When we have a large queue of blocked pipelines, we should cancel the validation/nightly pipeline on master since they'll consume the runner resource for too long, and runners should be prioritized for marge batches and regular MRs. Cheers, Cheng On Wed, Nov 6, 2024 at 11:34 AM Andreas Klebinger via ghc-devs < ghc-devs@haskell.org> wrote:

Hi Simon,

I took a brief look and it might just be CI being strained from ghc-9.12.1-alpha2 and related CI jobs.

It seems the number of pending jobs already went down slightly between when I first looked at it and just now. I would expect to resolve itself in the next few hours. But I will take a closer look in a few hours if the issue persists.

Cheers Andreas

Am 06/11/2024 um 09:37 schrieb Simon Peyton Jones:

Dear GHC devs

Is something stuck in CI?

https://gitlab.haskell.org/ghc/ghc/-/merge_requests/13553 has been stalled for 9 hrs. I have not seen that before.

Thanks

Simon

_______________________________________________ ghc-devs mailing listghc-devs@haskell.orghttp://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

_______________________________________________ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

I have been told by Equinix representatives that if we built something that dynamically scales up and down the number of machines we use, not only would it *not* impact their bottom line very much, but it would actually be *valuable* to Equinix. Their sponsorship of us comes out of their sales/marketing budget. If we can create cool things they can show off ("look at the cool things you can do on our platform"), they would appreciate it. If I had more time to spend on DevOps, this is something I'd be interested in building. On Wed, 6 Nov 2024, 14.44 Oleg Grenrus, wrote:

OTOH, if nightly pipeline is cancelled for a week in succession, and *then* it fails, we are screwed.

In my experience failing marge jobs is what consumes a lot of CI time.

In summary, until there is are analytics where the CI time is actually spent, it's at best guesses, at worst wrong guesses.

- Oleg On 6.11.2024 13.55, Cheng Shao wrote:

When we have a large queue of blocked pipelines, we should cancel the validation/nightly pipeline on master since they'll consume the runner resource for too long, and runners should be prioritized for marge batches and regular MRs.

Cheers, Cheng

On Wed, Nov 6, 2024 at 11:34 AM Andreas Klebinger via ghc-devs < ghc-devs@haskell.org> wrote:

...

Hi Simon,

I took a brief look and it might just be CI being strained from ghc-9.12.1-alpha2 and related CI jobs.

It seems the number of pending jobs already went down slightly between when I first looked at it and just now. I would expect to resolve itself in the next few hours. But I will take a closer look in a few hours if the issue persists.

Cheers Andreas

Am 06/11/2024 um 09:37 schrieb Simon Peyton Jones:

Dear GHC devs

Is something stuck in CI?

https://gitlab.haskell.org/ghc/ghc/-/merge_requests/13553 has been stalled for 9 hrs. I have not seen that before.

Thanks

Simon

_______________________________________________ ghc-devs mailing listghc-devs@haskell.orghttp://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

_______________________________________________ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

_______________________________________________ ghc-devs mailing listghc-devs@haskell.orghttp://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs

_______________________________________________ ghc-devs mailing list ghc-devs@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs