Rancher cluster behind A10 DSR VIP

I have an A10 hardware load balancer in my environment that I must use. I’d like to set up multiple VIPs that point to each of my rancher agent nodes, and let the rancher load balancer send the traffic to the appropriate containers.

When I set up a rancher load balancer to run on each of my agents with a SourceIP/port of 80, and a default target port of 8080, I can reach my container just fine. But each VIP ends up pointing to the same container.

It looks like I should set up the rancher load balancer SourceIP/port to :80. This way traffic would be sent to the appropriate container based on the vipaddress.

But when I do that, the requests time out. Also, looking at the haproxy.cfg that it generates, the frontend bind is still *:80, which I think is incorrect.

So my question is, how do I configure the rancher load balancer to route differently based on IP address.

The problem was related to the A10 load balancer’s health check looking for the service to exist on the floating ip, and the actual traffic would arrive on the VIP ip. Setting up two different rancher load balancers was the solution to my problem.