Before the sidecar would immediately do a sync and check the ready state. If the pod wasn't ready at that moment then the sidecar would die, which would tear down the pod. If you have a service which takes some time to be ready this is less-than-desirable. As a simple solution to this the sidecar will now retry indefinitely (until killed) to sync the service with consul. This keeps the katalog-sync code simple and allows the user to control the timing etc. through k8s configuration of the readiness.
Before the sidecar would immediately do a sync and check the ready state. If the pod wasn't ready at that moment then the sidecar would die, which would tear down the pod. If you have a service which takes some time to be ready this is less-than-desirable. As a simple solution to this the sidecar will now retry indefinitely (until killed) to sync the service with consul. This keeps the katalog-sync code simple and allows the user to control the timing etc. through k8s configuration of the readiness.