Proxmox VE by HTTP VM Not running Issue

[ssaner]

I am using the Proxmox VE by HTTP template on Zabbix to monitor a Proxmox VE host. On this host there are several VMs that are intentionally not running. As expected, the discovery rules find these VMs and an alert is triggered to say that the VM is not running. I can then acknowledge that alert.

But then 24 hours later, I get a Resolved alert followed almost immediately by another Problem alert, thus clearing my acknowledgement of the problem. I have not been able to figure out what is causing this. Is this a core Zabbix behavior that I'm not aware of? Is it something in the template, like maybe the discovery rules that are causing this? Or could it be the Proxmox API that is indicating an artificial recovery of the VM?

Running Zabbix 7.4.0 (but same thing happened with earlier versions)
Proxmox 7.4-19 (yeah it is an old)
Proxmox VE by HTTP template is the 7.0 version

[Markku]

I'm using the Zabbix-provided template with Zabbix 7.0 and PVE 8.x (various versions), haven't had similar problem.

The problem expression seems to be last(xxx) <> "running" (here in my template at least). If your trigger resolves, it means that the item returns "running". Have you checked the item history that it is indeed the case?

Markku

[ssaner]

Thanks for that suggestion. Sure enough I have one pole that returned "running" rather than stopped. So that would probably point to the proxmox api returning a false value. I'll look further from that point of view.

[Anejey]

Thanks for that suggestion. Sure enough I have one pole that returned "running" rather than stopped. So that would probably point to the proxmox api returning a false value. I'll look further from that point of view.

Do you have backups configured for those VMs? During a backup job, the VM is temporarily spun up (not fully, just enough to read the disks), which will report to Zabbix as the VM running. I run into this quite often, not really sure how to solve it short of removing the VMs from the backup job.