Bug? SNMP-polling shows 120Gbyte ram used, Proxmox shows 18.3.

[jorel83]

Hi peeps.

Been using Proxmox in a lab environment for about a year and recently transitioned from vmware into a proxmox for production use. So last week I finally enabled off site backup and something curious happend. I use Librenms for monitoring and it reports that the same time as the backup runs ram gets eaten and never frees up again. Proxmox gui shows nothing a like.

Htop shows same as proxmox gui 18.3 Gbyte.

The there are most vm:s on the proxmox host showing this snmp value, im looking in to the Librenms side of this as well. It is clearly related to the backup process happend the time the backup is scheduled, but seemingly it seems not to be a real issue only snmp-related.

Any clues?

Br/Joel

 

[Alwin]

The graph doesn't take any details for RAM usage. In Linux the conception of RAM is, free RAM is wasted RAM. Therefor cache and buffers are not counted to the usage, as they are freed as soon as the RAM is needed.
There was a discussion around a Linux RAM usage: https://github.com/librenms/librenms/issues/3179

 

[jorel83]

The graph doesn't take any details for RAM usage. In Linux the conception of RAM is, free RAM is wasted RAM. Therefor cache and buffers are not counted to the usage, as they are freed as soon as the RAM is needed.
There was a discussion around a Linux RAM usage: https://github.com/librenms/librenms/issues/3179

Indeed so it is. But until the backups was scheduled non of my server (have many) have shown this behaviour, snmp usually reports same as htop, and that is why I wrote this post.

But I agree, its clearly related to Librenms, not even Observium thats Librenms is based on have this issue on the servers Observium monitor at my day job.

Thanks for you post, Alwin