Proxmox VE shows high memory usage, how do I change the threshold?

[albatrosdk]

Hi,
I'm using the Proxmox VE by HTTP template, but I have an issue with zabbix complaining about the vms using too much memory.

But I can't figure out how to increase the threshold from 90% to 95%.
I changed it in the template's 'Macros', but it doesn't seem to have any affect on the already discovered vms. I've also tried re-running the discovery. but that hasn't changed anything either.

Any advice is much appreciated

/tony

[albatrosdk]

And it seems it just took some time for it to go through. The warnings are gone now, so setting the threshold in the template -> macros actually worked..

[albatrosdk]

Ok, I got it sorted I think. I unlinked and cleared the proxmox ve template from the hypervisor end re-added it.

Nope, that didn't help..

... still an issue

[albatrosdk]

And it seems it just took some time for it to go through. The warnings are gone now, so setting the threshold in the template -> macros actually worked..

[erikl_]

I had the exact same issue: immediately after adding the Proxmox monitoring (great template by the way!), the majority of my VM's supposedly were running on high memory consumption. I don't know why it gets reported this way, most of those VM's are also monitored from the inside with an Agent2 and it reports much healthier memory usage actually.

Anyway, my question was: would setting this macro on the host level actualy propagate through the already discovered items, AND resolve the problems that had already been created? Answer: yes. Yes, it does, and in my case it took less than 5 minutes.

I'm running 8.0.0rc2 so maybe that helps.

[erikl_]

By the way, just so posterity knows what we're talking about: go to the Host definition, tab Macros, select Host macros (this is where your {$PVE.TOKEN.ID} and .SECRET are defined), and add {$PVE.TRIGGER.MEMORY.WARNING}
It's set to 90 by default (representing 90% mem usage), and I set it to 98.

[irontmp]

I had the exact same issue: immediately after adding the Proxmox monitoring (great template, by the way!), the majority of my VMs supposedly were running on high memory consumption. I don't know why it gets reported this way. Most of those VMs are also monitored from the inside with an Agent2, and it reports much healthier memory usage actually. It feels like a ruleta aleatoria when comparing the memory readings between Proxmox and Agent2 without knowing which metrics are being used.

Anyway, my question was: would setting this macro on the host level actualy propagate through the already discovered items, AND resolve the problems that had already been created? Answer: yes. Yes, it does, and in my case it took less than 5 minutes.

I'm running 8.0.0rc2 so maybe that helps.

Have you checked whether the trigger threshold is actually controlled by a macro at the VM level or inherited from the template? In Zabbix, changing a template macro should normally update linked hosts, but if the trigger uses a hardcoded value or a different context-specific macro, the change might not take effect. I’d check the trigger expression for the memory usage alert and verify which macro it references. Also, check the discovered VM hosts to see whether a host-level macro is overriding the template value. That should help narrow down why the threshold is still at 90%.