I don't mess with containers very much--I prefer the much greater security and load isolation offered by VMs--but I pretty regularly get questions about the best way to set up Docker and/or LXC with ZFS datasets.

I'm happy to tell you, today I discovered a pretty thorough guide that includes--among other things--how to properly isolate ephemeral data from necessary data, in order to de-clutter and reduce the load of your backup and restoration routines.

https://du.nkel.dev/blog/2026-05-16_rootless_docker_virtiofs_proxmox/