Rui Almeida

This fix is very much needed, the LogCacheStore tends to die after a while and the service requires a manual restart. Could someone please review? @oavdeev @obgibson @jackie-ob

Hey @jackie-ob , setting the HTTP status code >= 400 is the standard way to signal that a service is unhealthy (in AWS ELB and k8s): - https://docs.aws.amazon.com/elasticloadbalancing/latest/network/target-group-health-checks.html - https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-a-liveness-http-request...

Sure, I can try the new version, I think the fix from https://github.com/Netflix/metaflow-service/pull/292/files#diff-17bee700a8b2eb91cc38dcb3d7e30cd88769966fdb21c724393a5746f664d3cf is not included tho. A new endpoint in the future with the purpose of health checking the...

Is there any update on this issue? We are currently monkey patching our MLflow Docker image to get the icons back.