Processor counts drop to zero
-
We're using PFSense 2.4, but also seeing this on other 2.3.4 versions as well. The issue is that the processes periodically drop out to 0 from a total of ~120.
This is a VM running on Hyper-V 2012 R2 on a Dell R610 with Intel e5520 procs, a built in 4 port Broadcom NIC (used for PFSense LAN interface and management) and an add-in Intel Dual port card (used for PFSense WAN).
Each processor is allocated 8 virtual CPUs. We have a few VMs running on a couple of identically configured servers, and we're seeing this behavior on all of them. We are not seeing this on other systems running on Dell R410s, R420s. One difference is that we are using NIC teaming on the setup having the issues, but I'm not sure how that would affect system processes.
We've turned off Virtual Machine Queues on the physical NICs, as we've seen issues with Broadcom cards randomly losing connectivity and requiring a disable and re-enable of the physical NIC on the VM Host.
We have 1 package installed: mtr-nox11
-
In case anybody runs into this issue, it was caused by having the time sync services turned on in Hyper-V. We disabled the time sync services offered by Hyper-V (in the Hyper-V manager) and the issue went away.