WIP: generalize post test debug

test/e2e/framework/framework.go

@@ -39,12 +39,14 @@ type Framework struct {
 	RootCA             *x509.CertPool
 	MetricsClientCert  *tls.Certificate
 	OperatorNamespace  string
+	ClusterVersion     *configv1.ClusterVersion
 }
 
 // Setup finalizes the initilization of the Framework object by setting
 // parameters which are specific to OpenShift.
 func (f *Framework) Setup() error {
 	clusterVersion := &configv1.ClusterVersion{}
+
 	if err := f.K8sClient.Get(context.Background(), client.ObjectKey{Name: "version"}, clusterVersion); err != nil {
 		if meta.IsNoMatchError(err) {
 			return nil
@@ -53,6 +55,7 @@ func (f *Framework) Setup() error {
 		return fmt.Errorf("failed to get clusterversion %w", err)
 	}
 
+	f.ClusterVersion = clusterVersion
 	f.IsOpenshiftCluster = true
 
 	// Load the service CA operator's certificate authority.
@@ -260,19 +263,58 @@ func (f *Framework) CleanUp(t *testing.T, cleanupFunc func()) {
 	})
 }
 
+// TODO: remove ForceFailure — temporary helper to exercise DumpOnFailure logging.
+func (f *Framework) ForceFailure(t *testing.T) {
+	t.Helper()
+	t.Error("forced failure to test debug dump output")
+}
+
+// DebugFunc is a diagnostic function invoked when a test fails.
+// Implementations should use t.Logf to emit relevant state.
+type DebugFunc func(t *testing.T)
+
+// DumpOnFailure registers a t.Cleanup that runs the given debug functions
+// when the test fails. It can be called multiple times to add more debug
+// functions as resources become available during the test.
+//
+// Cleanups run in LIFO order, so call DumpOnFailure before registering
+// resource deletions via CleanUp to ensure debug info is captured while
+// the resources still exist.
+func (f *Framework) DumpOnFailure(t *testing.T, fns ...DebugFunc) {
+	t.Helper()
+	t.Cleanup(func() {
+		if !t.Failed() {
+			return
+		}
+		for _, fn := range fns {
+			fn(t)
+		}
+	})
+}
+
+// DebugNamespace returns a DebugFunc that dumps deployments, pods, and events
+// for the given namespaces.
+func (f *Framework) DebugNamespace(namespaces ...string) DebugFunc {
+	return func(t *testing.T) {
+		t.Helper()
+		for _, ns := range namespaces {
+			t.Logf("--- Dumping debug info for namespace %s ---", ns)
+			f.DumpNamespaceDebug(t, ns)
+		}
+	}
+}
+
 // SkipIfClusterVersionBelow skips the test if the cluster version is below
 // minVersion. The minVersion string should be a semver-compatible version
 // (e.g. "4.19" or "v4.19").
 func (f *Framework) SkipIfClusterVersionBelow(t *testing.T, minVersion string) {
 	t.Helper()
-	cv := &configv1.ClusterVersion{}
-	err := f.K8sClient.Get(t.Context(), client.ObjectKey{Name: "version"}, cv)
-	if err != nil {
-		t.Fatalf("failed to determine cluster version: %v", err)
+	if f.ClusterVersion == nil {
+		t.Fatal("cluster version not available (non-OpenShift cluster?)")
 		return
 	}
 
-	actual := cv.Status.Desired.Version
+	actual := f.ClusterVersion.Status.Desired.Version
 	if actual == "" {
 		t.Fatal("cluster version is empty")
 		return
@@ -295,7 +337,7 @@ func (f *Framework) SkipIfClusterVersionBelow(t *testing.T, minVersion string) {
 	}
 
 	if semver.Compare(canonicalActual, canonicalMin) < 0 {
-		t.Skipf("Skipping: cluster version %s is below minimum required %s", cv.Status.Desired.Version, minVersion)
+		t.Skipf("Skipping: cluster version %s is below minimum required %s", f.ClusterVersion.Status.Desired.Version, minVersion)
 	}
 }
 

test/e2e/monitoring_stack_controller_test.go

@@ -48,6 +48,8 @@ func assertCRDExists(t *testing.T, crds ...string) {
 }
 
 func TestMonitoringStackController(t *testing.T) {
+	f.DumpOnFailure(t, f.DebugNamespace(e2eTestNamespace))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 	err := stack.AddToScheme(scheme.Scheme)
 	assert.NilError(t, err, "adding stack to scheme failed")
 	assertCRDExists(t,

test/e2e/observability_installer_test.go

@@ -45,6 +45,8 @@ func TestObservabilityInstallerController(t *testing.T) {
 }
 
 func testObservabilityInstallerTracing(t *testing.T) {
+	f.DumpOnFailure(t, f.DebugNamespace(f.OperatorNamespace, "tracing-observability"))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 	ctx := context.Background()
 
 	// The ObservabilityInstaller installs operators via subscriptions,

test/e2e/operator_metrics_test.go

@@ -11,6 +11,8 @@ import (
 )
 
 func TestOperatorMetrics(t *testing.T) {
+	f.DumpOnFailure(t, f.DebugNamespace(f.OperatorNamespace))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 	t.Run("operator exposes metrics", func(t *testing.T) {
 		pod := f.GetOperatorPod(t)
 

test/e2e/po_admission_webhook_test.go

@@ -12,6 +12,8 @@ import (
 )
 
 func TestPrometheusRuleWebhook(t *testing.T) {
+	f.DumpOnFailure(t, f.DebugNamespace(e2eTestNamespace))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 	assertCRDExists(t,
 		"prometheusrules.monitoring.rhobs",
 	)

test/e2e/prometheus_operator_test.go

@@ -29,6 +29,8 @@ type testCase struct {
 }
 
 func TestPrometheusOperatorForNonOwnedResources(t *testing.T) {
+	f.DumpOnFailure(t, f.DebugNamespace(e2eTestNamespace))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 	resources := []client.Object{
 		newPrometheus(nil),
 		newAlertmanager(nil),
@@ -74,6 +76,8 @@ func TestPrometheusOperatorForNonOwnedResources(t *testing.T) {
 }
 
 func TestPrometheusOperatorForOwnedResources(t *testing.T) {
+	f.DumpOnFailure(t, f.DebugNamespace(e2eTestNamespace))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 	resources := []client.Object{
 		newPrometheus(ownedResourceLabels),
 		newAlertmanager(ownedResourceLabels),

test/e2e/thanos_querier_controller_test.go

@@ -24,6 +24,8 @@ import (
 )
 
 func TestThanosQuerierController(t *testing.T) {
+	f.DumpOnFailure(t, f.DebugNamespace(e2eTestNamespace))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 	assertCRDExists(t, "thanosqueriers.monitoring.rhobs")
 
 	ts := []testCase{

test/e2e/uiplugin_cluster_health_analyzer_test.go

@@ -32,21 +32,21 @@ func clusterHealthAnalyzer(t *testing.T) {
 	err := monv1.AddToScheme(f.K8sClient.Scheme())
 	assert.NilError(t, err, "failed to add monv1 to scheme")
 
+	f.DumpOnFailure(t, f.DebugNamespace(uiPluginInstallNS))
+
 	plugin := resetMonitoringUIPlugin(t)
 	err = f.K8sClient.Create(t.Context(), plugin)
 	assert.NilError(t, err, "failed to create monitoring UIPlugin")
 
-	t.Cleanup(func() {
-		if t.Failed() {
-			dumpClusterHealthAnalyzerDebug(t, plugin.Name)
-		}
-	})
+	f.DumpOnFailure(t, dumpUIPluginDebug(plugin.Name))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 
 	t.Log("Waiting for health-analyzer deployment to become ready...")
 	haDeployment := appsv1.Deployment{}
 	f.GetResourceWithRetry(t, healthAnalyzerDeploymentName, uiPluginInstallNS, &haDeployment)
 	f.AssertDeploymentReady(healthAnalyzerDeploymentName, uiPluginInstallNS, framework.WithTimeout(5*time.Minute))(t)
 
+	// Use a unique suffix so re-runs don't conflict with leftover rules from prior executions.
 	suffix := strconv.FormatInt(time.Now().UnixNano()%100000, 10)
 	ruleName := "e2e-health-analyzer-" + suffix
 	alertName := "E2EHealthAnalyzer" + suffix
@@ -179,15 +179,16 @@ func newAlwaysFiringRule(t *testing.T, ruleName, alertName string) *monv1.Promet
 	return rule
 }
 
-func dumpClusterHealthAnalyzerDebug(t *testing.T, pluginName string) {
-	t.Helper()
-	ctx := context.WithoutCancel(t.Context())
+func dumpUIPluginDebug(pluginName string) framework.DebugFunc {
+	return func(t *testing.T) {
+		t.Helper()
+		ctx := context.WithoutCancel(t.Context())
 
-	// UIPlugin-specific diagnostics
-	var plugin uiv1.UIPlugin
-	if err := f.K8sClient.Get(ctx, client.ObjectKey{Name: pluginName}, &plugin); err != nil {
-		t.Logf("Failed to get UIPlugin %q: %v", pluginName, err)
-	} else {
+		var plugin uiv1.UIPlugin
+		if err := f.K8sClient.Get(ctx, client.ObjectKey{Name: pluginName}, &plugin); err != nil {
+			t.Logf("Failed to get UIPlugin %q: %v", pluginName, err)
+			return
+		}
 		t.Logf("UIPlugin %q generation=%d, resourceVersion=%s", pluginName, plugin.Generation, plugin.ResourceVersion)
 		t.Logf("UIPlugin spec.type=%s", plugin.Spec.Type)
 		if plugin.Spec.Monitoring != nil {
@@ -204,18 +205,15 @@ func dumpClusterHealthAnalyzerDebug(t *testing.T, pluginName string) {
 		for _, c := range plugin.Status.Conditions {
 			t.Logf("UIPlugin condition: type=%s status=%s reason=%s message=%s", c.Type, c.Status, c.Reason, c.Message)
 		}
-	}
 
-	var plugins uiv1.UIPluginList
-	if err := f.K8sClient.List(ctx, &plugins); err != nil {
-		t.Logf("Failed to list UIPlugins: %v", err)
-	} else {
-		t.Logf("Total UIPlugins in cluster: %d", len(plugins.Items))
-		for _, p := range plugins.Items {
-			t.Logf("  UIPlugin: name=%s type=%s conditions=%d", p.Name, p.Spec.Type, len(p.Status.Conditions))
+		var plugins uiv1.UIPluginList
+		if err := f.K8sClient.List(ctx, &plugins); err != nil {
+			t.Logf("Failed to list UIPlugins: %v", err)
+		} else {
+			t.Logf("Total UIPlugins in cluster: %d", len(plugins.Items))
+			for _, p := range plugins.Items {
+				t.Logf("  UIPlugin: name=%s type=%s conditions=%d", p.Name, p.Spec.Type, len(p.Status.Conditions))
+			}
 		}
 	}
-
-	// Generic namespace diagnostics (deployments, pods, events)
-	f.DumpNamespaceDebug(t, uiPluginInstallNS)
 }

test/e2e/uiplugin_test.go

@@ -46,6 +46,8 @@ func TestUIPlugin(t *testing.T) {
 }
 
 func dashboardsUIPlugin(t *testing.T) {
+	f.DumpOnFailure(t, f.DebugNamespace(uiPluginInstallNS))
+	f.ForceFailure(t) // TODO: remove — temporary, exercises debug dump
 	db := newDashboardsUIPlugin(t)
 	err := f.K8sClient.Create(context.Background(), db)
 	assert.NilError(t, err, "failed to create a dashboards UIPlugin")