[OpenAFS-devel] Re: slave suggestions for buildslaves?

Jason Edgecombe jason@rampaginggeek.com
Wed, 26 Jun 2013 10:57:30 -0400


On 06/26/2013 10:19 AM, Andrew Deason wrote:
> On Sun, 23 Jun 2013 20:40:17 -0400
> Jeffrey Hutzelman <jhutz@cmu.edu> wrote:
>
>> I guess I don't see an advantage to this.  The branches which only get
>> pullups once in a while get all the cycles, while the branch where the
>> real work happens gets to be starved?
> They could be for master, too; I'd like them to prioritize stable branch
> changes over master changes, if that's possible to do. The advantage for
> 1.6 is that it would speed up rebasing/submitting a lot of 1.6 changes,
> which can easily happen; I think speed there is more important, since
> delays can easily delay the release process. I don't really see what
> master submission delays impede.
>
> I don't think the concern about differing buildbot results would be a
> huge problem, though; we should notice it pretty quickly, and builds
> breaking for "external" reasons still happen sometimes.
>

FYI,

Our currently deployed buildbot does support user-defined algorithm for 
builder scheduling and choosing the next slave:
http://docs.buildbot.net/0.7.12/#Prioritizing-Builders

See "nextslave" option:
http://docs.buildbot.net/0.7.12/#Defining-Builders

That said, I would have to experiment to do this.

Jason