12:49 <walters> hartan: but you're still seeing `rpm-ostree cleanup -prb` failing with that error?

12:52 <walters> hartan: can you pastebin the output of `findmnt` ?

13:00 <walters> hartan: interesting...hmm. May relate to btrfs subvolumes somehow. Let's try `find /sysroot -name '*.boltdb'` (i.e. drop -xdev) but this will be more expensive as it will traverse everything

13:04 <walters> right, it shouldn't because the host daemon shouldn't be touching your home directory

13:04 <hartan[m]> Colin Walters: Finds exactly one result in my users podman cache. That doesn't count, right? :)

13:04 <walters> for the daemon we actually make /home read-only precisely to avoid potentially affecting user data...which may be the safety measure we're somehow hitting

13:05 <walters> when you say "users podman cache" you mean /home/username/.local/share/containers right ?

13:13 <walters> hartan: ok can you spin up a toolbox https://docs.fedoraproject.org/en-US/fedora-coreos/debugging-with-toolbox/

13:16 <walters> hartan: inside the toolbox shell, `dnf -y install perf`, then `perf trace &> /run/host/tmp/perf.trace&` and then `chroot /run/host rpm-ostree cleanup -prb`; then `fg` and we should have a system call trace

13:21 <walters> hmm maybe for this we do need pathname decoding which is going to want strace

13:23 <walters> try then `chroot /run/host/ systemctl restart rpm-ostreed; strace -s 2048 -f -p $(pidof rpm-ostree) -o /run/host/tmp/strace.log& chroot /run/host/ rpm-ostree cleanup -prb; fg`

15:21 <walters> hartan: please do e.g. `grep rpm-ostree perf.trace` (to filter out other stuff) and pastebin or so

15:25 <walters> hartan: sure

15:46 <dustymabe> hmm I feel like there are messages that aren't coming in here

15:46 <dustymabe> https://libera.irclog.whitequark.org/fedora-coreos/2023-05-11

15:46 <dustymabe> looks like walters is having a conversation with hartan, but I don't see hartan saying much

16:53 <walters> I also context switched, but let's indeed take this async to the issue

16:57 <dustymabe> ok yep. I just logged in to matrix and can see messages from hartan

16:57 <dustymabe> anyone else not seeing them on the IRC side?

18:48 <dustymabe> jlebon: I imagine we'll want to backport https://github.com/coreos/coreos-assembler/pull/3467 to older release branches but it probably won't apply cleanly with the second commit in there (i.e. /cherry-pick automation) should we drop that?

19:16 <jlebon> dustymabe: good point. i split it into https://github.com/coreos/coreos-assembler/pull/3468

19:17 <dustymabe> +1

20:07 <dustymabe> jlebon: can you explain to me what the guestfish debug command does? I swear I was trying to do bindmounts the other day with just guestfish sh by itself and was having the worst luck.

20:10 <jlebon> dustymabe: my understanding is `command` and `sh` assume a traditional guest OS and runs "as if" it ran in the guest environment. `guestfish debug sh` runs directly in the supermin environment.

20:12 <dustymabe> weird. i couldn't really find that in the docs much

20:18 <jlebon> yeah, it's underdocumented

20:18 <jlebon> (probably on purpose)

20:58 <dustymabe> thanks for the context! it helps alot