It's been working fine. I have the pi's set up to reboot on disk failures, which fixes 99% of the problems (looking at the up time, ony node still reboots once

It's been working fine. I have the pi's set up to reboot on disk failures, which fixes 99% of the problems (looking at the up time, ony node still reboots once ever 3 weeks or so), and the pods just get migrated, it's not a huge deal. I use longhorn though, not CEPH, and it's been relatively painless.