On 05/05/2015 07:31 AM, Anton Aylward wrote:
So we wold have this happeining with ext4! The issue isn't BtrFS in that case, its snapper.
We shouldn't be blaming BtrFS specifically.
Yes we should be blaming Btrfs specifically. These snapshots aren't a function of snapper, it is just the control mechanism. The Snapshots in question are a built in function of BTRFS, the snapshots thus created are not portable. If BTRFS better balanced its trees and was less subject to breakage and disk hogging this might not be an issue. But because Opensuse snapshots the root partition which includes some very volatile sub-partitions, and that volatility results rapid accumulation of disk space ALL out of a shared pool, the problem is compounded. These disk full situations started showing up as early as 2012, on a lot of distros when ever BTRFS was pressed into service. Btrfs snapshots are described to be diffs of just the changes since the prior snapshot. And the prior snapshot is diffs of changes since the one before that. Turtles all the way down. They should be better than most snapshot systems. The problem with this, is if you remove the oldest snapshot you have to roll all the contents forward into the next snapshot just to have a base image somewhere. With a very volatile /tmp included, each snapshot will be much bigger, and the purging of old snapshots more painful. You would expect marginal gains in free space as this trimming of old snapshots progresses, but this seems not to be happening. No clues beyond that, other than we can't absolve btrfs of blame yet. -- After all is said and done, more is said than done. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org