PBS as an archive: possible to restore a "gone" VM?

[greg]

Greetings

I'm a long time user of Proxmox but a recent user of PBS and I love it.
I was wondering: I have VMs which are off but I keep for archiving. Would it be possible to backup them, and delete the VM from the node? would I be able to restore them one day if needed?

Thanks in advance

Regards

GR

 

[UdoB]

...and delete the VM from the node? would I be able to restore them one day if needed?

Sure!

The whole point of a well-thought backup system is to be able to restore some "thing" when the original data is damaged/destroyed/stolen.

Every node which is connected to that PBS can see all backups made in the cluster. And via <node> --> <pbs storage> --> Backups --> "Restore" you can restore any VM/Container

The actual number of available backups of a specific VM will get lower during time - see "pruning" & "garbage collection", but NEVER will ALL of them get deleted. Execpt of course if this is done manually by the system administrator = you ;-)

That's my understanding. You may simply try it out with a test-vm under your own circumstances...

There is only one pitfall: the ID (e.g. 100) of the then vanished VM will get used again by the automatic algorithm. I for myself assign and document IDs manually. You may set a new "Next free VMID range" unter Datacenter --> Options to avoid collisions.

Have fun!

 

[greg]

Thanks for your answer.

I've been running a cluster for years, so I'm aware of the various pitfalls, including the IDs.
My concern was more about the VM/CT config. I noticed you cannot restore a backup from the backup server, you have to do it from the host, that means from a configured VM/CT. I was wondering if I had to keep that config manually (before deleting the VM/CT), or is it a cleaner way to keep that information for the purpose of a possible restore?

Thanks in advance

 

[UdoB]

you have to do it from the host, that means from a configured VM/CT

Maybe this is your pitfall: it is not required to have a corresponding VM/CT in beforehand.

For a restore go to the Node/Storage. That is select the storage in the server view in the tree in the lower area. "Server View --> Node --> lower area of the tree".

The right large body pane will show you ALL backups in that Storage, not only backups of VMs/CTs "living" on this node. You can select any of them and restore it. You can also assign any valid (conflict free) PMID in this step.

Of course the configuration is read from the backup. Your Node needs to be compatible. For example the relevant networks should be present. Other details can be modified on the fly, e.g. the actual datastore to restore to.

Best regards

 

[greg]

Thanks a lot for taking the time to explain. I understand it better now: you restore from the host side, but not necessarily the original host. it's really cool.
Does the backup includes all disks from the original CT/VM? I've seen the unused disks are skipped.

 

[UdoB]

Does the backup includes all disks from the original CT/VM?

Every virtual disk hast a clickable checkbox "Backup [x]" which allows you to choose...


Edit: be careful on Restore - you might loose the content of the not-backuped disk!