Skip to contentSkip to conversation

Lots of trouble hanging when rm files with many extents

DISCUSSION Tomasz Chmielewski 2 participants 3 messages Oct 20, 2013 – Oct 25, 2013 Btrfs file system

lore
Share & tools
lore
#0 lore raw
Quoted 4 lines

but I am not sure if that was the right thing to do. This is producing OOM issues and leading to system crashes (including eventual panics) with such alarming frequency that I wonder if perhaps there is something different about my setup than others.

FYI, I'm seeing similar / random hangs on systems with btrfs, compress-force=zlib, qgroups enabled. The systems have 32 GB RAM. Sometimes, I can review old logs (after force reboot) and see OOM, allocation failures and such.

The scenario there are snapshots, rsync with --inplace flag, which can fragment files a lot.

I'll do some more tests with lots of extents to see if it's reproducible here as well.

Signature · 2 lines
Tomasz Chmielewski
http://wpkg.org
#1 lore raw
Quoted 9 lines · Tomasz Chmielewski
Quoted 4 lines

but I am not sure if that was the right thing to do. This is producing OOM issues and leading to system crashes (including eventual panics) with such alarming frequency that I wonder if perhaps there is something different about my setup than others.

FYI, I'm seeing similar / random hangs on systems with btrfs, compress-force=zlib, qgroups enabled. The systems have 32 GB RAM. Sometimes, I can review old logs (after force reboot) and see OOM, allocation failures and such.

Quoted 7 lines

The scenario there are snapshots, rsync with --inplace flag, which can fragment files a lot.

I'll do some more tests with lots of extents to see if it's reproducible here as well.

Interestingly, I've generally had qgroups enabled here as well, possibly on all of these systems. Could that be the culprit?

#2 lore raw

rm with quotas can cause panic (was Re: Lots of trouble hanging when rm files with many extents)

Quoted 8 lines · John Goerzen <jgoerzen <at> complete.org> writes:
Quoted 4 lines · Tomasz Chmielewski

I'll do some more tests with lots of extents to see if it's reproducible here as well.

Interestingly, I've generally had qgroups enabled here as well, possibly on all of these systems. Could that be the culprit?

Updating the subject here, because it seems increasingly likely that quotas are the problem causing OOM panics.

I was able to reproduce the problem on a test filesystem with quotas enabled. After quotas were disabled and the system was rebooted, the problem no longer occurred. The dangerous combination seems to be many extents plus quotas.

Sadly, quotas are the only way to have btrfs keep track of how much space subvols are using, so it's very useful to be able to use that feature.

I also noted this at https://bugzilla.kernel.org/show_bug.cgi?id=63071

John