dead ssd drive on zfs points to initramfs

[menelaostrik]

I have an ssd drive with failed controller and which was a part of a mirrored zfs vdev. Now it boots into initramfs and says that it can't import pool "rpool". when i issue
zpool import -f rpool

it says that there's no such pool or dataset. However, in "zpool import" can see that rpool has status UNAVAIL because of insufficient replicas. i would like to replace the dead drive with a new one. I have tried:
1) zpool replace 2) zpool offline 3) zpool detach

and everytime i get that there's no such pool or dataset. The only way i got it working was by importing as read only by using:
echo "1" | sudo tee /sys/module/zfs/parameters/zfs_max_missing_tvds zpool import -o readonly=on rpool

screenshot of zpool import: https://prnt.sc/91X8xc3vG1SA
screenshot of zpool status after mounting as read-only: https://prnt.sc/vKzNL0nUFALc

Mounting as read-only boots into the OS. However since it's mounted as readonly i can't actually make any changes to the pool like replacing the damaged drive. Any suggestions?

 

[Dunuin]

But the pool status is just degraded and not failed?

 

[menelaostrik]

But the pool status is just degraded and not failed?

yes, its degraded.
please chckout this screenshot: https://prnt.sc/91X8xc3vG1SA

 

[Dunuin]

To me "rpool" looks like a stripe/raid0 so no parity. Otherwise there should be a "mirror-0" which is missing.

 

[menelaostrik]

To me "rpool" looks like a stripe/raid0 so no parity. Otherwise there should be a "mirror-0" which is missing.

it's a mirror. I don't know when on initramfs its shown like that.
I have it mounted as read-only and the zpool status output is: https://prnt.sc/vKzNL0nUFALc

if just attaching a new drive isn't possible, is there a way of migrating the VMs in a new pve over the network(samba maybe?)

 

[leesteken]

it's a mirror.

No it's RAID0 also known as stripe. Or as ZFS-people call it: "I hate my data"-mode.

I have it mounted as read-only and the zpool status output is: https://prnt.sc/vKzNL0nUFALc

That's why you have 10 data errors that cannot be repaired; it does not have copies for those files. Use zpool status -v rpool (as suggested in the screenshot) to see if your virtual disks are corrupted.

 

[menelaostrik]

No it's RAID0 also known as stripe. Or as ZFS-people call it: "I hate my data"-mode.

That's why you have 10 data errors that cannot be repaired; it does not have copies for those files. Use zpool status -v rpool (as suggested in the screenshot) to see if your virtual disks are corrupted.

omg. you're totally right.
any chance i could restore from this?

root@hyperlan:~# zfs list -t snapshot | grep 101
rpool/data/vm-101-disk-0@autosnap_2022-10-03_09:42:51_monthly   855M      -     89.9G  -
rpool/data/vm-101-disk-0@autosnap_2022-10-05_09:55:18_weekly    454M      -     89.9G  -
rpool/data/vm-101-disk-0@autosnap_2022-10-08_00:00:44_daily     380M      -     89.9G  -
rpool/data/vm-101-disk-0@autosnap_2022-10-09_00:00:51_daily     210M      -     90.2G  -
rpool/data/vm-101-disk-0@autosnap_2022-10-10_00:00:19_daily     145M      -     90.2G  -
rpool/data/vm-101-disk-0@autosnap_2022-10-11_11:44:45_weekly      0B      -     90.5G  -
rpool/data/vm-101-disk-0@autosnap_2022-10-11_11:44:45_daily       0B      -     90.5G  -
rpool/data/vm-101-disk-0@autosnap_2022-10-12_00:00:03_daily       0B      -     90.5G  -
rpool/data/vm-101-disk-0@autosnap_2022-10-13_00:00:03_daily     538M      -     90.4G  -
root@hyperlan:~#

I had also setup backups using syncoid long time ago on the same host. i have found this in crontab
0 15 * * * /usr/sbin/syncoid --quiet --no-sync-snap --recursive rpool backup-rpool/rpool
so it should backup rpool daily on an external hdd.
i have connected that hdd on another host but i can't find the backup sets.
shouldn't they appear using zfs list?