/apache/zookeeper/pull/637
> >>
> >> The old one just told 'Looks like there's a problem with this pull
> request'
> >> the new one reports ' FAILURE 1731 tests run, 1 skipped, 2 failed'
> >>
> >> Enrico
> >>
> >>
> >>>
> >>> is there any other trigger ?
> >>> I will continue to search
> >>>
> >>> Enrico
> >>>
> >>> Il giorno ven 21 set 2018 alle ore 09:33 Enrico Olivelli <
> eolivelli@gmail.com> ha scritto:
> >>>>
> >>>> I apologize if there will be some spam on the list
> >>>>
> >>>> When you enable this trigger asfgit starts to add comments like this:
> >>>> Can one of the admins verify this patch?
> >>>>
> >>>> I am looking for a solution. We had the same on BookKeeper jobs
> >>>>
> >>>> Enrico
> >>>>
> >>>> Il giorno ven 21 set 2018 alle ore 09:31 Enrico Olivelli <
> eolivelli@gmail.com> ha scritto:
> >>>>>
> >>>>>
> >>>>>
> >>>>> Il giorno ven 21 set 2018 alle ore 03:44 Michael Han <
> hanm@apache.org> ha scritto:
> >>>>>>
> >>>>>> Thanks Enrico for volunteering, it'll be cool if we can fix
this.
> >>>>>
> >>>>>
> >>>>> Ok, working now on
> >>>>> https://builds.apache.org/job/PreCommit-ZOOKEEPER-github-pr-build/
> >>>>>
> >>>>> Stay tuned
> >>>>> Enrico
> >>>>>
> >>>>>>
> >>>>>>
> >>>>>> On Wed, Sep 19, 2018 at 10:42 PM, Enrico Olivelli <
> eolivelli@gmail.com> wrote:
> >>>>>>>
> >>>>>>> Il gio 20 set 2018, 01:17 Michael Han <hanm@apache.org>
ha
> scritto:
> >>>>>>>
> >>>>>>> > We've seen lots of confusions recently around Jenkins
pre-commit
> check
> >>>>>>> > status on github pull request as well as questions
like how to
> trigger
> >>>>>>> > Jenkins job to get a green build. I added a section
"Jenkins
> Pre-commit
> >>>>>>> > Check" to our wiki page about this matter:
> >>>>>>> >
> >>>>>>> > https://cwiki.apache.org/confluence/display/ZOOKEEPER/
> HowToContribute
> >>>>>>> >
> >>>>>>> > I suspect the permission issue is about the fact that
apache
> zookeeper
> >>>>>>> > github is read only for most of us including committers.
Only
> asfbot has
> >>>>>>> > write access. When a Jenkins job is triggered through
"Rebuild"
> button the
> >>>>>>> > credential it's using is not asfbot thus the final
job result
> will fail to
> >>>>>>> > write back to github. I don't know how to solve this,
but
> hopefully we have
> >>>>>>> > workarounds (which though require PR creator to be
more diligent
> on
> >>>>>>> > obtaining a green build).
> >>>>>>> >
> >>>>>>>
> >>>>>>> In Bookkeeper we have enabled the magic 'retest this please'
and
> it is very
> >>>>>>> useful.
> >>>>>>>
> >>>>>>> I can take a look in order to back port to ZK, I have access
to
> ASF Jenkins
> >>>>>>> so I can check current configuration for ZK.
> >>>>>>>
> >>>>>>> In Bookkeeper we also have the configuration of Jenkins
jobs on
> git, using
> >>>>>>> Jenkins DSL language. I can also take a look and see how
to port
> to ZK
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> Enrico
> >>>>>>>
> >>>>>>>
> >>>>>>> > Michael.
> >>>>>>> >
> >>>>>>> --
> >>>>>>>
> >>>>>>>
> >>>>>>> -- Enrico Olivelli
> >>>>>>
> >>>>>>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message