ZFS problems, boot hangs!

Solouncapitano

New Member
Dec 18, 2024
4
0
1
Suddenly after a reboot due to electricity problem, proxmox hangs and doesn't boot.
I have a ZFS pool called rpool with data, I do not know how to "fix" the problem, tried ubuntu live but I was unable to mount zfs partition.

Would you please help ?

Some screenshots belowIMG_1305-min.jpg


error-min.jpg
 
ZFS panic on pve root after power outage ... that comes not unexpected. If you cannot import pool on any live-linux boot you need a new pve installation.
 
ZFS panic on pve root after power outage ... that comes not unexpected. If you cannot import pool on any live-linux boot you need a new pve installation.
would you please share the import command ? is something like "zpool import" ?
Also, another option, Launching Proxmox iso installer will recognize existent pool/installation?

Thks
 
What about "zpool import -o readonly=on rpool rpool2 -f -d /dev/disk/by-id" from your live-ubuntu ?
When you are using zfs you need to learn troubleshooting and to test how to restore a full system.
Operating a ZFS without UPS with server control is even negligent as you see yourself.
 
What about "zpool import -o readonly=on rpool rpool2 -f -d /dev/disk/by-id" from your live-ubuntu ?
When you are using zfs you need to learn troubleshooting and to test how to restore a full system.
Operating a ZFS without UPS with server control is even negligent as you see yourself.
if "zpool import" says nothing to import ?
 
It cannot read the 4 (anywhere 2 in front and 2 at end of disk !!) zfs metadata blocks ... how the hell could all 4 blocks (of each disk) couldn't be read while corrupt if zfs doesn't do it by itself - there is not bit rot which exactly get these 4 hits same time, that's zfs itself which destroys itself.
When you cannot repair their info (with backup info) you won't be able to import and need a new installation.
 
I strongly believe not as I think it would do new partitioning based on disk size. Even if it's the same disk the result would be the same but I would just dream if your vm/lxc images were intact in the data volume group as there theoretically would go no writes inside. You need to try but nevertheless the machine config files in /etc/pve must be recovered anyway. Good luck.