[users at bb.net] force scheduler skipping builds

Elliot Saba staticfloat at gmail.com
Wed Mar 8 23:24:07 UTC 2017


Bump.  Does anyone have any idea why this would be happening?  If you need
more information to diagnose this problem, I would be happy to provide it.
-E

On Sun, Feb 26, 2017 at 3:40 PM, Elliot Saba <staticfloat at gmail.com> wrote:

> Hello again,
>
> I have two schedulers defined to package my binaries.  The first, is an
> AnyBranchScheduler
> <https://github.com/staticfloat/julia-buildbot/blob/8c7859d3136c10f3236e4e11cbd408d773123029/master/package.py#L319-L328> that
> has a treeStableTimer defined.  The second, defined right below, is a
> ForceScheduler
> <https://github.com/staticfloat/julia-buildbot/blob/8c7859d3136c10f3236e4e11cbd408d773123029/master/package.py#L339-L356>
> .
>
> I am working on creating some tooling that allows us to force-schedule
> builds easily from the GitHub interface, and I discovered that if I
> force-schedule two builds, the first flips to results code 3, e.g. it is
> SKIPPED
> <http://docs.buildbot.net/latest/developer/results.html#buildbot.process.results.SKIPPED>
> .
>
> This is the exact behavior that I would expect if the ForceScheduler had a
> treeStableTimer applied to it, however, it does not, it's the
> AnyBranchScheduler that has one.
>
> It seems to me like the AnyBranchScheduler's `treeStableTimer` is messing
> with the builds from my ForceScheduler.  I may be conflating two entirely
> unrelated things, but if so, I'd appreciate any help in tracking this
> down.  I definitely want a tree stable timer on the AnyBranchScheduler,
> because otherwise merge commits can easily rack up days worth of CI time,
> as each build takes about an hour to complete, however I can't lose force
> scheduled builds willy-nilly.
>
> Thanks,
> -E
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.buildbot.net/pipermail/users/attachments/20170308/3996cfc0/attachment.html>


More information about the users mailing list