feat: Support custom ArgoCD namespace configuration (#134)

- Add ARGOCD_NAMESPACE and ARGOCD_OPERATOR_NAMESPACE environment variables
- Update GitOpsCluster controller to pass namespace config via AddOnDeploymentConfig
- Refactor addon install/cleanup to use configurable namespaces instead of hardcoded values
- Add e2e test for custom namespace scenarios (hub: notargocd, spoke: notargocd)
- Extract cluster setup into reusable make target (setup-e2e-clusters)
- Update Helm installation to use --create-namespace flag
- Update GitHub Actions workflow to include custom namespace e2e test
- Fix README with correct Helm installation commands

Signed-off-by: Mike Ng <[email protected]>

Author: mikeshng

.github/workflows/e2e.yml

@@ -136,3 +136,62 @@ jobs:
         env:
           KUBECONFIG: /home/runner/.kube/config
         run: make test-e2e-cleanup
+
+  e2e-custom-namespace:
+    name: e2e-custom-namespace
+    runs-on: ubuntu-latest
+    steps:
+      - name: checkout code
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+          path: go/src/open-cluster-management.io/argocd-pull-integration
+
+      - name: install Go
+        uses: actions/setup-go@v5
+        with:
+          go-version: ${{ env.GO_VERSION }}
+
+      - name: Add GOPATH/bin to PATH
+        run: echo "$GOPATH/bin" >> "$GITHUB_PATH"
+
+      - name: install imagebuilder
+        run: go install github.com/openshift/imagebuilder/cmd/[email protected]
+
+      - name: Set up Helm
+        uses: azure/setup-helm@v4
+        with:
+          version: v3.15.2
+
+      - name: Set up kubectl
+        uses: azure/setup-kubectl@v4
+        with:
+          version: v1.29.0
+
+      - name: build-images
+        run: make build-images
+
+      - name: setup kind (cluster1)
+        uses: engineerd/[email protected]
+        with:
+          version: v0.14.0
+          name: cluster1
+
+      - name: setup kind (hub)
+        uses: engineerd/[email protected]
+        with:
+          version: v0.14.0
+          name: hub
+
+      - name: Load image on the nodes of the hub
+        run: |
+          kind load docker-image --name=hub quay.io/open-cluster-management/argocd-pull-integration:latest
+
+      - name: Load image on the nodes of the cluster1
+        run: |
+          kind load docker-image --name=cluster1 quay.io/open-cluster-management/argocd-pull-integration:latest
+
+      - name: Run e2e custom namespace test
+        env:
+          KUBECONFIG: /home/runner/.kube/config
+        run: make test-e2e-custom-namespace

Makefile

@@ -69,6 +69,22 @@ E2E_IMG ?= quay.io/open-cluster-management/argocd-pull-integration:latest
 
 ##@ E2E Tests
 
+.PHONY: setup-e2e-clusters
+setup-e2e-clusters: ## Setup KinD clusters, build and load controller image
+	@echo "===== Cleaning up existing clusters ====="
+	$(KIND) delete clusters --all || true
+	@echo ""
+	@echo "===== Creating KinD clusters ====="
+	$(KIND) create cluster --name $(HUB_CLUSTER)
+	$(KIND) create cluster --name $(SPOKE_CLUSTER)
+	@echo ""
+	@echo "===== Building controller image ====="
+	$(MAKE) docker-build IMG=$(E2E_IMG)
+	@echo ""
+	@echo "===== Loading image to clusters ====="
+	$(KIND) load docker-image $(E2E_IMG) --name $(HUB_CLUSTER)
+	$(KIND) load docker-image $(E2E_IMG) --name $(SPOKE_CLUSTER)
+
 .PHONY: test-e2e
 test-e2e: manifests generate fmt vet ## Run e2e deployment tests only (checks pods running and logs)
 	@echo "===== Installing MetalLB ====="
@@ -79,12 +95,10 @@ test-e2e: manifests generate fmt vet ## Run e2e deployment tests only (checks po
 	@echo ""
 	@echo "===== Installing addon via Helm ====="
 	$(KUBECTL) config use-context kind-$(HUB_CLUSTER)
-	$(KUBECTL) create namespace argocd || true
-	$(KUBECTL) label namespace argocd app.kubernetes.io/managed-by=Helm --overwrite
-	$(KUBECTL) annotate namespace argocd meta.helm.sh/release-name=argocd-agent-addon meta.helm.sh/release-namespace=argocd --overwrite
 	helm install argocd-agent-addon \
 		./charts/argocd-agent-addon \
 		--namespace argocd \
+		--create-namespace \
 		--set image=quay.io/open-cluster-management/argocd-pull-integration \
 		--set tag=latest \
 		--wait \
@@ -117,19 +131,7 @@ test-e2e-integration: ## Run full e2e integration tests including AppProject and
 
 .PHONY: test-e2e-full
 test-e2e-full: ## Complete e2e test with kind cluster setup, build, deployment, and full integration tests
-	@echo "===== Cleaning up existing clusters ====="
-	$(KIND) delete clusters --all || true
-	@echo ""
-	@echo "===== Creating KinD clusters ====="
-	$(KIND) create cluster --name $(HUB_CLUSTER)
-	$(KIND) create cluster --name $(SPOKE_CLUSTER)
-	@echo ""
-	@echo "===== Building controller image ====="
-	$(MAKE) docker-build IMG=$(E2E_IMG)
-	@echo ""
-	@echo "===== Loading image to clusters ====="
-	$(KIND) load docker-image $(E2E_IMG) --name $(HUB_CLUSTER)
-	$(KIND) load docker-image $(E2E_IMG) --name $(SPOKE_CLUSTER)
+	$(MAKE) setup-e2e-clusters
 	@echo ""
 	@echo "===== Running deployment tests ====="
 	$(MAKE) test-e2e
@@ -171,12 +173,10 @@ test-e2e-cleanup: manifests generate fmt vet ## Run e2e cleanup tests (checks ad
 	@echo ""
 	@echo "===== Installing addon via Helm ====="
 	$(KUBECTL) config use-context kind-$(HUB_CLUSTER)
-	$(KUBECTL) create namespace argocd || true
-	$(KUBECTL) label namespace argocd app.kubernetes.io/managed-by=Helm --overwrite
-	$(KUBECTL) annotate namespace argocd meta.helm.sh/release-name=argocd-agent-addon meta.helm.sh/release-namespace=argocd --overwrite
 	helm install argocd-agent-addon \
 		./charts/argocd-agent-addon \
 		--namespace argocd \
+		--create-namespace \
 		--set image=quay.io/open-cluster-management/argocd-pull-integration \
 		--set tag=latest \
 		--wait \
@@ -189,19 +189,7 @@ test-e2e-cleanup: manifests generate fmt vet ## Run e2e cleanup tests (checks ad
 
 .PHONY: test-e2e-cleanup-full
 test-e2e-cleanup-full: ## Complete e2e test with cleanup verification including Application (cluster setup + deployment + cleanup)
-	@echo "===== Cleaning up existing clusters ====="
-	$(KIND) delete clusters --all || true
-	@echo ""
-	@echo "===== Creating KinD clusters ====="
-	$(KIND) create cluster --name $(HUB_CLUSTER)
-	$(KIND) create cluster --name $(SPOKE_CLUSTER)
-	@echo ""
-	@echo "===== Building controller image ====="
-	$(MAKE) docker-build IMG=$(E2E_IMG)
-	@echo ""
-	@echo "===== Loading image to clusters ====="
-	$(KIND) load docker-image $(E2E_IMG) --name $(HUB_CLUSTER)
-	$(KIND) load docker-image $(E2E_IMG) --name $(SPOKE_CLUSTER)
+	$(MAKE) setup-e2e-clusters
 	@echo ""
 	@echo "===== Installing MetalLB ====="
 	./test/e2e/scripts/install_metallb.sh
@@ -211,12 +199,10 @@ test-e2e-cleanup-full: ## Complete e2e test with cleanup verification including
 	@echo ""
 	@echo "===== Installing addon via Helm ====="
 	$(KUBECTL) config use-context kind-$(HUB_CLUSTER)
-	$(KUBECTL) create namespace argocd || true
-	$(KUBECTL) label namespace argocd app.kubernetes.io/managed-by=Helm --overwrite
-	$(KUBECTL) annotate namespace argocd meta.helm.sh/release-name=argocd-agent-addon meta.helm.sh/release-namespace=argocd --overwrite
 	helm install argocd-agent-addon \
 		./charts/argocd-agent-addon \
 		--namespace argocd \
+		--create-namespace \
 		--set image=quay.io/open-cluster-management/argocd-pull-integration \
 		--set tag=latest \
 		--wait \
@@ -229,6 +215,42 @@ test-e2e-cleanup-full: ## Complete e2e test with cleanup verification including
 	@echo "Hub context: kind-$(HUB_CLUSTER)"
 	@echo "Spoke context: kind-$(SPOKE_CLUSTER)"
 
+.PHONY: test-e2e-custom-namespace
+test-e2e-custom-namespace: manifests generate fmt vet ## Run e2e test with custom ArgoCD namespaces (hub: notargocd, spoke: argocdnot)
+	@echo "===== Installing MetalLB ====="
+	./test/e2e/scripts/install_metallb.sh
+	@echo ""
+	@echo "===== Setting up OCM environment ====="
+	./test/e2e/scripts/setup_ocm_env.sh
+	@echo ""
+	@echo "===== Installing addon via Helm with custom namespaces ====="
+	$(KUBECTL) config use-context kind-$(HUB_CLUSTER)
+	helm install argocd-agent-addon \
+		./charts/argocd-agent-addon \
+		--namespace notargocd \
+		--create-namespace \
+		--set global.argoCDNamespace=notargocd \
+		--set global.argoCDOperatorNamespace=argocd-operator-system \
+		--set controller.namespace=notargocd \
+		--set gitOpsCluster.namespace=notargocd \
+		--set image=quay.io/open-cluster-management/argocd-pull-integration \
+		--set tag=latest \
+		--wait \
+		--timeout 10m
+	@echo ""
+	@echo "===== Running e2e tests with custom namespaces ====="
+	HUB_ARGOCD_NAMESPACE=notargocd go test -tags=e2e ./test/e2e/ -v -ginkgo.v --ginkgo.label-filter="custom-namespace"
+	@echo ""
+	@echo "===== E2E Custom Namespace Tests Complete ====="
+	@echo "Hub context: kind-$(HUB_CLUSTER) (ArgoCD in notargocd namespace)"
+	@echo "Spoke context: kind-$(SPOKE_CLUSTER) (ArgoCD in notargocd namespace)"
+
+.PHONY: test-e2e-custom-namespace-full
+test-e2e-custom-namespace-full: ## Complete e2e test with custom namespaces (cluster setup + custom namespace deployment + tests)
+	$(MAKE) setup-e2e-clusters
+	@echo ""
+	$(MAKE) test-e2e-custom-namespace
+
 ##@ Build
 
 .PHONY: build

README.md

@@ -138,7 +138,7 @@ For detailed argocd-agent architecture and operational modes, see [argocd-agent
 helm repo add ocm https://open-cluster-management.io/helm-charts
 helm repo update
 helm search repo ocm
-helm install argocd-agent-addon ocm/argocd-agent-addon
+helm install argocd-agent-addon ocm/argocd-agent-addon --namespace argocd --create-namespace
 ```
 
 This installs the GitOpsCluster controller and creates a GitOpsCluster resource that automatically deploys argocd-agent to your managed clusters.

charts/argocd-agent-addon/templates/argocd-operator/operator.yaml

@@ -8,14 +8,6 @@ metadata:
   name: {{ .Values.global.argoCDOperatorNamespace }}
 ---
 apiVersion: v1
-kind: Namespace
-metadata:
-  labels:
-    control-plane: argocd-operator
-    app.kubernetes.io/managed-by: argocd-agent-addon
-  name: {{ .Values.global.argoCDNamespace }}
----
-apiVersion: v1
 kind: ServiceAccount
 metadata:
   name: argocd-operator-controller-manager

internal/addon/addon_cleanup.go

@@ -33,12 +33,25 @@ import (
 )
 
 const (
-	operatorNamespace = "argocd-operator-system"
-	argoCDNamespace   = "argocd"
 	// PauseMarkerName is the name of the ConfigMap used to pause the addon controller
 	PauseMarkerName = "argocd-agent-addon-pause"
 )
 
+// getNamespaceConfig returns namespace configuration from environment variables or defaults
+func getNamespaceConfig() (operatorNamespace, argoCDNamespace string) {
+	operatorNamespace = os.Getenv("ARGOCD_OPERATOR_NAMESPACE")
+	if operatorNamespace == "" {
+		operatorNamespace = "argocd-operator-system"
+	}
+
+	argoCDNamespace = os.Getenv("ARGOCD_NAMESPACE")
+	if argoCDNamespace == "" {
+		argoCDNamespace = "argocd"
+	}
+
+	return operatorNamespace, argoCDNamespace
+}
+
 // uninstallArgoCDAgent uninstalls the ArgoCD agent addon in reverse order
 // Does NOT delete namespaces, only deletes operator after ArgoCD CR is gone
 func (r *ArgoCDAgentAddonReconciler) uninstallArgoCDAgent(ctx context.Context) error {
@@ -59,6 +72,10 @@ func uninstallArgoCDAgentInternal(ctx context.Context, c client.Client) error {
 	klog.Info("Starting ArgoCD agent addon uninstall")
 	var cleanupErrors []error
 
+	// Get namespace configuration from environment
+	operatorNamespace, argoCDNamespace := getNamespaceConfig()
+	klog.Infof("Using namespaces - operator: %s, argocd: %s", operatorNamespace, argoCDNamespace)
+
 	// Step 0: Create pause marker to prevent the addon controller from reconciling resources
 	klog.Infof("Step 0: Creating pause marker in namespace: %s", operatorNamespace)
 	if err := createPauseMarker(ctx, c, operatorNamespace); err != nil {
@@ -122,7 +139,7 @@ func uninstallArgoCDAgentInternal(ctx context.Context, c client.Client) error {
 
 	// Step 2: ArgoCD CR is gone (or timed out), now delete operator resources
 	klog.Info("Step 2: Deleting operator resources from namespace: " + operatorNamespace)
-	if err := deleteOperatorResourcesInternal(ctx, c); err != nil {
+	if err := deleteOperatorResourcesInternal(ctx, c, operatorNamespace); err != nil {
 		klog.Errorf("Error deleting operator resources (continuing with cleanup): %v", err)
 		cleanupErrors = append(cleanupErrors, fmt.Errorf("failed to delete operator resources: %w", err))
 	}
@@ -133,7 +150,7 @@ func uninstallArgoCDAgentInternal(ctx context.Context, c client.Client) error {
 	waitDuration := getCleanupVerificationWaitDuration()
 	if waitDuration > 0 {
 		klog.Infof("Step 3: Waiting and re-verifying that resources stay deleted (%v)...", waitDuration)
-		if err := waitAndVerifyCleanup(ctx, c, waitDuration); err != nil {
+		if err := waitAndVerifyCleanup(ctx, c, waitDuration, argoCDNamespace, operatorNamespace); err != nil {
 			klog.Errorf("Error during cleanup verification (resources may have been recreated): %v", err)
 			cleanupErrors = append(cleanupErrors, fmt.Errorf("cleanup verification failed: %w", err))
 		}
@@ -153,11 +170,12 @@ func uninstallArgoCDAgentInternal(ctx context.Context, c client.Client) error {
 
 // deleteOperatorResources deletes operator resources (deployment, RBAC, etc.)
 func (r *ArgoCDAgentAddonReconciler) deleteOperatorResources(ctx context.Context) error {
-	return deleteOperatorResourcesInternal(ctx, r.Client)
+	_, operatorNamespace := getNamespaceConfig()
+	return deleteOperatorResourcesInternal(ctx, r.Client, operatorNamespace)
 }
 
 // deleteOperatorResourcesInternal deletes operator resources (deployment, RBAC, etc.)
-func deleteOperatorResourcesInternal(ctx context.Context, c client.Client) error {
+func deleteOperatorResourcesInternal(ctx context.Context, c client.Client, operatorNamespace string) error {
 	klog.Info("Deleting operator resources")
 
 	// Delete resources in the operator namespace with our management label
@@ -287,7 +305,7 @@ func IsPaused(ctx context.Context, c client.Client, namespace string) bool {
 
 // waitAndVerifyCleanup waits for a specified duration and periodically verifies that
 // resources stay deleted. If resources are recreated, it deletes them again.
-func waitAndVerifyCleanup(ctx context.Context, c client.Client, waitDuration time.Duration) error {
+func waitAndVerifyCleanup(ctx context.Context, c client.Client, waitDuration time.Duration, argoCDNamespace, operatorNamespace string) error {
 	klog.Infof("Starting cleanup verification for %v", waitDuration)
 
 	checkInterval := 20 * time.Second
@@ -325,7 +343,7 @@ func waitAndVerifyCleanup(ctx context.Context, c client.Client, waitDuration tim
 
 		if operatorResourcesExist {
 			klog.Warning("Resources detected in operator namespace, attempting to delete again...")
-			if err := deleteOperatorResourcesInternal(ctx, c); err != nil {
+			if err := deleteOperatorResourcesInternal(ctx, c, operatorNamespace); err != nil {
 				klog.Errorf("Failed to re-delete operator resources: %v", err)
 			}
 		}