Exit On Error does not exit if LaunchDarkly SaaS is unreachable

[gravufo]

Describe the bug

We are currently implementing LaunchDarkly Relay Proxy on our clusters and were expecting EXIT_ON_ERROR to fail with an error at startup if LaunchDarkly SaaS is unreachable. The reason we want this behavior is that if LaunchDarkly is down and our proxies are restarted during that time and are only configured with in-memory cache, we end up with unusable pods. Ideally, they would never become ready unless they have an external cache or area able to reach the LaunchDarkly SaaS.

To reproduce

Deploy the LD-Relay and setup some network policy to prevent the pods from reaching LaunchDarkly's URLs. You will see the pod will still become healthy and ready.

Expected behavior

I would want the pod to fail at startup if LD is unreachable and no external cache is configured.

[keelerm84]

Thank you for raising this request. This is something we are aware of, and have plans to address. As we progress on this issue, I will update back here.

Tracking this internally as SDKPD-56.