Swapping out a drive in proxmox, without knowing which physical drive.

The instructions on how to replace and resilver a drive in the command line for zfs

See the current devices.

zpool status
zpool status 'poolname'

Take a note of all the device ID in the pool you want to replace.

Shutdown the server and replace one drive.

zpool status 'poolname'

should show you that the drive is unavailable. We need to stop ZFS from looking for the file by setting it to offline, using the driveID shown in the status command.

zpool offline 'poolname' 'driveid'
zpool offline rpool 15935140517898495532

We need to identify the new drive.

lsblk

You should be able to find the new drive in here, likely that it does not have a mount point or any partitions on it.

To add it to the ZFS pool, we must 'replace' the old drive so that this new one starts to 'resilver' or rebuild the raid array.

zpool replace 'poolname' 'old-drive-id' 'new-drive-id'
zpool replace rpool 15935140517898495532 nvme1n1

This will begin the resilvering automatically, and you can manually check the stauts of the rebuild with.

zpool status 'poolname'

you can use the watch command to display automatic updates.

watch -n 5 'zpool status poolname'

Proxmox zpool of boot drives

[[proxmox]]

Changing the SSD because of wearout.

We need to create the same UEFI boot partitions for the system to be able to boot from either of the drives, Otherwise the drive will not be a bootable device and will only have the 'other' proxmox data. Technically we are only putting the 'data' partition in raid. the EUFI partition is not. We rely on the proxmox tool to keep the EUFI partition in sync for us.

1. Change out one of the physical drives.

2. Find the device ID for the new drive you have entered, can be veified with these 2 commands.

   lsblk #shows the actively connected block devices ls -l /dev/disk/by-id #Shows all system links, by name. ``` 
   

3. You should be able to use the 2 to find the new device, likely nvme#n# with the # representing an incrementing number. You also need to find the ID of the current existing drive.

   We need to copy the partition layout from the existing device to the new one, nvme0n1 is the old device and vnme1n1 is the new. Replace them with your disk IDs.

   sgdisk info found here: https://linux.die.net/man/8/sgdisk

   sgdisk /dev/nvme0n1 -RG /dev/nvme1n1
   

4. Now we setup the UEFI bootloader. Again replace the drive ID with your ID. The p# on the end is choosing the partition to use. Make sure you pick the correct one.

   More info on exactly what this does here. https://pve.proxmox.com/wiki/Host_Bootloader

   The status command should show you the new EFI partition and the old one should no longer be found, The clean function should remove it.

   pve-efiboot-tool format /dev/nvme1n1p2 --force
   pve-efiboot-tool init /dev/nvme1n1p2
   pve-efiboot-tool status
   pve-efiboot-tool clean
   pve-efiboot-tool status
   

5. Next we get the status of the existing ZFS pool. rpool is the name of your pool

   zpool status rpool
   

6. This should show you which of the device is unavailable. We need to set it to offline, replace the drive with your ID

   zpool offline rpool nvme-eui.6479a7304169387a-part3
   

7. Now we need to replace the drive with our new one. This will start resilvering automatically, use the ID of your respective disks. nvmep3, replace with the p3 with the number of the partition that is relevant to you, most likely it is 3.

   zpool replace rpool nvme-eui.6479a7304169387a-part3 nvme1n1p3
   

8. Check the status of your pool, it should be resilvering.

   zpool status rpool
   

9. When it is finished the offline drive should not be in the status list anymore and the status of the pool should return to 'ONLINE'.