Feature Request / Bug report / Workaround virtio-rng

Hello all,

First time post, I am running TrueNas Core 13.3 and recently created an Ubuntu 22.04.5 VM which is being used as a Veeam Repository running the latest transport version 12.2.0.334 for my company’s backups. I kept running into issues since provisioning in which I would receive in DMESG and on the console CPU Soft locks on all 16 vCPUs provisioned within minutes of the backups being sent over to the VM for storage. At first, I believed DMESG and went down the rabbit-hole of CPU contention however the host machine running TrueNAS is not heavily used hence why I selected this particular host. Some background of my configuration of the VM:

3 Virtual Disks
Boot disk /dev/vdc 50GB virtio formatted as EXT4 supported by ZFS
Internal Array /dev/vda (Dell Perc) 15TB formatted as XFS (Veeam requirement) supported by ZFS
External Array /dev/vdb (Dell MD1400/Perc) 25TB formatted as XFS supported by ZFS
Serial console enabled
VNC Disabled
16vCPU
16 GB RAM

My workaround was as follows to stop the issue

In /etc/sysctl.conf
kernel.random.write_wakeup_threshold=1500
sysctl kernel.task_delayacct=1

install rng-tools and enable the daemon at boot
Configured /etc/default/rng-tools-debian
HRNGDEVICE=/dev/urandom

Rebooted VM and tests OK

I do see that bhyve supports the virtio-rng device however, it is not available to be added in the GUI please add this feature editing the VM’s xml using virsh gets overwritten almost immediately.

Thank you,
Dan