-

Mon, 19 Dec 2022 09:48:42 +0100

Sprite_tm

sprite_tm@social.spritesmods.com

On the btrfs issues I had earlier (btrfs subvolume delete made my kernel use up the entirety of the 4GiB of RAM on my NAS and paniced), looks like btrfs quotas were to blame... disabled quotas, subvolume delete is instant and no more GBs of ram use. It makes it impossible to see how large (as in: how much unique data) each snapshot has, but them's the breaks I guess. The SBC I use for the NAS is a FriendlyARM NanoPC T4 with a M2-to-SATA board, so perhaps it's time to upgrade that anyway when they make a new version with at least more RAM...

Sat, 24 Dec 2022 02:50:48 +0100

Sprite_tm

sprite_tm@social.spritesmods.com

@Richard "mtfnpy" Harman Yeah, I was thinking about that as well... on another system i use Dirvish which is effectively similar to your script, and that works decently. The advantage of the btrfs approach (in theory at least) is that it's block-based and not file-based, so e.g. VM images with only a few bytes changed won't take up the size of the entire VM image on disk.

Sat, 24 Dec 2022 03:01:14 +0100

Sprite_tm

sprite_tm@social.spritesmods.com

@Richard "mtfnpy" Harman Can't do that as my VM server and my NAS are two different systems... VM system runs off SSDs, NAS runs off spinning rust. Otoh, if btrfs entirely doesn't work out, I can always revert to LVM snapshots for the backups.