If you turn on proxy protocol support two apps in the same rack can no longer communicate with each other unless they manually set proxy protocol headers on their own requests.
The NGiNX ingress/load balancer that routes traffic is configured to reject all requests without the proxy protocol header, but since the rack’s router routes the traffic locally it is missing the header which would be applied at the AWS LB level.
There’s no good fix for this AFAIK.
But I just wanted to call this out in case other people are having this issue.
It’s also worth noting that you need to be careful if you’ve been setting your rack to “Local” instead of “Cluster” traffic mode at the service level. You need to revert to “Cluster” before turning on proxy protocol or you’ll blow your rack offline with no way to fix it.