Kernel Panic: mbuf_cluster kern.ipc.nmbclusters limit reached

yctn

It seems that the amount of mbuf clusters keeps on rising and rising.

https://puu.sh/KILvE/cbaa51d1fb.png

After a week, I'm getting closer to 1M and it will result in a kernel panic. What can I do to stop this mbuf leak?

I am running:

2.7.2-RELEASE (amd64)
built on Wed Dec 6 20:10:00 UTC 2023

inside Proxmox 8.1.10

4 GB of RAM
4 cores
Virtio network interfaces (I already tried switching them to Intel E1000, but that resulted in the same behavior)

Any advice is welcome. Thank you in advance.

chrcoluk

I got no idea on this, I have never had a workload that increases mbuf above very low levels,

But for sure switch away from E1000 emulation, go back to virtio.

SteveITS

@yctn me neither, but they mention it some at https://docs.netgate.com/pfsense/en/latest/hardware/tune.html#mbuf-exhaustion

2.7.2 is over 2 years old, why not try a later version?