Since a couple updates (or perhaps a few) my OPNsense server (proxmox vm) memory keeps going to 90+ percent. Previously it never got very high at all. This last patch it seemed ok for a couple days but I logged in today via web gui and it was over 90% again.
I only use Unbound DNS, Kea DHCP (for Cisco switch DHCP relay), QEMU guest agent, and Insight to see what is eating bandwidth. How can I track this down?
I had similar a few 25.7 revs back, but now running the latest, seems ok.
I think I notice fastest throughput when I was running 25.7.3_7
Valid and recurring question. Aside from times when an OPN change requires a large increase in memory use, which is very rare, essentially comes to the fact that the qemu guest agent for freebsd is very limited. Note this is outside the control of opnsense. Any ballooning is unreported back to the host.
In short, all the memory use and freeing from the VM is unreported to the host by the agent.
You should be able to check differences by using top or htop IN the VM and compare the memory use with what is reported by Proxmox.
If you're referring to the memory use reported by Proxmox and you've recently upgraded to PVE9, this may be due to the change in the way PVE9 gathers memory stats vs what the FreeBSD qemu-agent provides.
VM Memory Consumption Shown is Higher (https://pve.proxmox.com/wiki/Upgrade_from_8_to_9#VM_Memory_Consumption_Shown_is_Higher)
I've seen this after upgrading to PVE9 and now rely on telegraf stats to monitor opnsense memory use.
Suricata looks strange to me. I upgraded to 25.7.5 couple days ago, rebooted. Today suricata showed 11G size and 6G res (lobby mem widget showed 9% ram used). I restart Suricata and it comes back as 6.2G size and 3.99 res, now the lobby mem widget shows 24% used. In top suricata res keeps growing.
The widget showing 9% when Suricata was 6G res makes no sense to me, I only have 16G of ram. 6 of 16 is 37.5%, so something in that widget seems off.
Right now top tells me I have 8921M free mem. That's near 50% used, widget says 3827M used.
So something is not right in Lobby Dashboard stuff, I tend to believe the OS tools.
Disk widget seems right, shows me 1%, which matches df, zroot/ROOT/default 1%