[hibernate-dev] ORM CI jobs - erroneous github triggers

[Sanne Grinovero]

Unfortunately we have to reopen this thread. The issue is affecting us
again (since some weeks I think) and in larger scale, affecting all
builds. I guess it means the previous workaround is no longer
effective after some of the recent Jenkins / plugins updates.

On IRC we where wondering if we could fix this issue ourselves, or if
it's even possible with the metadata that GitHub provides. I'm not
sure of that but since we have the impression this previously worked
fine, maybe the metadata is available?

I won't have time for this this week but if someone else wishes to
look into it I can give some pointers, let me know.

Thanks,
Sanne


On 4 January 2018 at 12:09, Sanne Grinovero <sanne at hibernate.org> wrote:
> Also these jobs were configured to build automatically every 5 hours:
>  - hibernate-orm-4.2-h2
>  - hibernate-orm-4.3-h2
>
> I removed the schedule, they will be built when (and only when)
> anything is committed to their respective branches.
>
>
> On 3 January 2018 at 19:15, Steve Ebersole <steve at hibernate.org> wrote:
>> Nice!  Glad you found something.  Thanks for making the changes.
>>
>>
>>
>> On Wed, Jan 3, 2018 at 12:16 PM Sanne Grinovero <sanne at hibernate.org> wrote:
>>>
>>> I've made the change on:
>>>    - hibernate-orm-5.0-h2
>>>    - hibernate-orm-5.1-h2
>>>    - hibernate-orm-master-h2-main
>>>
>>> Let's see if it helps, then we can figure out a way to check all jobs
>>> are using this workaround.
>>>
>>>
>>> On 3 January 2018 at 18:12, Sanne Grinovero <sanne at hibernate.org> wrote:
>>> > Hi Steve,
>>> >
>>> > this rings a bell, we had this bug in the past and apparently it's
>>> > regressed again :(
>>> >
>>> > The latest Jenkins bug seems to be:
>>> >  - https://issues.jenkins-ci.org/browse/JENKINS-42161
>>> >
>>> > I'll try the suggested workarount, aka to enable SCM poll without any
>>> > frequency.
>>> >
>>> > Thanks,
>>> > Sanne
>>> >
>>> >
>>> > On 3 January 2018 at 17:35, Steve Ebersole <steve at hibernate.org> wrote:
>>> >> So I just pushed to the ORM master branch, which has caused the
>>> >> following
>>> >> jobs to be queued up:
>>> >>
>>> >>
>>> >>    - hibernate-orm-5.0-h2
>>> >>    - hibernate-orm-5.1-h2
>>> >>    - hibernate-orm-master-h2-main
>>> >>
>>> >> Only one of those jobs is configured to "watch" master.  So why do
>>> >> these
>>> >> other jobs keep getting triggered?
>>> >>
>>> >> I see the same exact thing on my personal fork as well.  At the same
>>> >> time I
>>> >> pushed to my fork's 5.3 branch, which triggered the 6.0 job to be
>>> >> queued.
>>> >>
>>> >>
>>> >>
>>> >> On Tue, Jan 2, 2018 at 1:54 PM Steve Ebersole <steve at hibernate.org>
>>> >> wrote:
>>> >>
>>> >>> The legacy ORM jobs (5.1-based ones at least) are getting triggered
>>> >>> when
>>> >>> they should not be.  Generally they all show they the run is triggered
>>> >>> by a
>>> >>> "SCM change", but it does not show any changes.  The underlying
>>> >>> problem
>>> >>> (although I am at a loss as to why) is that there has indeed been SCM
>>> >>> changes pushed to Github, but against completely different branches.
>>> >>> As
>>> >>> far as I can tell these job's Github setting are correct.  Any ideas
>>> >>> what
>>> >>> is going on?
>>> >>>
>>> >>> This would not be such a big deal if the CI environment did not
>>> >>> throttle
>>> >>> all waiting jobs down to one active job.  So the jobs I am actually
>>> >>> interested in are forced to wait (sometimes over an hour) for these
>>> >>> jobs
>>> >>> that should not even be running.
>>> >>>
>>> >>>
>>> >>>
>>> >> _______________________________________________
>>> >> hibernate-dev mailing list
>>> >> hibernate-dev at lists.jboss.org
>>> >> https://lists.jboss.org/mailman/listinfo/hibernate-dev