integration tests failing

[geemus]

@Temikus - do you have any ideas on what might be happening with the integration tests? Seems to be failing when trying to bundle and reference the local stuff. It's not obvious to me that my recent merge impacted it, so I wasn't sure what else may have changed. You can see an example of the failure here: https://github.com/fog/fog-google/actions/runs/4767247086. Thanks!

[Temikus]

Bundler fails for some reason:

Using fog-google 1.19.0 from source at `.`
Error: The process '/opt/hostedtoolcache/Ruby/2.7.8/x64/bin/bundle' failed with exit code null
##[debug]Node Action run completed with exit code 1
##[debug]Finishing: Set up Ruby

I suspect the new Ruby 2.7.8 has some caveats. This is a local runner, so I should be able to debug a bit.

[Temikus]

Cannot repro on the runner either 🤔

Using fog-google 1.19.0 from source at `.`
Bundle complete! 14 Gemfile dependencies, 78 gems now installed.
Bundled gems are installed into `./vendor/bundle`

[Temikus]

Lemme try running from master to exclude any weirdness.

[Temikus]

The only difference with unit that passed is:

        env:
          # Needs to be set up for self-hosted runners, see:
          # https://github.com/ruby/setup-ruby#using-self-hosted-runners
          # Image used in runners: summerwind/actions-runner
          ImageOS: ubuntu20

Maybe actions-runner no longer needs it? Let's try it out I guess 🤔

[Temikus]

That didn't work either - kinda bizarre 🤔 I'll dig a bit more this evening.

[Temikus]

Now it's running fine in #594 🙃 I think there's something weird with our runners 🤔

[geemus]

Weird. Maybe just some kind of intermittent issue...

[Temikus]

I'm gonna try and add some debugging now and see how it goes on the main branch again.

[Temikus]

Pushed a new debug workflow in, I'll see what it spits out 🤔

[Temikus]

Ok, so something looks to be killing the runners:

Installing parser 3.2.2.1
/runner/_work/_temp/b9ca9963-76ef-47c5-b939-eab94c68eea3.sh: line 1:   354 Killed                  bundle install --verbose
Error: Process completed with exit code 137.

Looks like we're OOM'ing:

Memory cgroup out of memory: Killed process 2996984 (bundle) total-vm:777476kB, anon-rss:325672kB, file-rss:4792kB, shmem-rss:0kB, UID:1000 pgtables:860kB oom_score_adj:937

[Temikus]

Resized everything, will push a change soon as soon as the cluster rescales.

[Temikus]

Tests are looking much better - I will wait for current queue to finish and then restore caching and see if that was the reason for ruby/setup-ruby failing.

[Temikus]

Aight, tests are all green, trying with the ruby/setup-ruby 👍

[geemus]

@Temikus Awesome, thanks again for digging in and getting to the bottom of this.

[Temikus]

No worries, everything's looking good now ( except the usual Google API jitter :/ ) - I'll merge some fixes I made along the way soon 👍

[geemus]

Sounds good.