There seem to be an issue with kong, how to work on this kind of scenario, is it a bug?
Problem: We have 2 services, ServiceA and ServiceB both are deployed in AWS ECS Container. And both of them have /status as there healthcheck point.
So Lets assume
ServiceA Deployed on IP-1:PortA
ServiceB Deployed on IP-1:PortB
both of them are successfully registered to kong gateway.
Now suppose ServiceA and ServiceB got restarted multiple times being in same ECS cluster Now serviceB got respaned With IP-1:PortA as that was free port and got reused.
Supposing the serviceB die abrubtly and dont have any gracefull termination policy.
Now when we access
kong.apigwatey.internal.com/serviceA → This traffic is served by both A and B service.
Reason being that Targets are not cleaned by kong and they are actively checked constantly if they turn healthy.
How to deal in this kind of scenario.