Backup of PVE config

ns33

New Member
Apr 4, 2024
8
0
1
I'm sure this was asked in the past and maybe missing it within the webui for pbs, but is there a way to backup the configs for each pve node directly from pbs? I can't seem to find a way within pbs or even pve to just capture the host config. Doing a vm/ct is straightforward but what about the node itself?

I can use rescuezilla or something similar to clone the boot drives but is there a way for pbs to do a capture of config files?
 
I run ZFS on all my Proxmox VE and the a cron-rsync-job every 15 min to rsync /etc/pve to a zfs dataset with zfs-snap-shot enable.
So i get with every 15 min a entry into that dataset.
 
I love how you found an old post where the staff actually admit that there's no real backup solution for the hosts themselves.

I don't back up PVE hosts. If they die I burn em down and spin up a new one. I customize the host as little as possible, and treat them like an appliance, replaceable.
 
  • Like
Reactions: joishw
I was searching for the same, and ended up with this for stand-alone-hosts:
- seperated disks for PVE-OS and virtual machines
- a manual backup of /etc/pve/, here you can find all the relevant config stuff.
- like tcabernoch mentioned, if having trouble, you can "burn it down" and reinstall the pve on the os-drive(s) and then re-attach the machines.

Did some tests recently, worked like a charm ....

see : https://forum.proxmox.com/threads/p...nstallation-os-repair-right-way-to-go.166455/
 
So I read the pbs documentation more and there looks to be a cli way to do a host level backup. This feature really should be within the gui. Maybe I missed it cause I clearly missed this section in the documentation

From the cli within pve, run the following:
proxmox-backup-client backup root.pxar:/ --repository <ip_of_pbs>:<pbs_datastore_name> --backup-type host --include-dev /etc/pve

You need to force include /etc/pve because by default it skips it. Now I have no idea how this will play out during a restore but from pbs you can download a compressed dir of the /etc/pve, and ideally copy that over to a fresh pve instance and be back up and running.

This is acceptable in my book.
 
I love how you found an old post where the staff actually admit that there's no real backup solution for the hosts themselves.

I don't back up PVE hosts. If they die I burn em down and spin up a new one. I customize the host as little as possible, and treat them like an appliance, replaceable.
I'm not sure what I'm in for on the cyber side, but I can see mods being made to stig each node. Keeping those changes backed up and in a state where redeploying is easy is ideal. Of course, only apply what doesn't break anything.