Monitoring Resource Usage

[rtlarose]

Hello,

I have Proxmox installed on a remote dedicated server. Currently I am using it for personal/testing purposes. I have deployed VM's to friends to test and want to make sure that they aren't overutilizing allocated resources.

Is there a way to set set thresholds on VM's resources, i.e. (Memory, CPU, Diskspace) for the purpose of sending email alerts to me? This way I do not have to continually log in to check each VM.

I appreciate the input!

Regards,

Robert LaRose

 

[H4R0]

icinga2, zabbix, grafana, munin, there are many monitoring solutions that can send mails.

just set hard limits, every vm can be restricted for cpu, ram, disk io, nic bandwith.

 

[rtlarose]

H4R0,

Thank you for your response. I have the resources limited manually for the VM's. I was hoping to be able to set thresholds to trigger an email alert. I have configured zabbix before but find that it's use is better suited for clusters. It's not too big a deal... just wanted to see if there was a simple solution via the GUI. Still learning Proxmox...

 

[H4R0]

H4R0,

Thank you for your response. I have the resources limited manually for the VM's. I was hoping to be able to set thresholds to trigger an email alert. I have configured zabbix before but find that it's use is better suited for clusters. It's not too big a deal... just wanted to see if there was a simple solution via the GUI. Still learning Proxmox...

There is nothing built in to proxmox for alerts.

But proxmox supports export to influxdb and graphite databases.

https://pve.proxmox.com/wiki/External_Metric_Server

 

[dooferorg]

Pretty sad this still needs to be managed by an external program rather than something within Proxmox that will collect and show resource usages quite happily in the GUI - but only if you go look for it.

 

[UdoB]

Pretty sad this still needs to be managed by an external program

Is it?

Let PVE do its chosen job - it does this well! And use well established monitoring tools do their job too. Unix has had the philosophy "one problem --> one tool" / "do one thing, and do it well". And while this is not true for all tools nowadays (systemd + zfs don't obey) it stays valid. It is definitely better than reinventing the wheel again and gain - which just implements the well know NIH-syndrom (Not-Invented-Here), some large companies suffer from.

Influxdb/Grafana, Zabbix, Prometheus, checkmk, LibreNMS, and so on and so on. Look at them - you'll find something that does the job better than a quickly integrated "only PVE" monitoring tool.

Just my two €¢ on a Sunday evening...

 

[dooferorg]

I see what you're saying about the unix philosophy. Given how much Proxmox team likes systemd though I'm not sure they hold to that. Lol.

I just felt like Proxmox has the information needed (stats) and capability (send an email) already right there.