diff --git a/cmd/kk/pkg/etcd/module.go b/cmd/kk/pkg/etcd/module.go
index 0de29475c..f454cba03 100644
--- a/cmd/kk/pkg/etcd/module.go
+++ b/cmd/kk/pkg/etcd/module.go
@@ -311,15 +311,6 @@ func handleExistCluster(c *ConfigureModule) []task.Interface {
 		Parallel: false,
 	}
 
-	restart := &task.RemoteTask{
-		Name:     "RestartETCD",
-		Desc:     "Restart etcd",
-		Hosts:    c.Runtime.GetHostsByRole(common.ETCD),
-		Prepare:  &NodeETCDExist{Not: true},
-		Action:   new(RestartETCD),
-		Parallel: true,
-	}
-
 	newETCDNodeHealthCheck := &task.RemoteTask{
 		Name:     "NewETCDNodeHealthCheck",
 		Desc:     "Health check on new etcd",
@@ -360,7 +351,6 @@ func handleExistCluster(c *ConfigureModule) []task.Interface {
 		existETCDHealthCheck,
 		generateETCDConfig,
 		joinMember,
-		restart,
 		newETCDNodeHealthCheck,
 		checkMember,
 		allRefreshETCDConfig,
diff --git a/cmd/kk/pkg/etcd/tasks.go b/cmd/kk/pkg/etcd/tasks.go
index 400ca426e..052991e55 100644
--- a/cmd/kk/pkg/etcd/tasks.go
+++ b/cmd/kk/pkg/etcd/tasks.go
@@ -349,6 +349,12 @@ func (j *JoinMember) Execute(runtime connector.Runtime) error {
 	} else {
 		return errors.New("get etcd cluster status by pipeline cache failed")
 	}
+
+	// After adding a new member for etcd, it is necessary to start the new member as the etcd cluster may experience abnormal behavior.
+	if _, err := runtime.GetRunner().SudoCmd("systemctl daemon-reload && systemctl restart etcd && systemctl enable etcd", true); err != nil {
+		return errors.Wrap(errors.WithStack(err), "start etcd failed")
+	}
+
 	return nil
 }