Backend maxconn config not working properly when running multiple ingress controllers

[m4nu56]

Hi all,
I have an HAProxy ingress controller set up as a Load Balancer on my k8s cluster.
Deployed using haproxytech/helm-charts and using the latest version 1.35.3 with haproxytech/kubernetes-ingress:1.10.10.

I've noticed that, with only 1 server registered to my backend, when I set the following config to my core.haproxy.org/v1alpha2:Backend:

  config:
    default_server:
      maxconn: 1

And when I run only 1 haproxy instance, then the limit works properly and I can see the subsequent requests piling up in the queue.

But if I make my ingress controller scale up to 2, then I get about 50% chance having a following request pile up in the queue or being sent to the actual server.

As you can see in the screenshot: limit is set to 1 but the max went to 2.

And I can actually port-forward into the 2 different ingress controller individual stats pages, shouldn't I have only one ?

Do you know of any config that would prevent this unwanted behaviour ?

Thanks for your help
Cheers

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.