Use Patch to Scale Deployments

When scaling cluster Deployments, and Patch operation is now used
instead of an Update.  This prevents potential conflicts from
occurring when attempting to scale the Deployments for a cluster
(e.g. the Deployments for PG pods and/or the pgBackRest repository).

Author: andrewlecuyer

internal/operator/cluster/clusterlogic.go

@@ -41,6 +41,7 @@ import (
 	kerrors "k8s.io/apimachinery/pkg/api/errors"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/fields"
+	"k8s.io/apimachinery/pkg/types"
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/rest"
 )
@@ -762,11 +763,12 @@ func ScaleClusterDeployments(clientset kubernetes.Interface, cluster crv1.Pgclus
 		log.Debugf("scaling deployment %s to %d for cluster %s", deployment.Name, replicas,
 			clusterName)
 
-		// Scale the deployment accoriding to the number of replicas specified.  If an error is
-		// encountered, log it and move on to scaling the next deployment.
-		deployment.Spec.Replicas = kubeapi.Int32(int32(replicas))
-		if _, err = clientset.AppsV1().Deployments(deployment.Namespace).Update(&deployment); err != nil {
-			log.Errorf("Error scaling deployment %s to %d: %w", deployment.Name, replicas, err)
+		// Scale the deployment according to the number of replicas specified.  If an error is
+		// encountered, log it and move on to scaling the next deployment
+		patchString := fmt.Sprintf(`{"spec":{"replicas":%d}}`, replicas)
+		if _, err := clientset.AppsV1().Deployments(namespace).Patch(deployment.GetName(),
+			types.MergePatchType, []byte(patchString)); err != nil {
+			log.Errorf("Error scaling deployment %s to %d: %v", deployment.Name, replicas, err)
 		}
 	}
 	return