Removal of unhealthy server pods on java.net.SocketTimeoutException: Connect Timeout

Hello Team,

Is your feature request related to a problem? Please describe.
We are using Spring Cloud Load Balancer version 2.0.5 on Kubernetes v1.20.0.
When a server pod terminates the requests that are load balanced towards the pod start getting the java.net.SocketTimeoutException: Connect Timeout exception.

Describe the solution you'd like
On a event of a Connect timeout the pod should be removed from the service list immediately.
This will still result in 1st packet getting the exception but subsequent requests will load balanced among remaining healthy pods.

Describe alternatives you've considered
To solve this problem we had enabled health-check.refresh-instances-interval but until the health check timeout the requests are routed to the deleted pod.

Additional context
Please let us know if there is a way to remove the service instances based on Connect timeout exception or is there a feature which is planned to achieve this.

Thanks and Regards,
Vijit

Start by reviewing Spring Cloud Load Balancer's instance lifecycle and how java.net.SocketTimeoutException connection failures are handled. Define the required behavior for removing an unhealthy Kubernetes pod from the service list, then verify that subsequent requests avoid it while the first failed request remains expected.