It's a locally managed pair of gateways.
The DHCP issue is probably not cluster-related.
It was serious enough that the webui was dead and I was unable to SSH into the box when this happened.
Also, HA still did not see anything wrong with this, so no failover happened until I shut the ports down from the switch.
Happened on both units, which were bought separately, always on the active one, so definitely not a hardware issue.
If I get the opportunity to look into this again while onsite, I guess I'll try plugging into the serial console.