Many random errors on CPU generic workers

[eu9ene]

Example runs:

https://firefox-ci-tc.services.mozilla.com/tasks/groups/b8_PxNNiSKWyuugT0V8faA
https://firefox-ci-tc.services.mozilla.com/tasks/groups/BFgoyXuvRHe0kK4J9d-Lgw
https://firefox-ci-tc.services.mozilla.com/tasks/groups/LphXRhZoScGYSNAPKCiVfQ

See #629

It might be related to #549

[bhearsum]

The common thread here seems to running out of some sort resources. Examples include:

[task 2024-05-23T23:56:57.158Z] [34/12:fasttext_filter] OpenBLAS blas_thread_init: pthread_create failed for thread 22 of 32: Resource temporarily unavailable

[task 2024-05-24T04:15:45.087Z] [24/4:deescape-special-chars] Error: can't start new thread

[task 2024-05-24T01:12:29.457Z] [81/12:fasttext_filter] OpenBLAS blas_thread_init: pthread_create failed for thread 22 of 32: Resource temporarily unavailable
[task 2024-05-24T01:12:29.457Z] [81/12:fasttext_filter] OpenBLAS blas_thread_init: RLIMIT_NPROC 1031641 current, 1031641 max

I suspect podman is enforcing some resource limits. For example, by default there's a limit of 2048 pids in the container:

--pids-limit=limit

Tune the container’s pids limit. Set to -1 to have unlimited pids for the container. The default is 2048 on systems that support “pids” cgroup controller.

(From https://docs.podman.io/en/latest/markdown/podman-run.1.html.)

I'm not sure if we're doing this intentionally or not; I'll look into it.

[bhearsum]

taskcluster/taskcluster#7120 has been filed for this.

[bhearsum]

taskcluster/taskcluster#7120 has been filed for this.

This is fixed. We need to wait for it to be released, and pick up a new version of generic-worker for the cpu workers before to call this ticket fixed. I may wait on updating that image until some of the other things blocking generic worker for cpu workers are dealt with, in case we have other fixes that need to get into the image.