WIP: Partitioning locking and refactoring

Open tbzatek opened this issue 4 years ago • 1 comments

Constantly failing tests in #933 led me to chase for an explanation and managed to find a deficiency in object update after partitioning. Just another race condition that only reproduced in the CI, never locally.

The biggest difference made changing the advisory lock from shared to exclusive.

Jan 08 '22 21:01 tbzatek

I'm still thinking whether we should try harder with acquiring exclusive lock when busy - 500ms might not be enough. Specifically the mentioned blogpost says:

... in case systemd-udevd is still processing the device the tool will wait for it to finish.

Jan 09 '22 12:01 tbzatek