Following applying the 2020-05 Cumulative Update for Windows Server 2016 (1607), a PFSense 2.4.5 Hyper-V virtual machine on this host OS suddenly became very unstable.
The machine appears to experience high load averages immediately following boot, with a large percentage of the vCPU usage (>99%) being shown as “system” in top.
Additionally, sometimes it is seen that the pfctl process is also using close to 100% of a CPU. The system is unresponsive, or intermittently responsive throughout this time — both in terms of the console, the web UI, and packet forwarding that the machine should be doing.
After a few minutes, Hyper-V appears to restart the guest OS. This isn’t a PFSense kernel panic and no dump is made by PFSense.
I spun up a new VM of the unstable daily snapshot of PFSense, a Generation 2 UEFI-based VM. This completely eliminated these problems for us in this new VM — it works fine, routes just fine for what we’re doing and is responsive.
I don’t have a solution — I just want to throw this issue out there in case others have experienced it, in case we might work together to figure out what’s going on here.