Comment # 2 on bug 1162471 from Wenruo Qu

(In reply to Jiri Slaby from comment #1)
> (In reply to Dominique Leuenberger from comment #0)
> > for Snapshot 0131, the kernel was updated to version 5.5.0. The test fails
> > with 'no space left' on the disk. Considering that this test runs on VM with
> > HDD configured for 40GB, this is unlikely to happen.
> > 
> > I reproduced this locally in a VM (based on the not-published snapshot).
> > After the machine was up for a while and I keep on installing programs
> > (while monitoring the disk usage) I saw an immediate jump at one point from
> > 31GB free storage to 0 bytes free. After running btrfs balance, the 31GB
> > where quite immediately reclaimed.
> 
> Guys, any ideas what could break btrfs this way in 5.5?

A known bug, caused by v5.4 kernel where metadata over-commit behavior changed.

The fix is already in David's tree, would be sent to upstream in next pull
request.

SLE kernels aren't affected yet.