Do not delete pods and let stateful sets handle restarts (#153)

With the right timeout setting (part of PR #152)
there is no need to restart pods part of the old zone manually as all pods get restarted.

pkg/controller/cluster/main-controller.go

@@ -23,7 +23,6 @@ import (
 	"crypto/sha256"
 	"encoding/hex"
 	"fmt"
-	"strconv"
 	"time"
 
 	"k8s.io/klog"
@@ -469,7 +468,6 @@ func (c *Controller) syncHandler(key string) error {
 		if mi.MinIOReplicas() != *ss.Spec.Replicas && mi.Status.CurrentState != addingZone {
 			// save current replicas before creating new statefulset
 			// this is used later to delete only the older pods in statefulset
-			currentReplicas := mi.MinIOReplicas()
 			mi, err = c.updateMinIOInstanceStatus(ctx, mi, addingZone, 0)
 			if err != nil {
 				return err
@@ -496,12 +494,6 @@ func (c *Controller) syncHandler(key string) error {
 			if _, err := c.kubeClientSet.AppsV1().StatefulSets(mi.Namespace).Update(ctx, ss, uOpts); err != nil {
 				return err
 			}
-			// remove all the existing Pods so StatefulSet creates new pods with proper secrets/cli args etc
-			for i := 0; i < int(currentReplicas); i++ {
-				if err := c.kubeClientSet.CoreV1().Pods(mi.Namespace).Delete(ctx, "minio-"+strconv.Itoa(i), metav1.DeleteOptions{}); err != nil {
-					return err
-				}
-			}
 		}
 
 		// If this container version on the MinIOInstance resource is specified, and the