Does router cache subgraph url (Load balancer DNS/IP)?

[purushottam-anand]

We have a federated graphql setup having 18 different subgraphs. All the subgraphs follow blue green deployment. Shifting traffic from blue stack to green stack (or vice-versa) involves switching of route53 weights from 0 to 100. Now with this setup and TTL for R53 being 60s, we expect all the traffic from router to this subgraph to be coming to the new stack (say green) after a period of almost a minute. But on the contrary we have observed that router continues to send the requests to old stack (blue) for a very prolonged period (~40-50 mins). Does this mean router caches subgraph's load balancer DNS(IP)? If yes how to flush that cache in say a minute ? Is there any way to configure that ?