[TUTORIAL] FabU: I have two (or at least one) additional HDDs in spare - how can I use them to create Backups of my VMs and Containers?

UdoB · Aug 27, 2025

Target audience: Beginners

There are no secret tips-n-tricks here, just basic knowledge.

Precondition

you have an intact PVE installation, be it a single computer or a cluster with several nodes; I am using the current 9.x release
you have HDDs which may contain old data - that data will get erased here completely! The disks may have slightly different sizes
you have free connectors to attach all new disks

Disclaimer

There is no way I could foresee each and every detail of a user’s system and there may be errors in my writing. Whatever you do with your system: you are responsible, not me. Make a backup first! (If possible..., we are talking about creating a new backup destination here --> chicken-and-egg problem.)

Notes

everything here is common knowledge; everything here should be documented in the reference documentation (#1) or in the wiki (#2); only the specific sequence in “walk-through”-style makes it worth being "postable"
I will talk about adding two drives to be used as a ZFS mirror. Everything will work too if you have only one drive.
a PBS, built upon a separate computer is recommended - however this FabU is not about that approach
we are talking HDDs because that is what the target audience might have recycled from an old computer, everything here works with SSD too
(nearly) everything is possibly through the CLI or alternatively the GUI - except noted

Hardware

I use a virtual PVE instance named “pnm” for the following. A cluster (of whatever size and shape) is probably not what a beginner has available at this stage of the journey. This is a fresh machine, without any additions I do usually add. And it has quirks: I have installed the PVE operating system with ZFS-on-ZFS on a single disk - both is a big no-no for any other use cases than teaching/learning. Being virtual is also the reason why my “physically” added disks are named “QEMU harddisk”.

Step-by-step

1 - document which drives you currently have

write down manufacturer, model name and serial number
check which drives are currently active:

CLI:

lsblk -o+FSTYPE,MODEL | grep -v zd | tee ~/lsblk-pre-adding-drives.txt

Example:

Code:

root@pnm:~# lsblk -o+FSTYPE,MODEL | grep -v zd  | tee ~/lsblk-pre-adding-drives.txt
NAME   MAJ:MIN RM  SIZE RO TYPE MOUNTPOINTS FSTYPE     MODEL
sda      8:0    0   32G  0 disk                        QEMU HARDDISK
├─sda1   8:1    0 1007K  0 part        
├─sda2   8:2    0  512M  0 part             vfat
└─sda3   8:3    0 31.5G  0 part             zfs_member
sr0     11:0    1 1024M  0 rom                         QEMU DVD-ROM

GUI:

look at “Server View” --> “<yournodename>“ --> “Disks” and write down what you see (or print it)

2 - connect both drives

Turn off power, connect everything, turn on. The system should boot as usual. If not:

try to access a “boot menu”
search for “boot problems”... as I won’t talk about it here

3 - learn which drives you actually added

CLI:

lsblk -o+FSTYPE,MODEL | grep -v zd | tee ~/lsblk-after-adding-drives.txt

Example:

Code:

root@pnm:~# lsblk -o+FSTYPE,MODEL | grep -v zd  |  tee ~/lsblk-after-adding-drives.txt
NAME   MAJ:MIN RM  SIZE RO TYPE MOUNTPOINTS FSTYPE     MODEL
sda      8:0    0   32G  0 disk                        QEMU HARDDISK
├─sda1   8:1    0 1007K  0 part        
├─sda2   8:2    0  512M  0 part             vfat
└─sda3   8:3    0 31.5G  0 part             zfs_member
sdb      8:16   0   32G  0 disk                        QEMU HARDDISK
├─sdb1   8:17   0    1G  0 part        
└─sdb2   8:18   0   28G  0 part             ntfs
sdc      8:32   0   32G  0 disk                        QEMU HARDDISK
├─sdc1   8:33   0  512M  0 part        
├─sdc2   8:34   0    1K  0 part        
└─sdc5   8:37   0   25G  0 part             vfat
sr0     11:0    1 1024M  0 rom                         QEMU DVD-ROM

Code:

root@pnm:~# diff ~/lsblk-pre-adding-drives.txt ~/lsblk-after-adding-drives.txt
5a6,12
> sdb      8:16   0   32G  0 disk                        QEMU HARDDISK
> ├─sdb1   8:17   0    1G  0 part        
> └─sdb2   8:18   0   28G  0 part             ntfs
> sdc      8:32   0   32G  0 disk                        QEMU HARDDISK
> ├─sdc1   8:33   0  512M  0 part        
> ├─sdc2   8:34   0    1K  0 part        
> └─sdc5   8:37   0   25G  0 part             vfat

GUI:

look at “Dashboard” --> “Storage/Disks” and compare it with the version from above

Obviously in my setup sdb + sdc are my newly added (but formerly used) disks.

4 - prepare those disks

CLI:

sgdisk --zap-all /dev/sdb # THIS IS DESTRUCTIVE - tripple check the device name! (#3)

Example:

Code:

root@pnm:~# sgdisk  --zap-all /dev/sdb
GPT data structures destroyed! You may now partition the disk using fdisk or
other utilities.

Repeat for all newly added disks.

The result:

Code:

root@pnm:~# lsblk -o+FSTYPE,MODEL | grep -v zd
NAME   MAJ:MIN RM  SIZE RO TYPE MOUNTPOINTS FSTYPE     MODEL
sda      8:0    0   32G  0 disk                        QEMU HARDDISK
├─sda1   8:1    0 1007K  0 part        
├─sda2   8:2    0  512M  0 part             vfat
└─sda3   8:3    0 31.5G  0 part             zfs_member
sdb      8:16   0   32G  0 disk                        QEMU HARDDISK
sdc      8:32   0   32G  0 disk                        QEMU HARDDISK
sr0     11:0    1 1024M  0 rom                         QEMU DVD-ROM

Note that sdb+sdc don’t have any partition anymore.

GUI:

on the “Disks” view select one of sdb/sdc and click the button at the top: “Wipe Disk”; repeat for all added disks.

Only now we have “officially” empty disks available

5 - create a new ZFS pool

CLI:
We will NOT use “sdb”/“sdc” here. That is considered bad practice as those names may change dynamically, e.g. when adding/removing other PCI cards. Instead we look up the right identifier like this:

Code:

root@pnm:~# ls -Al /dev/disk/by-id/
total 0
lrwxrwxrwx 1 root root  9 Aug 27 09:37 ata-QEMU_DVD-ROM_QM00003 -> ../../sr0
lrwxrwxrwx 1 root root  9 Aug 27 09:37 scsi-0QEMU_QEMU_HARDDISK_drive-scsi0 -> ../../sda
lrwxrwxrwx 1 root root 10 Aug 27 09:37 scsi-0QEMU_QEMU_HARDDISK_drive-scsi0-part1 -> ../../sda1
lrwxrwxrwx 1 root root 10 Aug 27 09:37 scsi-0QEMU_QEMU_HARDDISK_drive-scsi0-part2 -> ../../sda2
lrwxrwxrwx 1 root root 10 Aug 27 09:37 scsi-0QEMU_QEMU_HARDDISK_drive-scsi0-part3 -> ../../sda3
lrwxrwxrwx 1 root root  9 Aug 27 09:45 scsi-0QEMU_QEMU_HARDDISK_drive-scsi1 -> ../../sdb
lrwxrwxrwx 1 root root  9 Aug 27 09:46 scsi-0QEMU_QEMU_HARDDISK_drive-scsi2 -> ../../sdc

With that information we can create a new pool:

Code:

root@pnm:~# zpool create backuppool mirror /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_drive-scsi1  /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_drive-scsi2

GUI:

“Disks” --> submenu “ZFS”; then the button “Create: ZFS” is visible --> click it and it opens the creation dialog
You’ll see a table with both disks listed; if there are no “Device” entries in the table you did something wrong ;-)

Fill in some data in the dialog:

Name: “backuppool” - or whatever you want
RAID Level: “Mirror” - or “Single Disk” if you have only one device ;-)
Add Storage: [ ] - leave it checked if you want to store something else here; for OUR goal this is NOT necessary
Device: select all drives

The resulting pool:

Code:

root@pnm:~# zpool list backuppool  -v
NAME                                       SIZE  ALLOC   FREE  CKPOINT  EXPANDSZ   FRAG    CAP  DEDUP    HEALTH  ALTROOT
backuppool                                31.5G   118K  31.5G        -         -     0%     0%  1.00x    ONLINE  -
  mirror-0                                31.5G   118K  31.5G        -         -     0%  0.00%      -    ONLINE
    scsi-0QEMU_QEMU_HARDDISK_drive-scsi1  32.0G      -      -        -         -      -      -      -    ONLINE
    scsi-0QEMU_QEMU_HARDDISK_drive-scsi2  32.0G      -      -        -         -      -      -      -    ONLINE

5.1 - if you have different sized disks...

... the GUI will tell you: “

command '/sbin/zpool create -o 'ashift=12' backuppool mirror /dev/disk/by-id/scsi-0QEMU\_QEMU\_HARDDISK\_drive-scsi1 /dev/disk/by-id/scsi-0QEMU\_QEMU\_HARDDISK\_drive-scsi2' failed: exit code 1

”.

The CLI fails also, but it tells you a way out:

Code:

root@pnm:~# zpool create backuppool mirror /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_drive-scsi1  /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_drive-scsi2
invalid vdev specification
use '-f' to override the following errors:
mirror contains devices of different sizes

Adding -f allows to step forward:

Code:

root@pnm:~# zpool create -f backuppool mirror /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_drive-scsi1  /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_drive-scsi2

root@pnm:~# zpool list backuppool -v
NAME                                       SIZE  ALLOC   FREE  CKPOINT  EXPANDSZ   FRAG    CAP  DEDUP    HEALTH  ALTROOT
backuppool                                31.5G   118K  31.5G        -         -     0%     0%  1.00x    ONLINE  -
  mirror-0                                31.5G   118K  31.5G        -         -     0%  0.00%      -    ONLINE
    scsi-0QEMU_QEMU_HARDDISK_drive-scsi1  32.0G      -      -        -         -      -      -      -    ONLINE
    scsi-0QEMU_QEMU_HARDDISK_drive-scsi2  40.0G      -      -        -         -      -      -      -    ONLINE

Note that I got "31.5G" net capacity, which is slightly below the size of the smaller disk. The difference of 8G is just not used in this case.

6 - create a dedicated data set

We could skip this. The previous step allows to create a directory inside the root-dataset of the new pool. It would be something like “/backuppool/myfolder”. But I really recommend to create a separate dataset as this allows to set independent parameters. (No, we are not actually doing that for now.) Different use cases inside of this single pool may want different settings. For this to be possible we separate “things” in datasets (or in ZVOLs if we have the need for a block device --> used for virtual disks of the VMs).

CLI:

zfs create backuppool/backupspace (#4)

Code:

root@pnm:~# zfs create backuppool/backupspace
root@pnm:~# zfs list -r backuppool
NAME                     USED  AVAIL  REFER  MOUNTPOINT
backuppool               564K  30.5G    96K  /backuppool
backuppool/backupspace    96K  30.5G    96K  /backuppool/backupspace

GUI:

not available... as far as I know

7 - tell PVE we have new storage

Until now PVE-the-software does not “know” what we have prepared and what we want to do with it. (At least if you did not check the checkbox from above.) We need to create a new “Directory”-Storage. We can store different things in it but we will focus only on “Backup”.
CLI:

pvesm add dir mybackupstorage --path /backuppool/backupspace --content backup

Example:

Code:

root@pnm:~# pvesm add dir mybackupstore --path /backuppool/backupspace  --content backup

GUI:

“Datacenter” --> “Storage” --> drop-down menu at the top “Add” --> “Directory”
"ID:" "mybackupstore"
"Directory:" "/backuppool/backupspace"
"Content:" "Backup" (DESELECT "Disk Image" for now)
"Nodes:" "All" - we have only one node, in a cluster choose this local node explicitly as the others do not have access to these local disks!
Tab "Backup Retentions:" skip for now

Result:

Code:

root@pnm:~# pvesm status
Name                 Type     Status           Total            Used       Available        %
local                 dir     active        29385344          123776        29261568    0.42%
local-zfs         zfspool     active        29571552          309968        29261584    1.05%
mybackupstore         dir     active        31997440             128        31997312    0.00%

8 - empty

Numbering ain't easy...

9 - create a first backup, manually

GUI:

select any VM or Container --> Backup --> notice the drop-down menu “Storage: local”, which is wrong! Select “mybackupstore”! Click “Backup now”.

This one is visible in the GUI immediately. You can cross-check in the CLI:

Code:

root@pnm:~# pvesm list mybackupstore
Volid                                                            Format  Type        Size VMID
mybackupstore:backup/vzdump-qemu-101-2025_08_27-14_22_36.vma.zst vma.zst backup    342187 101

Look at it directly on the file system:

Code:

root@pnm:~# ls -Al /backuppool/backupspace/dump/
total 340
-rw-r--r-- 1 root root   1129 Aug 27 14:22 vzdump-qemu-101-2025_08_27-14_22_36.log
-rw-r--r-- 1 root root 342187 Aug 27 14:22 vzdump-qemu-101-2025_08_27-14_22_36.vma.zst
-rw-r--r-- 1 root root      6 Aug 27 14:22 vzdump-qemu-101-2025_08_27-14_22_36.vma.zst.notes

10 - setup an automatic schedule

GUI:

"Datacenter" --> "Backup" --> "Add" ...

Addendum

Additional topics, not handled here

removable storage - those are not supported here
retention settings: per job and/or per storage (fall-back)
power consumption / power saving
external storage via NFS/CIFS/whatever
Proxmox-Backup-Server - THE recommended tool for backups ;-)

References

#1 - https://pve.proxmox.com/pve-docs/
#2 - https://pve.proxmox.com/
#3 - man sgdisk
#4 - man zfs-create

Disclaimer

As stated at the beginning: I am not responsible for what happens if you follow this text. This is by far not an exhaustive description. If you see something stupid (or errors) please reply.

Thanks

Thanks for reading

To find other articles like this search for “FabU” with “[x] Search titles only” = https://forum.proxmox.com/search/8543442/?q=FabU&c[title_only]=1
(( "FabU" has basically no meaning - I was just looking for a unique search term - and "Frequently answered by Udo" made sense - for me

))

Edited:

for some small corrections - this article is just too long to be 100% from the beginning
to add 5.1
extending step 7 - and stumbling about missing step 8...?

Search

Search

[TUTORIAL] FabU: I have two (or at least one) additional HDDs in spare - how can I use them to create Backups of my VMs and Containers?

UdoB

Distinguished Member

Target audience: Beginners

Precondition

Disclaimer

Notes

Hardware

Step-by-step

1 - document which drives you currently have

2 - connect both drives

3 - learn which drives you actually added

4 - prepare those disks

5 - create a new ZFS pool

5.1 - if you have different sized disks...

6 - create a dedicated data set

7 - tell PVE we have new storage

8 - empty

9 - create a first backup, manually

10 - setup an automatic schedule

Addendum

Additional topics, not handled here

References

Disclaimer

Thanks

We value your privacy

[TUTORIAL] FabU: I have two (or at least one) additional HDDs in spare - how can I use them to create Backups of my VMs and Containers?

UdoB

Distinguished Member

Target audience: Beginners​

Precondition​

Disclaimer​

Notes​

Hardware​

Step-by-step​

1 - document which drives you currently have​

2 - connect both drives​

3 - learn which drives you actually added​

4 - prepare those disks​

5 - create a new ZFS pool​

5.1 - if you have different sized disks...​

6 - create a dedicated data set​

7 - tell PVE we have new storage​

8 - empty​

9 - create a first backup, manually​

10 - setup an automatic schedule​

Addendum​

Additional topics, not handled here​

References​

Disclaimer​

Thanks​

We value your privacy

Target audience: Beginners

Precondition

Disclaimer

Notes

Hardware

Step-by-step

1 - document which drives you currently have

2 - connect both drives

3 - learn which drives you actually added

4 - prepare those disks

5 - create a new ZFS pool

5.1 - if you have different sized disks...

6 - create a dedicated data set

7 - tell PVE we have new storage

8 - empty

9 - create a first backup, manually

10 - setup an automatic schedule

Addendum

Additional topics, not handled here

References

Disclaimer

Thanks