[Buildbot-devel] Waiting for two builders to finish beforetriggering a third
JKelly at vectorworks.net
Fri Jul 8 16:35:48 UTC 2011
Ah, thanks for letting me know! I find the dependent schedulers slightly easier to configure, a little cleaner (as you don't need to add a step on the builder you are dependent on) but I suppose the triggerables are much more versatile and the functionality is largely a duplicate.
From: Amber Yust [mailto:ayust at yelp.com]
Sent: Thu 7/7/2011 4:14 PM
To: Jim Kelly
Cc: Alec Douglas; buildbot-devel at lists.sourceforge.net; Alex Escott
Subject: Re: [Buildbot-devel] Waiting for two builders to finish beforetriggering a third
Triggers+Triggerable schedulers are preferred over Dependent schedulers
(which to some extent should be considered deprecated).
On Thu, Jul 7, 2011 at 1:06 PM, Jim Kelly <JKelly at vectorworks.net> wrote:
> Perhaps have a specific scheduler that kicks off X and Y, and a dependent
> scheduler for T, whose upstream was the scheduler is the one you have
> feeding X and Y? I'm doing something similar in a series, where as i have
> builds occur on builders on two different platforms, and then if that
> succeeds, I do a longer, more involved step on another builder, and then if
> that succeeds, I package up all the results.
> -----Original Message-----
> From: Alec Douglas [mailto:AD at malaspina-labs.com <AD at malaspina-labs.com>]
> Sent: Thu 7/7/2011 3:00 PM
> To: buildbot-devel at lists.sourceforge.net
> Cc: Alex Escott
> Subject: [Buildbot-devel] Waiting for two builders to finish
> beforetriggering a third
> My problem is as follows: I have two builders (let's call them X and Y)
> that are triggered simultaneously when code is committed. X and Y build two
> different versions of the committed code. I also have a third builder (call
> it T) that runs tests on both versions which is currently triggered on the
> completion of X. The problem is, it is difficult to guarantee that Y has
> finished before T begins, and if the tests are run before Y is done then the
> tests will fail unnecessarily (by that I mean: failing based on not being
> built yet, not based on a problem in the code).
> What I would like to do is only have T run after both X and Y have
> I would like to refrain from simply moving the buildsteps from Y onto X, as
> the builds take a bit of time and running them sequentially instead of
> simultaneously would slow things down considerably. I would also like to
> avoid splitting the tests into two builders (T1 running on X and T2 running
> on Y), as that would cause a great deal of bloat.
> Any suggestions?
> Alec Douglas
> All of the data generated in your IT infrastructure is seriously valuable.
> Why? It contains a definitive record of application performance, security
> threats, fraudulent activity, and more. Splunk takes this data and makes
> sense of it. IT sense. And common sense.
> Buildbot-devel mailing list
> Buildbot-devel at lists.sourceforge.net
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the devel