Timeout errors since 2.4.1

[aduh95]

Describe the bug

In nodejs/node#62219, it was reported our GHA runners show a ~9% timeout rate since we upgrade from 2.4.0 to 2.4.1. We are using the ubuntu-slim runners that have a 15m timeout, but AFAICT it never was a problem until we updated.
I don't know exactly what is the problem, it looks like it sometimes just hand randomly.

To Reproduce

Steps to reproduce the behavior:

I'm not sure, here are the steps that take the Node.js runner: https://github.com/nodejs/node/blob/5e417551e2dbfff0664c4783df16cc32fd8625e5/.github/workflows/linters.yml#L147-L157

Expected behavior

The runner exits successfully in less than a minute 100% of the time

System information

ubuntu-slim GHA runner

Additional context

[bitbloxhub]

Just mentioning that some additional discussion is going on in #661 (comment) if anyone else is looking at this.

[brianmcgee]

I think I might have fixed this with #678 but having a look at what you shared above now and seeing if I can replicate it

[aduh95]

I've tried the PR with https://github.com/nodejs/node-auto-test/actions/runs/23312784437/job/67804213357, looks like we still have a occasional timeout, although it's much better already:

Version	Result	Link
2.4.1	22 / 36 ~ 61%	https://github.com/nodejs/node-auto-test/actions/runs/23313003681
2.4.1 - #661	36 / 36 ~ 100%	https://github.com/nodejs/node-auto-test/actions/runs/23313017583
2.4.1 + #678	35 / 36 ~ 97%	https://github.com/nodejs/node-auto-test/actions/runs/23312784437/

More runs would be needed to get more reliable figures, but hopefully that already gives a good overview.

[brianmcgee]

Ok cool. I spent some time replicating the checkout but I need to call it a day. I'll have some more time tomorrow to dig into it.

[aduh95]

I did some more testing with larger samples, and got more accurate results:

Version	Result	Link
2.4.1	220 / 256 ~85.9%	https://github.com/nodejs/node-auto-test/actions/runs/24000317184
2.4.1 - #661	250 / 256 ~97.7%	https://github.com/nodejs/node-auto-test/actions/runs/24000296288
2.4.1 + #678	244 / 254 ~96%	https://github.com/nodejs/node-auto-test/actions/runs/24000243638
2.4.1 + 8f0a545	245 / 256 ~95.7%	https://github.com/nodejs/node-auto-test/actions/runs/24086953972
2.5.0	256 / 256 ~100%	https://github.com/nodejs/node-auto-test/actions/runs/24103455562

I was wrong saying #661 introduced the problem: apparently the timeout was already happening before, although marginally, and #661 made it noticeable. AFAICT #678 does bring back the reliability almost to the same level as before, so I'd be happy to switch to that if it lands.

[brianmcgee]

Just merged #678. I will find some time to keep digging into the timeouts at some point. I'll try to cut a release later this evening.

[brianmcgee]

@aduh95 https://github.com/numtide/treefmt/releases/tag/v2.5.0

[brianmcgee]

Had a bit of a hiccup with the release, had to recut it to squeak in #679 before it required a breaking change.