We have a situation where have very regular traffic through our AHC client to a particular endpoint. It appears that if the endpoint we are routing to has a dns change (e.g. it starts resolving to a new IP due to a failover scenario), the connection in the pool using the old IP is never dropped even if all HTTP responses are errors (502's).

Setting org.asynchttpclient.connectionTtl to something other than -1 addresses this, but just wanted to check if there are other options. Ideally we want to see it something close to the DNS TTL (e.g. 30000 for 30 seconds) but am concerned that there could be a performance impact from restablishing connections so frequently. Wasn't sure if there was, say, any AHC configuration to drop a connection after N consecutive failed responses. I didn't see any in ahc-default.properties.