New PPPoE module (if_pppoe) causes high cpu usage and cause lagging

[tlex]

Current setup :

PfSense CE 2.8.0 (ufs install) virtualized on Proxmox.
No device passthrough used.
Host is a I5-8500T, 96gb of ram, nvme disks and with dual X710 10gbps SFP card.
VM has 4 cores / 8gb ram and 3 virtio 10gbps interfaces.
mtu is set to 9000 on the lan interface and left to default for the wan.

My isp is Bell Canada (3gbps) with a modem bypass using a XGS-PON SFP+ from ECI

Now, the vm cpu usage used to be between 2% and 5% before I tried to enable if_pppoe and now it stays steady at 20% without any major network load. I used to go up to 20-30% under more intense network activity but now this goes up to 70-90%.
I know this isn't scientific but browsing is lagging, accessing PfSense webui is also pretty laggy some times (situation that I never had before). So I guess I must have something set wrong but I can't figure... Looking at the regular logs, I don't see anything different or evident.

This returns nothing under a speedtest / load test:

dtrace -n 'fbt::if_sendq_enqueue:return / arg1 != 0 / { stack(); printf("=> %d", arg1); }'

I do have the following tuning set :

net.link.ifqmaxlen="2048"
net.isr.defaultqlimit="2048"
net.isr.dispatch="deferred"
net.isr.maxthreads="-1"
net.isr.bindthreads="1"
legal.intel_ipw.license_ack="1"
legal.intel_iwi.license_ack="1"
machdep.hyperthreading_allowed="0"

running Speedtest from the vm itself i can see the error count increase drastically in the interface statistics :

[tlex]

I found my issue !
I used to have an old VIP attached to the WAN interface.
It wasn't making any issue when using MPD but ended up creating problems with IF_PPPOE.
VIP removed, all good now :)

[mr_nets]

Interesting, does it means that the new PPPoE driver is not HA compatible, your VIP was CARP or ALIAS ?

[tlex]

@mr_nets Alias