This message was deleted.

https://longhorn.io/kb/space-consumption-guideline/#disk-exhaustion

To do recover nodes and disk, we would recommend directly removing some redundant replica directories for the full disk. Here redundant replicas means that the corresponding volumes have healthy replicas in other disks.

How do I do this? Is it safe to just delete files in

/var/lib/longhorn/replicas/

??

Don’t you use a lvm for longhorn data?

Not currently, though we probably should 🙃 For production deployments we plan on it but our testbed was just a PoC and we went without an LVM partition. I ended up getting things back. Turns out you can delete

/var/lib/longhorn/replicas/

if nessecary and longhorn will re-create the replicas once it's back Doing a docker image prune / crictl rmi --prune saved a ton of space since we've been cycling docker images often

I suggest you to have the same setup in test as you plan to have in prod. - It’ll help you out a lot in the future.