Hi @stephenw10, Thank you!
You're correct regarding using bridges with HA configuration.
As sample, below is the scenario that I have tested.
[image: 1623309908498-screen-shot-2021-06-10-at-15.46.59.png]
As shown the switches SW3 and SW4 aren't interconnected to avoid loops.
For redundancy I use a combination of LACP in failover mode and VRRP IP as default gateway instead of use pfSense CARP configuration.
The reason to enable HA is to have the pfSense Sessions, Alias and Rules synchronized.
For the servers behind, as shown WB1, there's two connectivity ways;
Master to SW3 (SW4 as Backup) WB1
Master to SW4 (SW3 as Backup) * WB2
Image
It means, both pfSense can handle traffic simultaneously. Although being configured as HA Master / Backup they work as Active / Active. Did you had some experience like that before?
Regarding the hight CPU interrupt time % issue.
After change the parameters below, the performance looks better then before. I still monitoring it.
System Tunables
net.link.bridge.pfil_bridge = 0 to 1
net.link.bridge.pfil_member = 1 to 0
MY