Did a high-level test between a PVE+Ceph cluster and a single PVE node with the remote-migration of a Windows Server 2022 VM with EFI & TPM disks.

Ceph RBD -> remote LVM thin
LVM thin -> remote Ceph RBD

Worked in both directions, and the VM booted up as expected after each migration.

One thing I ran into, only tangentially related to this series, is that we don't support the 'raw+size' option for ZFS. Maybe we can get it working on ZFS at least for VM disk images (zvol)?

Maybe it might also be time to consider if we want to handle CT volumes differently on ZFS in the long term (file based dataset). In all other storage options we have a block dev or raw file that we loop mount into the CT. Aligning this with ZFS would probably simplify things quite a bit.


With the above mentioned tests, partially:
Tested-By: Aaron Lauterer <a.laute...@proxmox.com>


_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Reply via email to