[SOLVED] ZFS: Changing label after replacing harddisk?

[bowman]

Hi,

after a harddisk-failure i replaced the disk in a zfs-pool and resilvered it (everything is working fine so far). Unfortunately, i choose the device notation for the replacement-harddisk (/dev/sda) and not the ID (containing the serial number) and now the pool looks like this:

    NAME                                          STATE     READ WRITE CKSUM
    local-pool                                    ONLINE       0     0     0
      mirror-0                                    ONLINE       0     0     0
        sda                                       ONLINE       0     0     0
        ata-WDC_WD30EFRX-68EUZN0_WD-WMC4N0N10NER  ONLINE       0     0     0
      mirror-1                                    ONLINE       0     0     0
        ata-WDC_WD30EFRX-68EUZN0_WD-WCC4N6JNVNX4  ONLINE       0     0     0
        ata-WDC_WD30EFRX-68EUZN0_WD-WCC4N6CYKSR1  ONLINE       0     0     0

Is there a simple way to replace the 'sda' with the ID of the replacement-harddisk? the only way i figured out that might (?) work is this:

zpool detach local-pool /dev/sda

zpool attach -f local-pool /dev/disk/by-id/ata-WDC_WD30EFRX-68EUZN0_WD-WCC4N4JNCD8E /dev/disk/by-id/ata-WDC_WD30EFRX-68EUZN0_WD-WMC4N0N10NER

I'd like to avoid another full resilvering (or a complete rebuild of the pool from the backup), so i better ask first... ;-)

 

[Nemesiz]

Have you tried to import pool by disk-id ?

# zpool import -d /dev/disk/by-id/ pool

 

[bowman]

...i tried, but it seems a bit complicated to me: It would require to export the pool first, then to rename it, then to import it as you suggested and then to rename it again. The Problem: The pool is busy and therefore refuses to export. I guess i'd have to shut downs all VM's and to remove the pool from Proxmox - and i'm not shure how to reattach it properly to avoid loosing the VM-Configurations...

 

[Nemesiz]

It is enough to export and import. It does not request to rename or shutdown system nor you loose your VM configurations. If your system is running you don`t need to worry. Export and import the pool after next system reboot.

 

[bowman]

...hmmm: I switched off the autostart of the VM's and rebooted the host: I can export the pool now, but when i try to import it with

zpool import -d /dev/disk/by-id/ local-pool

it says

cannot import 'local-pool': a pool with that name already exists
use the form 'zpool import <pool | id> <newpool>' to give it a new name

I guess it would work if i give it a new name (say 'local-pool2'), but then i will have to rename it back to 'local-pool' to match the VM-configuration, right? (...i'm quite new to ZFS and Proxmox, sorry for the annoying questions... ;-) I just want to avoid risks because i don't have a 2nd system for testing here... ;-)

 

[bowman]

...oh wait! Just after doing the export and a reboot everything is fine now:

    NAME                                          STATE     READ WRITE CKSUM
    local-pool                                    ONLINE       0     0     0
      mirror-0                                    ONLINE       0     0     0
        ata-WDC_WD30EFRX-68EUZN0_WD-WCC4N4JNCD8E  ONLINE       0     0     0
        ata-WDC_WD30EFRX-68EUZN0_WD-WMC4N0N10NER  ONLINE       0     0     0
      mirror-1                                    ONLINE       0     0     0
        ata-WDC_WD30EFRX-68EUZN0_WD-WCC4N6JNVNX4  ONLINE       0     0     0
        ata-WDC_WD30EFRX-68EUZN0_WD-WCC4N6CYKSR1  ONLINE       0     0     0

I have no idea why (the import seems to have been refused), i just did the export...

Anyhow, thanks a lot!

 

[mir]

Maybe this can help you: https://forums.freebsd.org/threads/50481/

 

[bowman]

Maybe this can help you: https://forums.freebsd.org/threads/50481/

...thats where i got my first idea how to fix it from (see initial posting). After switching off the Autostart of the VM's, doing a reboot, the export and starting over the ID of the first disk is shown. Great!

 

[Nemesiz]

zpool import -d /dev/disk/by-id/ local-pool

it says

cannot import 'local-pool': a pool with that name already exists
use the form 'zpool import <pool | id> <newpool>' to give it a new name

To finish this thread I can say that this zpool message you can see only if the pool with name 'local-pool' has been imported already.
#zpool list - can show you it.

 

[GadgetPig]

Don't mean to resurrect this old thread, I thought to share a possible solution from another post here. It might be a similar problem where the disk "label" is missing and/or false disk degraded pool message (but disk checks fine with smartctl)

https://forum.proxmox.com/threads/faulty-disk-or-false-zfs.26956/

Basically he took the GUID of the affected disk and replaced it with itself (using dev-by-id)

(per the post above, the command would have looked like this)

zpool replace -f rpool /dev/GUIDof"SDA" /dev/disk/by-id/ata-WDC_WD30EFRX-68EUZN0_WD-WMC4N0N10NER

For those new to proxmox, you can find GUID using "zbl" command or
blkid (where the GUID is listed under “UUID_SUB=xxxxx”

(They mention to wait until resilver is done before rebooting)

cheers!