@popa.bogdanp not sure. I just want racket/racket:pre-release and racket/racket:snapshot images, somehow

Hmm. The CI has been failing with

find: '/__w/racket/racket/racket/bin': No such file or directory since Monday

https://github.com/racket/racket/runs/3137538120

Thanks - on it now.

Somehow the build failed in the previous step — before the find one

compiling <http://cpprim.ss\|cpprim.ss> with output to cpprim.ta6le make[13]: *** [Mf-base:589: bootall] Killed make[12]: *** [Mf-cross:37: xboot] Error 2 make[11]: *** [Mf-boot:25: ta6le.boot] Error 2 make[10]: *** [Makefile:66: ta6le.bootquick] Error 2 make[9]: *** [Makefile:58: ta6le.bootquick] Error 2 make[9]: Leaving directory '/__w/racket/racket/racket/src/build/cs/c/ChezScheme' make[8]: *** [Makefile:156: pb-bootquick] Error 2 ...

yes, just noticed the same thing.

Shouldn’t make returns an exit code that makes the build fails right there, instead of letting it go one more step?

Also this:

https://github.com/racket/racket/runs/3137538119#step:8:25

It’s deep inside the makefile not sure if it’s a makefile issue or a workflow issue

@mflatt probably knows about this already

this has been happening for a few days…

hmm, there’s a make ... \| tee in the command

If this is bash, we probably need something like set -o pipefail

first time this failed was for https://github.com/racket/racket/actions/runs/1050087507

https://github.com/racket/racket/commit/6901f8d5511fd45516aa0a85ae070292e64226cc

it feels like that this commit introduced the bug.

This is very weird: https://github.community/t/pipefail-doesnt-work-with-pipes/17397

The answer claims that set -o pipefail is already set

strange indeed.

GHA sometimes is quite flaky…

I mean this https://github.com/racket/racket/runs/3135807712?check_suite_focus=true

has been running for over 4hrs. and the interface doesn’t even show the logs.

I need to go to the raw logs to see:

2021-07-22T15:49:18.1470267Z Found online and idle hosted runner(s) in the current repository's organization account that matches the required labels: 'ubuntu-18.04' 2021-07-22T15:49:18.1470317Z Waiting for a hosted runner in 'organization' to pick this job...

what’s strange though… is that this is not one of my self-hosted runners. this should be running on github machines.

huh - unless they deprecated ubuntu-18.04

nope - should still be available

https://github.com/racket/racket/issues/3932

https://github.com/actions/virtual-environments/issues/1877#issuecomment-716627187

excessive artifact usage - first time I hear about that.

I pinged the github.community https://github.community/t/hosted-runners-not-picking-up-jobs/191663

I think that just got fixed

Relatedly, do people get notified by the CI failures? Another point is, do they generate too many false positive that you ignore them?

I notice them but arranging for more of them to be posted to slack would be good

I think I would have picked up on this one faster with less CI noise. Also, a different problem created too much noise on DrDr, but I guess I would have noticed this problem there soon (e.g., if I had looked at the most recent DrDr run).

I think we’ve had a bunch of CI noise because we had a bunch of CI failures for genuine reasons recently, but I also agree that we have too many spurious failures

I think a :snapshot image would be doable via a GHA workflow on a timer that rebuilds and pushes the image when the snapshot version changes (at the top of the page https://www.cs.utah.edu/plt/snapshots/\|here).