On our 4.3 Azure periodic jobs, I've noticed some GRPC flakes with errors like below

    grpc_test.go:377: The endpoint for Route g-r-p-c-unary-ping-xrouodqa at http://g-r-p-c-unary-ping-xrouodqa.serving-tests.apps.ci-op-7wkb3bfx-fd231.ci.azure.devcluster.openshift.com didn't return success: failed get the cluster endpoint: expected exactly one ingress load balancer, instead had 0: []
--- FAIL: TestGRPCAutoscaleUpDownUp (8.58s)

When manually deploying OpenShift Serverless on a 4.3 Azure cluster, I do get an ingress load balancer. My suspicion is it's taking longer for this to happen on Azure than expected. I did see a number of rate limit events for the kourier Kubernetes Service in the knative-serving-ingress namespace around Azure networking resources. Perhaps under certain situations it takes an abnormally time for these to get created.