fix: increase timeout for wait namespace delete (#2359)

framework/app-service/.olares/config/cluster/deploy/appservice_deploy.yaml

@@ -170,7 +170,7 @@ spec:
       priorityClassName: "system-cluster-critical"
       containers:
       - name: app-service
-        image: beclab/app-service:0.4.70
+        image: beclab/app-service:0.4.72
         imagePullPolicy: IfNotPresent
         securityContext:
           runAsUser: 0

framework/app-service/controllers/load.go

@@ -146,7 +146,7 @@ func LoadStatefulApp(ctx context.Context, appmgr *ApplicationManagerController,
 		case appv1alpha1.ApplyingEnvCanceling:
 			return appstate.NewApplyingEnvCancelingApp(appmgr, &am)
 		case appv1alpha1.Uninstalling:
-			return appstate.NewUninstallingApp(appmgr, &am, 15*time.Minute)
+			return appstate.NewUninstallingApp(appmgr, &am, 30*time.Minute)
 		case appv1alpha1.StopFailed:
 			return appstate.NewSuspendFailedApp(appmgr, &am)
 		case appv1alpha1.UninstallFailed:

framework/app-service/pkg/appinstaller/helm_ops_upgrade.go

@@ -2,7 +2,6 @@ package appinstaller
 
 import (
 	"encoding/json"
-	"fmt"
 	"time"
 
 	appv1alpha1 "github.com/beclab/Olares/framework/app-service/api/app.bytetrade.io/v1alpha1"
@@ -12,7 +11,6 @@ import (
 	"github.com/beclab/Olares/framework/app-service/pkg/helm"
 	"github.com/beclab/Olares/framework/app-service/pkg/users/userspace"
 	apputils "github.com/beclab/Olares/framework/app-service/pkg/utils/app"
-	helmrelease "helm.sh/helm/v3/pkg/release"
 	apierrors "k8s.io/apimachinery/pkg/api/errors"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/types"
@@ -23,15 +21,7 @@ import (
 
 // Upgrade do a upgrade operation for release.
 func (h *HelmOps) Upgrade() error {
-	status, err := h.status()
-	if err != nil {
-		klog.Errorf("get release status failed %v", err)
-		return err
-	}
-	if status.Info.Status == helmrelease.StatusDeployed {
-		return h.upgrade()
-	}
-	return fmt.Errorf("cannot upgrade release %s/%s, current state is %s", h.app.Namespace, h.app.AppName, status.Info.Status)
+	return h.upgrade()
 }
 
 func (h *HelmOps) upgrade() error {

framework/app-service/pkg/appstate/types.go

@@ -20,6 +20,7 @@ import (
 	apierrors "k8s.io/apimachinery/pkg/api/errors"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/types"
+	utilwait "k8s.io/apimachinery/pkg/util/wait"
 	"k8s.io/client-go/rest"
 	"k8s.io/klog/v2"
 	ctrl "sigs.k8s.io/controller-runtime"
@@ -138,6 +139,13 @@ func (p *baseStatefulApp) forceDeleteApp(ctx context.Context) error {
 			return err
 		}
 	}
+
+	// Wait for namespace to be fully deleted before updating status
+	if err = p.waitForNamespaceDeleted(ctx); err != nil {
+		klog.Errorf("wait for namespace %s deleted failed %v", p.manager.Spec.AppNamespace, err)
+		return err
+	}
+
 	err = p.updateStatus(ctx, p.manager, appsv1.Uninstalled, nil, appsv1.Uninstalled.String(), "")
 	if err != nil {
 		klog.Errorf("update app manager %s to state %s failed", p.manager.Name, appsv1.Uninstalled)
@@ -146,6 +154,32 @@ func (p *baseStatefulApp) forceDeleteApp(ctx context.Context) error {
 	return nil
 }
 
+// waitForNamespaceDeleted waits for the namespace to be completely deleted
+func (p *baseStatefulApp) waitForNamespaceDeleted(ctx context.Context) error {
+	namespace := p.manager.Spec.AppNamespace
+	if apputils.IsProtectedNamespace(namespace) {
+		return nil
+	}
+
+	klog.Infof("waiting for namespace %s to be fully deleted", namespace)
+	err := utilwait.PollImmediate(time.Second, 30*time.Minute, func() (done bool, err error) {
+		var ns corev1.Namespace
+		err = p.client.Get(ctx, types.NamespacedName{Name: namespace}, &ns)
+		if err != nil && !apierrors.IsNotFound(err) {
+			klog.Errorf("failed to get namespace %s: %v", namespace, err)
+			return false, err
+		}
+		if apierrors.IsNotFound(err) {
+			klog.Infof("namespace %s has been fully deleted", namespace)
+			return true, nil
+		}
+		klog.Infof("namespace %s still exists, waiting...", namespace)
+		return false, nil
+	})
+
+	return err
+}
+
 type OperationApp interface {
 	StatefulApp
 	IsTimeout() bool

framework/app-service/pkg/appstate/uninstalling_app.go

@@ -100,7 +100,7 @@ func (p *UninstallingApp) waitForDeleteNamespace(ctx context.Context) error {
 	if apputils.IsProtectedNamespace(p.manager.Spec.AppNamespace) {
 		return nil
 	}
-	err := utilwait.PollImmediate(time.Second, 15*time.Minute, func() (done bool, err error) {
+	err := utilwait.PollImmediate(time.Second, 30*time.Minute, func() (done bool, err error) {
 		klog.Infof("waiting for namespace %s to be deleted", p.manager.Spec.AppNamespace)
 		nsName := p.manager.Spec.AppNamespace
 		var ns corev1.Namespace

framework/app-service/pkg/appstate/upgrading_app.go

@@ -176,6 +176,7 @@ func (p *UpgradingApp) exec(ctx context.Context) error {
 			return err
 		}
 		appConfig.Ports = cfg.Ports
+		appConfig.TailScale = cfg.TailScale
 
 	} else {
 		_, err = apputils.GetIndexAndDownloadChart(ctx, &apputils.ConfigOptions{

framework/app-service/pkg/helm/helm.go

@@ -87,7 +87,8 @@ func UpgradeCharts(ctx context.Context, actionConfig *action.Configuration, sett
 	client.Namespace = namespace
 	client.Timeout = 300 * time.Second
 	client.Recreate = false
-	client.Atomic = true
+	// Do not use Atomic, this could cause helm wait all resource ready.
+	//client.Atomic = true
 	if reuseValue {
 		client.ReuseValues = true
 	}