Eureka Client not resilient to socket read timeouts not getting triggered

[NiteshKant]

Eureka client uses thread pool schedulers to schedule tasks:

• Fetch registry
• Heartbeats

This assumes that every scheduled task finishes either with success or error. If this assumption is broken, no new tasks (for the same type) gets scheduled.
In some scenarios we have seen inside AWS that the socket read timeouts do not get triggered which makes the thread reading HTTP response stuck and hence break the above assumption.

The client should be able to identify this scenario and be resilient.