Delay in state propagation of ATP from OCI to ServiceInstance

[lake-of-dreams]

It was observed while both creating and deleting ServiceInstances for ATP that their is a random delay for state propagation to the ServiceInstance from actual state of ATP in OCI. The behaviour is intermittent but should be addressed.

While create - The ATP might already have provisioned in OCI - but the Status on ServiceInstance will still be Provisioning for as long as 4-5 minutes after the ATP might have provisioned.

Similarly while delete - the Status on ServiceInstance will still be Terminating even after the ATP is terminated

[ashokkkannan]

Each time the oci-service-broker is polled for the status we go to the oci service and get the current status. The delay you are seeing is from the service-catalog. They are polling the oci-service-broker with an exponential back-off. Please tweak the service-catalog configuration to change this polling time to reduce the delay. For example, if you change the below value to 20s then the service-catalog will pool oci-service-broker for status at least once every 20s.
https://github.com/kubernetes-sigs/service-catalog/blob/master/charts/catalog/values.yaml#L70

[jayasheelankumar]

@lake-of-dreams Please let us know if we can close this issue?