Zfs: lxd/deleted/custom/{uuid} - How is deleted/ managed?

Incus
1

I’m running incus 6.5 on Ubuntu 22.04 from Zabbly packages. This was initially an LXD system, migrated via lxd-to-incus around 6.0.

zfs list
NAME                                                                                                     USED  AVAIL     REFER  MOUNTPOINT
...
pricey-pool/encrypted/lxd/deleted/custom/cca44792-20e7-4700-bf99-60c46cce7df7                           5.71T  7.75T     5.71T  legacy
...

I’ve noticed this odd filesystem which is huge, larger than anything else in the pool. I’m not sure where it came from.

The uuid segment doesn’t mean anything to me. Another pool (created after the incus migration?) contains similar images with what look like hashes as names, rather than uuids.

I have read through the incus docs but haven’t found anything that helps me understand the lifecycle of these deleted filesystems. Can I expect it to be pruned by some process? Is there any documentation on that process?

Thanks for any insight.

2

Normally the delete/XYZ stuff is what happens when you do:

  • incus storage volume create default foo
  • incus storage volume copy default/foo default/bar
  • incus storage volume delete default foo

When doing that, even though you’re deleting foo, it can’t actually go away because it was copied into bar. It will get automatically deleted when bar is deleted.

Can you check zfs list -t all to see if it has any snapshot on it?
Then for any of those snapshots, use zfs get clones XYZ@XYZ to see what’s relying on that snapshot.

3

Thank you, yes you’re right. I copied a storage volume between projects before deleting the original.

I think I’d like to clean that up so space reporting is a little less surprising, but can figure that out for myself.