@polarathene I'm a bit afraid that #2729 shows that there still are timing-related problems with our tests, but test failures are now happening in PRs for which tests should run fine. Anything we can do here as a follow-up?

See #2752 (comment).

but test failures are now happening in PRs for which tests should run fine. Anything we can do here as a follow-up?

Can you link to the PR with failing test?

I saw this one, but not sure if this was due to the issues you fixed in later commits? (due to a package you had omitted from the build)

Early setup of test/tests.bats is the area that's been mostly sensitive to timing.

There's also a check-for-changes.sh test for remove_lock() which failed one time likely due to timing issues, but the method didn't output anything to log when the conditions were not met IIRC. If it happens again, I'd patch it, otherwise when I have time later in the year I would like to tackle switching back to flock().

See #2752 (comment).

Overlooked that. I thought the comment was about the failing tests 😟

See #2752 (comment).

Overlooked that. I thought the comment was about the failing tests worried

No worries. Due to the large feedback, I will provide separate PRs and close #2752 :)