[SOLVED] Official way to backup proxmox VE itself?

[promoxer]

Is there any official guide on how to backup proxmox VE itself?

My main concern is the amount of work that I have put in to make the current setup work, e.g. GPU passthrough etc.
If that breaks due to updates etc, I would like to have a way to quickly revert back to the last known working state.

Is Proxmox backup server able to do it? and is it the only way?
Or is there some way that is built-in to Proxmox VE itself?

 

[Dunuin]

Is there any official guide on how to backup proxmox itself?

Nothing official except for restoring the config.db: https://pve.proxmox.com/wiki/Proxmox_Cluster_File_System_(pmxcfs)#_recovery

Is Proxmox backup server able to do it?

Not yet, but it's on the roadmap.

My main concern is the amount of work that I have put in to make the current setup work, e.g. GPU passthrough etc.

If that breaks due to updates etc, I would like to have a way to quickly revert back to the last known working state.

I would backup the /etc folder so you got a backup of the config files. And for a complete backup you could boot into a clonezilla pen drive and create an image of the whole system disk on block level.

 

[promoxer]

I would backup the /etc folder so you got a backup of the config files. And for a complete backup you could boot into clonezilla pen drive and create an image of the whole system disk on block level.

Which brings up a good point... my PVE is installed on 1 SSD together with the local and local-zfs... I'm not familiar with clonezilla, but it would seem like I have to be cloning the whole 2TB SSD (!!)

However, during the installation of PVE, it did not seem like I have much options. Any pointers on handling this?

Thank you.

 

[Dunuin]

Which brings up a good point... my PVE is installed on 1 SSD together with the local and local-zfs... I'm not familiar with clonezilla, but it would seem like I have to be cloning the whole 2TB SSD (!!)

Jup, ZFS uses a single partition and you need to backup on block level, so the whole 2TB partition. Thats why I personally got a 100GB SSD mirror just for the PVE system, so I don't have to backup all those VMs/LXCs too, with the VMs/LXCs on another pool using other SSDs. With a new installation you could create two ZFS pools using two partitions on the same SSD, but that won'T help you now with an existing system.

 

[promoxer]

Hmm, actually it might, I just want to restore PVE to as-was. My main and biggest concern is GPU passthrough failing to work due to any reason.

I was hoping there is an easier way though. Clonezilla means that I have to do this manually (and often enough) to get a recent snapshot of my PVE.

Maybe I should just stay away from doing updates

 

[Dunuin]

I was hoping there is an easier way though. Clonezilla means that I have to do this manually (and often enough) to get a recent snapshot of my PVE.

I combine block level and file level backups. First I restore the older but complete block-level backup. Then I restore the backups of my more recent PVE config files on top of that.

Maybe I should just stay away from doing updates

And running an insecure system just waiting to be hacked. And even if you don't update, your SSD will fail sooner or later and you will need to start from scratch without a proper backup.

 

[LnxBil]

Hmm, actually it might, I just want to restore PVE to as-was. My main and biggest concern is GPU passthrough failing to work due to any reason.
[...]
Maybe I should just stay away from doing updates

Yes, those two things have a close relation. I also lost my passthrough and I dumped it completely. It was not worth the hassle and my time.

Which brings up a good point... my PVE is installed on 1 SSD together with the local and local-zfs...

Easiest way would be to have just one ZFS pool so that you only have to backup the one pool just with internal send/receive. I only do this for non-clustered machines. You can have snapshots so a rollbacks are easier (with some extent), but you do have consistent backups if you sync them otherwise. You can also just upload them encrypted to the internet and store them in a remote ZFS pool.

 

[promoxer]

Easiest way would be to have just one ZFS pool so that you only have to backup the one pool just with internal send/receive. I only do this for non-clustered machines. You can have snapshots so a rollbacks are easier (with some extent), but you do have consistent backups if you sync them otherwise. You can also just upload them encrypted to the internet and store them in a remote ZFS pool.

I'm not worried about the VMs, I have a pretty comprehensive automated way to deal with them.

My concern here is PVE itself. What can I do (apart from approaches similar to clonzilla) to turn a freshly installed PVE back to what it was, with GPU passthrough working etc.

 

[Dunuin]

PCI passthrough is not part of the PVE config files, so not sure if that even would be backupable, even when the Proxmox team actually adds PVE host backup feature from the PBS roadmap. Maybe if they add root-filesystem snapshots, but then still something could go wrong that you lose your whole pool/LV/dataset/partition where snapshots won't help as they would be lost too.

Some people here use Veeam to back up the PVE host, but not sure how that will work. If they just backup the config files, the disk on block level or if it is making use of some more complex stuff like creating backups based on snapshots. I think I've read something about snapshots.

 

[patch]

Is there any official guide on how to backup proxmox VE itself?

I suggest starting with this thread https://forum.proxmox.com/threads/how-to-backup-proxmox-configuration-files.67789/
Particularly posts #27 and #32

 

[spirit]

backup:

/var/lib/pve-cluster/config.db
/etc/corosync/*
/etc/network/interfaces
/etc/hostname
/etc/hosts

 

[LnxBil]

I'm not worried about the VMs, I have a pretty comprehensive automated way to deal with them.

My concern here is PVE itself. What can I do (apart from approaches similar to clonzilla) to turn a freshly installed PVE back to what it was, with GPU passthrough working etc.

You don't need VM backup if you have ZFS send/receive to another host, that was my whole point.

 

[promoxer]

You don't need VM backup if you have ZFS send/receive to another host, that was my whole point.

I do not plan to have another host.

 

[promoxer]

By chance, this video "How I upgraded and reverted Proxmox from 6 to 7 to 6 with UEFI boot and GRUB" popped up on my youtube feed.

He had exactly the same issue: loss of GPU passthrough. But his steps are still technically beyond my understanding and not directly addressing my backup concern. But I think the idea is to keep the PVE installation small and separate.

Ideally, I can just backup the settings and reinstall PVE each time I need to restore. However, I think my best option now is to keep PVE small and separate, backed up by clonzilla.

Thank you all.

 

[leesteken]

He had exactly the same issue: loss of GPU passthrough. But his steps are still technically beyond my understanding and not directly addressing my backup concern. But I think the idea is to keep the PVE installation small and separate.

If the passthrough problem caused by kernel 5.15 (or later), then this work-around might solve it. Apologies if this is off-topic for this thread.

 

[LnxBil]

I do not plan to have another host.

... and where is your backup then? If not on another host, it is per definition not a backup.

 

[promoxer]

... and where is your backup then? If not on another host, it is per definition not a backup.

What meant was, I don't plan to have another PC sort of machine that requires OS maintenance.

I backup via samba to a NAS (that I do not believe can run ZFS).

 

[promoxer]

Nothing official except for restoring the config.db: https://pve.proxmox.com/wiki/Proxmox_Cluster_File_System_(pmxcfs)#_recovery


Not yet, but it's on the roadmap.


I would backup the /etc folder so you got a backup of the config files. And for a complete backup you could boot into a clonezilla pen drive and create an image of the whole system disk on block level.

Do you know if clonezilla recognizes ZFS and only copies the used blocks?

 

[yjjoe]

Using btrfs snapshots is already a way to go (even tho not fully supported yet)

 

[aanjaneyam]

Hi. I just installed proxmox for the first time. The first thing after installing proxmox that came to my mind is "How to backup proxmox VE itself?". Off course I did some research before deciding if proxmox is right for me. I am not not aware of zfs too as this is the first time I would be using it even though I have been using Linux mostly in desktop environment, with some server level experience. I went though zfs(8) man and zpool(1M) man pages. The zfs(8) man pages states the following:

ZFS File System Hierarchy

A ZFS storage pool is a logical collection of devices that provide space for datasets. A storage pool is also the root of the ZFS file system hierarchy.
The root of the pool can be accessed as a file system, such as mounting and unmounting, taking snapshots, and setting properties. The physical storage characteristics, however, are managed by the zpool(1M) command.

Isn't is possible to just create a snapshot of the whole rpool (pve installation) and use zfs send and receive to to backup to another destination?