Merge pull request #6550 from gandhipr/prachigandhi/azure-manager-retry

Add retry for creatingAzureManager in case of throttled requests

cluster-autoscaler/cloudprovider/azure/azure_manager.go

@@ -26,10 +26,13 @@ import (
 	"time"
 
 	"github.com/Azure/go-autorest/autorest/azure"
+	"k8s.io/apimachinery/pkg/util/wait"
 	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider"
 	"k8s.io/autoscaler/cluster-autoscaler/config"
 	"k8s.io/autoscaler/cluster-autoscaler/config/dynamic"
+	kretry "k8s.io/client-go/util/retry"
 	klog "k8s.io/klog/v2"
+	"sigs.k8s.io/cloud-provider-azure/pkg/retry"
 )
 
 const (
@@ -106,7 +109,18 @@ func createAzureManagerInternal(configReader io.Reader, discoveryOpts cloudprovi
 		return nil, err
 	}
 
-	if err := manager.forceRefresh(); err != nil {
+	retryBackoff := wait.Backoff{
+		Duration: 2 * time.Minute,
+		Factor:   1.0,
+		Jitter:   0.1,
+		Steps:    6,
+		Cap:      10 * time.Minute,
+	}
+
+	err = kretry.OnError(retryBackoff, retry.IsErrorRetriable, func() (err error) {
+		return manager.forceRefresh()
+	})
+	if err != nil {
 		return nil, err
 	}