I have a 10T zfs pool set up on the host (that was necessary due to a migration - see my post history...) with a bunch of datasets. I am currently sharing them all via nfs to a proxmox VM, the only VM running on this particular server. The VM runs Debian. nfs is akward due to UID mapping and I'd rather keep the Hypervisor as just that. I have 2 clustered proxmox instances.
Questions:
Questions:
- I believe I can zfs pool export, umount the pool from the host, do a qm set for all disks towards the QEMU VM, and do a zfs import there and everything should work as before. Is that best practice?
- For backing up the VMs themselves (all 32-64GB disks, i.e. fairly small), will having disk passthrough enabled cause any issues? I want to use proxmox for that and back up all VMs to the respective other server. The backup disks for the VMs will not be part of the aforementioned ZFS pool, of course.
- How do I do data/zfs backups? I have a separate proxmox instance where I can replicate the same process (another 10T zfs pool) - would a zfs send | ssh ... be a good route or are there better tools within the proxmox universe? I assume the answer is "no", given that proxmox wouldn't be aware of the zfs pool anymore.