On Tue, Jan 5, 2021 at 3:21 PM Jorge Cardoso Leitão
<jorgecarlei...@gmail.com> wrote:
>
> Hi Krisztian,
>
> Could you describe which flows you are referring to?
>
> Rust builds are taking 4-6m on non-(windows/mac) and 8-15m on windows/mac,
> with almost immediate feedback. E.g. I force-pushed recently
> https://github.com/apache/arrow/pull/9089 and the first complete test came
> 5m later.
On the actions page I see builds with 30m queue time:
https://github.com/apache/arrow/actions
>
> One hypothesis is that, on personal repos, many builds are being triggered
> when any change happens. On my repo, R
> <https://github.com/jorgecarleitao/arrow/actions/runs/463436428>, Python
> <https://github.com/jorgecarleitao/arrow/actions/runs/463436416>, C++
> <https://github.com/jorgecarleitao/arrow/actions/runs/463436422>, Java JNI
> <https://github.com/jorgecarleitao/arrow/actions/runs/463436419>, C GLib
> and Ruby <https://github.com/jorgecarleitao/arrow/actions/runs/463436420>
> jobs ran when a Rust-only change was force pushed into a branch.
I'm not sure what could we do against it, but it could worth some investigation.
>
> Another hypothesis is that there are too many builds being triggered with a
> blanket `ci/*`, which covers all languages, and thus, when changing the
> comment bot, they all get triggered.
I think we improved that, see the cpp paths as an example:
https://github.com/apache/arrow/blob/master/.github/workflows/cpp.yml#L22
>
> What I usually do when playing with builds that can't be tested on my
> personal repo (e.g. because they require credentials), is to delete all
> unrelated workflows during testing, and then checkout them when the PR is
> ready. A bit hackish I must say, but works.
To prevent triggering the upstream builds you can add `WIP` to the PR's title:
https://github.com/apache/arrow/blob/master/.github/workflows/cpp.yml#L52
>
> Best,
> Jorge
>
>
> On Tue, Jan 5, 2021 at 1:34 PM Krisztián Szűcs <szucs.kriszt...@gmail.com>
> wrote:
>
> > Hi,
> >
> > I'm concerned about the overall feedback time we have on pull requests.
> > I have a simple PR to make the comment bot working again, but no
> > builds are running even after 30 minutes.
> > I can see 2-4 running builds, which will make our work much harder
> > right before the release.
> >
> > I wasn't following the build queue's state lately, but I think we
> > should consolidate the build configurations.
> > Possible candidates are the PR* workflows and good to have tests which
> > we could trigger on master instead.
> >
> > Opinions?
> >
> > Regards, Krisztian
> >
