I am running 1.3.4 LTS since last 2 weeks in proxmox with 4 vCPU and 6GB RAM. Ever since I installed it, the memory utilization of the vyos keeps increasing with no significant changes in the network Please look at the Grafana chart. What can be the probable cause??
How does “free” and “cat /proc/meminfo” change over time?
Also do you have any increase in conntrack like “conntrack -L”?
A dump like every 15min or so of “ps auxwww” would probably be handy aswell to see if its kernel, buffers or a specific application that increases in memory over time.
I checked the trend for conntrack and I see the average entries are on rise day by day. I will try to dump the ‘ps auxwww’
Linux kernels memory utilization is not that simplistic. The kernel will usually not start page swapping and reclaiming which is an expensive and destructive operation until it’s short on free pages. This means that in most cases the kernel will continue to acquire additional memory as it’s free (you see it on your VM manager as being gradually claimed, but the kernel looks at it as free RAM). Only when the acquired memory reaches a specific minimum (called “low watermark”) the kernel will wake a special kernel level daemon called kswapd that will start scanning the memory and reclaiming back unused parts of the memory to be re-used by the kernel. Monitoring memory leaks and over consumption of RAM at the VM management level is pretty useless and does not indicate anything. To truly monitor memory bottlenecks you need to install an agent in the guest that will monitor memory activity inside the OS (as @Apachez indicated by using native Linux commands within the OS) and not outside the VM black box.
