I use a local worker which its only job is to coordinate the triggers and the
triggered builds and wait for all of that to finish.
Nothing is actually built on the local builders, so, having them wait is not an
issue and it never occupies an actual worker.
Makes any sense for your use case?
On February 14, 2018 12:45:43 AM GMT+00:00, Will Rosecrans
>I have a buildbot job that builds some software, makes a binary, and
>triggers a separate job to run some tests that only need the binary
>it continues running some tests that require the full source. I am
>impatient, so I want the triggered job to run asynchronously from the
>of the build. (Both take ages). But I would still like the overall
>fail if the triggered builder fails.
>So is there some way for me to set a property in the builder saying
>something like the job ID # of the triggered build, that I could use in
>step at the end of the triggering build to poll for completion status
>the triggered build?
>Having to refactor things so that the controlling build's tests could
>as another triggered build would mean having to upload the whole
>half-built source tree to the master, and then schlep it to another
>machine to run those tests -- and it would leave the machine that
>runs the tests just sitting idle occupying a build machine waiting for
>triggered jobs to finish. So I'd like to avoid going that route in
>particular scenario, even if the master-controller that just triggers
>usually recommended for this sort of thing generally. (I already have
>master controller that triggers builds on all the OS's)
>Any clever ideas?
Pedro Algarvio @ Phone
users mailing list