We are running our spring boot on AWS ECS with 2 instances on production. We have put an Application Load Balancer on top of it with round-robin algorithm for 2 targets but I am seeing that most the requests are going to only one instance. Requests are sequential and stickiness is off.
There are two requests coming from the same client in one it is just to pre-warm the local ehcache and second is the actual request which will create a deal. Now as per my knowledge in round-robin it cannot be possible as request 1 will go to Instance A while request 2 will go to Instance B -- but it is happening.
Please let me know where my understanding is wrong.