bug#51787: GC takes more than 9 hours on berlin

[Ludovic Courtès]

Hi,

Mathieu Othacehe <othacehe@gnu.org> skribis:

> What is problematic though is that in 20 hours, the free space has
> bumped from 9.6T to 9.7T in the store partition. As the GC lock is
> preventing most of Berlin services from running, almost all the machine
> IO is dedicated to removing this directory, as shown by iotop.
>
> I'm not sure to understand why this removing process is so long, but if
> someone has an idea, I'm all ears. In the meantime, I plan to let the GC
> run and keep monitoring it.

This is the first time GC runs since we’ve increased the threshold from
10 TiB to 15 TiB free.  There are at least 5 more TiBs to delete than
usual, so it’s expected to take more time.

Still, I’m surprised a mere ‘rm -rf’ can take this long.  ‘strace -T’ on
the child guix-daemon process doesn’t reveal anything obviously wrong,
pause times or similar.

Ludo’.