Thanks for investigating these @samth . I am trying to reproduce them locally.

Quite frustrating that act doesn’t seem to work too well with our workflows for some reason.

I can’t seem to really reproduce them on my machine.

I re-ran a couple of jobs to see if it keeps happening in the same place.

ubsan got stuck on test macro

Locally test (section) macro doesn’t take that much time.

Although I do get an error in Section module:

Section(module) Performed 641 expression tests (356 value expressions, 285 exn expressions) and 818 exception field tests. Errors were [1]: (Section (got expected (call))) ((module) (#f #t ((has #<path:/var/tmp/the-file.rkt>) #f)))

Starting to wonder if for some reason they are throttling the project due to the amount of workflows, in which case I will contact support directly.

Also locally on ubsan 3m: ```

Section(parameters) Performed 387 expression tests (238 value expressions, 149 exn expressions) and 234 exception field tests. Errors were [1]: (Section (got expected (call))) ((parameters) (#f #t (#<procedure:regexp-match?> #rx"repeats[^\n]+[0-9]+[^\n]+times[^\n]*\n [.][.][.]\n$" "error: repctx\n context...:\n /home/pmatos/dev/racket/pkgs/racket-test-core/tests/racket/param.rktl:556:0: repctx\n [repeats 17 more times]\n")))

Otherwise ubsan finished without issues.

Build failed again after 6hrs. Or better, were cancelled when they hit the time limit.

Cannot understand why suddenly they started failing.

I have seen before a way to ssh to hosted runners to debug…

I don’t think it was : https://github.com/marketplace/actions/debug-via-ssh

but this could do the trick.

Last time CI Linux x86 worked was : https://github.com/racket/racket/actions/runs/1037588418

12 days ago.

it lasted 1h31m.

Now they all get cancelled after 6 hours.

Exactly same story for ubsan.

after commit “add binding-space support to provide and require” : https://github.com/racket/racket/commit/0d526f54c77b878de64ebc4fa1b052bc9cc16a57 things just started failing.