Solved: Weird Packetloss

Router - Switch - VM. Two different public subnets on the same hardware and cables. From router to VM, subnet 1 has packetloss, while subnet 2 does not.

Both subnets are on the same interface.
Same thing happens to multiple VMs on multiple independent environments as well as some physical machines.
I can change the IP from the “good” subnet to the “bad” subnet and back again… packetloss only occurs on the “bad” subnet.
I disable all queues, filters, NATs and mangles, problem persists.
I tried different IPs within the “bad” subnet and the test IP I’m not using at the time doesn’t respond to pings, nor show up in MT’s ARP table.
I disabled the gateway IP on the “good” subnet… no dice.

Captured from the perspective of the router: Router → VM, all requests in the packet capture were replied to, so apparently 24% of the requests never even leave the router. VM → Router 100% of requests hit the router, but the router didn’t respond to 70% of them.

I use 100 pings as my test. If I get a low number, I test again incase it was a fluke, either direction.

I found my answer to the packetloss problem! From Forest Christian (PacketFlux), “Sounds like the type of thing which happens when the router thinks it has multiple paths to the segment. Is the segment with packet loss by chance in the router twice?”. I had an IP in that subnet assigned to a similarly spelled interface.