Feature request: Proxmox exclusions by TAG or name regex

[oernii]

Hi,

As part of our migration from VMware to Proxmox, we’re facing a challenge with VM backup exclusions. In VMware, we used a NOBACKUP folder that Veeam automatically ignored. However, in Proxmox, we must manually specify which VMs to exclude from backups.

This is particularly problematic for test and dev VMs, which are frequently created by teams without access to Veeam job configurations. As a result, our backup storage fills up quickly before we can intervene. Additionally, since Proxmox uses a per-VM licensing model, this can also lead to unnecessary licensing costs.

Do you have any recommendations for managing this more efficiently?


Please add TAG based and name/regex backup exclusions to Proxmox.


Thank you.

[rovshan.pashayev]

Hi,

Have you tried to put those dev VMs into the resource pool and then exclude it?
If you added VM into the Backup job, and same VM is also added into the resource pool, which is excluded, then Veeam will not process that VM.

[oernii]

Thank you, this seems to solve my issue.

[oskarb]

Hi,

Have you tried to put those dev VMs into the resource pool and then exclude it?

This is not a satisfactory solution, because resource pools are exclusive in Proxmox. A VM can be part of at most 1 resource pool.

This is unlike both vmware and proxmox *tags*, which can be combined in different ways.

To let a job backup only a subset of VMs (or all VMs but use different guest credentials for different machines), Veeam provides only these options for Proxmox: Cluster, physical node, VM, or Resource pool.

USE CASE
Backup all VMs, using AAIP for Windows guests but not for Linux guests.

Veeam for VMware job 1 (with AAIP): Include the vmware "datacenter". Exclude machines having tag "backup-no-AAIP".
Veeam for VMware job 2 (without AAIP): Include the vmware tag "backup-no-AAIP".
Result: Automatically discovers all new machines in either job. If a new machine needed the tag but it was forgotten, job 1 will raise an alert so it can be fixed. The tag does not prevent use of other tags on the same vm. There is no risk of accidentally not backing up a VM because 1 job is guaranteed to pick it up.

Try to do the same for Proxmox:
In job 1, include the proxmox cluster.

Now we can exclude:
1) the cluster - pointless
2) physical node - useless because VMs move around
3) individual VMs - technically possible to exclude the VMs that should not have AAIP, but requires Veeam job reconfiguration every time a new VM is created. Doesn't work in practice.
3) resource pool - on the surface similar to tags, but if we have a resource pool veeam-no-aaip, then resource pools cannot be used for any other purpose for those VM.
Result: Pain and suffering.

Now with Veeam 13.1 there is also the ability to set different credentials "per guest" in the same job but then we get the same issues there. Imagine having an AD account that Veeam uses on all the Windows machines. Then for Linux machine we could create a local account of the same name and password in each guest and let Veeam use that. But to tell Veeam to use those Linux credentials for Linux machines, we either need to set the exception *per VM* which means reconfiguring Veeam every week, or we can have a resource pool like "LinuxVMs" and refer to that in Veeam, but again, then we can't use any other resource pools on those machines.

This feels like a quite significant oversight in the Veeam for proxmox design. Please implement support for include/exclude/override by tag, and/or if there is some other possible mechanism other than resource pools.

[HenrikS.]

I second this RFE, managing inclusions, exclusions and credentials based on tags or any other solutions is most important. Resource group is not adequate as a VM cannot be in more than one.
It is a severe lack of functionality regarding manageability and a deal breaker when comparing Veeam to other solutions for backup of PVE.