The weird thing is that proxmox itself runs very well, but pfSense randomly freezes/reboots.
It would be great to get some more ideas on how to get this thing stabilized.
Im pretty much in the same boat lately. I get a random crash that has no useful logs in Proxmox or OPNsense. Internet goes down until I log in and restart the VM.
So I picked up the Topton N5105 V3 in May and I seem to be in the same boat except I am running Unraid (6.10.3 Linux kernel 5.15.46) with pfsense (2.6.0 FreeBSD 12.3-STABLE) in a VM and 3 of the NICs passed through.
I previously had the exact same setup but running on an Intel i3-11400 with a 4 port i211 card without issue.
Since moving to the N5105 I have had 4 random kernel panics from the pfsense VM (over 4 weeks) but Unraid server remains unaffected.
Each time the VM has rebooted cleanly by itself. I only noticed because pfsense sends me an email letting me know it has rebooted.
I have tried all the obvious stuff;
- Temps: nothing is getting hot but to be sure I added heatsinks to the NVME drives and a fan to the chassis.
- Switched to a good quality PSU.
- Connected it to a sine wave UPS.
- Run memtest without issues.
After the 3rd time I decided that I need to actually look into this, and setup syslog on both Unraid and the VM. Took a look at the
pfsense kernel panic guide and started to copy the dump files.
The most recent (4th) panic was "
Panic String: spin locks can only use msleep_spin". Every time has been a different message.
Of course now that I am actually paying attention, it hasn't had an issue since (2 weeks).
If I ignore the hardware change then the only other thing I can think of, is that the original was setup a year ago and was running on QEMU Q35-5.1. The new build is running QEMU Q35-6.2. So it is possible that FreeBSD is having issue with the VM.
Given this is happening to a number of people and seems to only effect *nix OSs this feels more like a low-level hardware/kernel incompatibility rather than a bunch of near identical hardware faults.
Anyway I am just going to keep an eye on things for now. If it comes back I will downgrade the QEMU VM emulation as my next step.