feat: record events when a feature fails and use the shared event reasons

controllers/clustersummary_controller.go

@@ -57,6 +57,7 @@ import (
 	libsveltosv1beta1 "github.com/projectsveltos/libsveltos/api/v1beta1"
 	"github.com/projectsveltos/libsveltos/lib/clusterproxy"
 	"github.com/projectsveltos/libsveltos/lib/deployer"
+	"github.com/projectsveltos/libsveltos/lib/eventreasons"
 	license "github.com/projectsveltos/libsveltos/lib/licenses"
 	logs "github.com/projectsveltos/libsveltos/lib/logsettings"
 	predicates "github.com/projectsveltos/libsveltos/lib/predicates"
@@ -607,23 +608,13 @@ func (r *ClusterSummaryReconciler) prepareForDeployment(ctx context.Context,
 func (r *ClusterSummaryReconciler) proceedDeployingClusterSummary(ctx context.Context,
 	clusterSummaryScope *scope.ClusterSummaryScope, logger logr.Logger) (reconcile.Result, error) {
 
-	// Snapshot existing failure messages before deploying so we can detect
-	// new conflicts vs. ongoing ones and avoid re-raising an event every retry.
-	preDeployFailures := collectFailureMessages(clusterSummaryScope.ClusterSummary)
-
 	err := r.deploy(ctx, clusterSummaryScope, logger)
 	if err != nil {
 		var conflictErr *deployer.ConflictError
 		ok := errors.As(err, &conflictErr)
 		if ok {
+			// The ProfileConflict event is recorded by updateFeatureStatus, when the conflict is persisted
 			logger.V(logs.LogInfo).Error(err, "failed to deploy because of conflict")
-			if _, alreadyKnown := preDeployFailures[conflictErr.Error()]; !alreadyKnown {
-				clusterSummary := clusterSummaryScope.ClusterSummary
-				r.eventRecorder.Eventf(clusterSummary, nil, corev1.EventTypeWarning, "Conflict",
-					configv1beta1.ClusterSummaryKind, "Conflict detected for cluster %s %s/%s: %s",
-					clusterSummary.Spec.ClusterType, clusterSummary.Spec.ClusterNamespace,
-					clusterSummary.Spec.ClusterName, conflictErr.Error())
-			}
 			r.setNextReconcileTime(clusterSummaryScope, r.ConflictRetryTime)
 			return reconcile.Result{RequeueAfter: r.ConflictRetryTime}, nil
 		}
@@ -2143,14 +2134,23 @@ func grantsPullModeEligibility(result *license.LicenseVerificationResult,
 func (r *ClusterSummaryReconciler) updateStatusWithMissingLicenseError(
 	clusterSummaryScope *scope.ClusterSummaryScope, logger logr.Logger) {
 
-	notEligibleError := errors.New("license is required to manage SveltosCluster in pull mode")
+	// All features fail for the same reason: one event, recorded the first time
+	_, previousMessage := getFeatureFailure(clusterSummaryScope.ClusterSummary, libsveltosv1beta1.FeatureHelm)
+
 	failed := libsveltosv1beta1.FeatureStatusFailed
 	r.updateFeatureStatus(clusterSummaryScope, libsveltosv1beta1.FeatureHelm, &failed, nil,
-		notEligibleError, logger)
+		errPullModeLicenseRequired, logger)
 	r.updateFeatureStatus(clusterSummaryScope, libsveltosv1beta1.FeatureKustomize, &failed, nil,
-		notEligibleError, logger)
+		errPullModeLicenseRequired, logger)
 	r.updateFeatureStatus(clusterSummaryScope, libsveltosv1beta1.FeatureResources, &failed, nil,
-		notEligibleError, logger)
+		errPullModeLicenseRequired, logger)
+
+	if previousMessage == nil || *previousMessage != errPullModeLicenseRequired.Error() {
+		clusterSummary := clusterSummaryScope.ClusterSummary
+		recordEventf(r.eventRecorder, clusterSummary, nil, corev1.EventTypeWarning, eventreasons.LicenseInvalid,
+			eventreasons.ActionDeploying, "Cluster %s %s/%s: %s", clusterSummary.Spec.ClusterType,
+			clusterSummary.Spec.ClusterNamespace, clusterSummary.Spec.ClusterName, errPullModeLicenseRequired.Error())
+	}
 }
 
 // removeFinalizer removes the finalizer
@@ -2272,16 +2272,3 @@ func getClusterSummaryWithInstantiatedCharts(ctx context.Context, cs *configv1be
 
 	return &csCopy, nil
 }
-
-// collectFailureMessages returns the set of FailureMessages currently recorded
-// across all FeatureSummaries. Used to detect whether a conflict is new (not
-// yet in the status) versus ongoing (already reported in a previous reconcile).
-func collectFailureMessages(cs *configv1beta1.ClusterSummary) map[string]struct{} {
-	msgs := make(map[string]struct{}, len(cs.Status.FeatureSummaries))
-	for i := range cs.Status.FeatureSummaries {
-		if msg := cs.Status.FeatureSummaries[i].FailureMessage; msg != nil {
-			msgs[*msg] = struct{}{}
-		}
-	}
-	return msgs
-}

controllers/clustersummary_deployer.go

@@ -37,6 +37,7 @@ import (
 	"github.com/projectsveltos/libsveltos/lib/clustercache"
 	"github.com/projectsveltos/libsveltos/lib/clusterproxy"
 	"github.com/projectsveltos/libsveltos/lib/deployer"
+	"github.com/projectsveltos/libsveltos/lib/eventreasons"
 	logs "github.com/projectsveltos/libsveltos/lib/logsettings"
 	"github.com/projectsveltos/libsveltos/lib/pullmode"
 )
@@ -179,8 +180,8 @@ func (r *ClusterSummaryReconciler) proceedDeployingFeature(ctx context.Context,
 				clusterSummary.Spec.ClusterType, clusterSummary.Spec.ClusterNamespace, clusterSummary.Spec.ClusterName)
 			cluster, _ := clusterproxy.GetCluster(ctx, getManagementClusterClient(),
 				clusterSummary.Spec.ClusterNamespace, clusterSummary.Spec.ClusterName, clusterSummary.Spec.ClusterType)
-			r.eventRecorder.Eventf(clusterSummary, cluster, corev1.EventTypeNormal, "Sveltos",
-				configv1beta1.ClusterSummaryKind, message)
+			recordEventf(r.eventRecorder, clusterSummary, cluster, corev1.EventTypeNormal, eventreasons.FeatureDeployed,
+				featureAction(eventreasons.ActionDeploying, f.id), "%s", message)
 			return nil
 		}
 
@@ -251,8 +252,8 @@ func (r *ClusterSummaryReconciler) handleDeployerError(deployerError error, clus
 	if errors.As(deployerError, &nonRetriableError) {
 		nonRetriableStatus := libsveltosv1beta1.FeatureStatusFailedNonRetriable
 		r.updateFeatureStatus(clusterSummaryScope, f.id, &nonRetriableStatus, currentHash, deployerError, logger)
-		r.eventRecorder.Eventf(clusterSummary, nil, corev1.EventTypeWarning, "FailedNonRetriable",
-			configv1beta1.ClusterSummaryKind,
+		recordEventf(r.eventRecorder, clusterSummary, nil, corev1.EventTypeWarning, eventreasons.FeatureFailedNonRetriable,
+			featureAction(eventreasons.ActionDeploying, f.id),
 			"Feature %s for cluster %s %s/%s failed with non-retriable error: %s",
 			f.id, clusterSummary.Spec.ClusterType, clusterSummary.Spec.ClusterNamespace,
 			clusterSummary.Spec.ClusterName, deployerError.Error())
@@ -262,8 +263,8 @@ func (r *ClusterSummaryReconciler) handleDeployerError(deployerError error, clus
 	if errors.As(deployerError, &templateError) {
 		nonRetriableStatus := libsveltosv1beta1.FeatureStatusFailedNonRetriable
 		r.updateFeatureStatus(clusterSummaryScope, f.id, &nonRetriableStatus, currentHash, deployerError, logger)
-		r.eventRecorder.Eventf(clusterSummary, nil, corev1.EventTypeWarning, "FailedNonRetriable",
-			configv1beta1.ClusterSummaryKind,
+		recordEventf(r.eventRecorder, clusterSummary, nil, corev1.EventTypeWarning, eventreasons.FeatureFailedNonRetriable,
+			featureAction(eventreasons.ActionDeploying, f.id),
 			"Feature %s for cluster %s %s/%s failed to instantiate template: %s",
 			f.id, clusterSummary.Spec.ClusterType, clusterSummary.Spec.ClusterNamespace,
 			clusterSummary.Spec.ClusterName, deployerError.Error())
@@ -327,8 +328,8 @@ func (r *ClusterSummaryReconciler) handlePullModeProvisioned(ctx context.Context
 		clusterSummary.Spec.ClusterNamespace, clusterSummary.Spec.ClusterName, clusterSummary.Spec.ClusterType)
 	message := fmt.Sprintf("Feature: %s deployed to cluster %s %s/%s", f.id,
 		clusterSummary.Spec.ClusterType, clusterSummary.Spec.ClusterNamespace, clusterSummary.Spec.ClusterName)
-	r.eventRecorder.Eventf(clusterSummary, cluster, corev1.EventTypeNormal, "Sveltos",
-		configv1beta1.ClusterSummaryKind, message)
+	recordEventf(r.eventRecorder, clusterSummary, cluster, corev1.EventTypeNormal, eventreasons.FeatureDeployed,
+		featureAction(eventreasons.ActionDeploying, f.id), "%s", message)
 	r.updateFeatureStatus(clusterSummaryScope, f.id, pullmodeProvisionedStatus(), currentHash, nil, logger)
 	clusterSummaryScope.SetFailureMessage(f.id, nil)
 	now := metav1.NewTime(time.Now())
@@ -610,10 +611,15 @@ func (r *ClusterSummaryReconciler) processUndeployResult(ctx context.Context, cl
 func (r *ClusterSummaryReconciler) markUndeployFailureRetriable(clusterSummaryScope *scope.ClusterSummaryScope,
 	featureID libsveltosv1beta1.FeatureID, failureErr error, logger logr.Logger) {
 
+	// What the ClusterSummary recorded for this feature before this update tells whether this failure is new
+	_, previousMessage := getFeatureFailure(clusterSummaryScope.ClusterSummary, featureID)
+
 	removing := libsveltosv1beta1.FeatureStatusRemoving
 	r.updateFeatureStatus(clusterSummaryScope, featureID, &removing, nil, failureErr, logger)
 	errorMsg := failureErr.Error()
 	clusterSummaryScope.SetFailureMessage(featureID, &errorMsg)
+	r.recordFeatureFailure(clusterSummaryScope.ClusterSummary, featureID, eventreasons.ActionRemoving,
+		previousMessage, failureErr)
 }
 
 func cleanPreviousActiveTracking(ctx context.Context, clusterSummary *configv1beta1.ClusterSummary, f feature,
@@ -956,11 +962,16 @@ func (r *ClusterSummaryReconciler) updateFeatureStatus(clusterSummaryScope *scop
 		// non-terminal statuses above in case that ever changes.
 		clusterSummaryScope.SetFeatureStatus(featureID, libsveltosv1beta1.FeatureStatusBlocked, hash, nil)
 	case libsveltosv1beta1.FeatureStatusFailed, libsveltosv1beta1.FeatureStatusFailedNonRetriable:
+		// What the ClusterSummary recorded for this feature before this update tells whether this failure is new
+		previousStatus, previousMessage := getFeatureFailure(clusterSummary, featureID)
 		failed := true
 		clusterSummaryScope.SetFeatureStatus(featureID, *status, hash, &failed)
 		if statusError != nil {
 			err := statusError.Error()
 			clusterSummaryScope.SetFailureMessage(featureID, &err)
+			if *status == libsveltosv1beta1.FeatureStatusFailed {
+				r.recordFailedStatus(clusterSummary, featureID, previousStatus, previousMessage, statusError)
+			}
 		}
 		clusterSummaryScope.SetLastAppliedTime(featureID, &now)
 		trackReconcileOutcome(clusterSummary.Spec.ClusterNamespace, clusterSummary.Spec.ClusterName,
@@ -1087,8 +1098,8 @@ func (r *ClusterSummaryReconciler) markMaxConsecutiveFailuresReached(clusterSumm
 	nonRetriableStatus := libsveltosv1beta1.FeatureStatusFailedNonRetriable
 	resultError := errors.New("the maximum number of consecutive errors has been reached")
 	r.updateFeatureStatus(clusterSummaryScope, f.id, &nonRetriableStatus, currentHash, resultError, logger)
-	r.eventRecorder.Eventf(clusterSummary, nil, corev1.EventTypeWarning, "FailedNonRetriable",
-		configv1beta1.ClusterSummaryKind,
+	recordEventf(r.eventRecorder, clusterSummary, nil, corev1.EventTypeWarning, eventreasons.FeatureFailedNonRetriable,
+		featureAction(eventreasons.ActionDeploying, f.id),
 		"Feature %s for cluster %s %s/%s will no longer be retried: maximum consecutive failures reached",
 		f.id, clusterSummary.Spec.ClusterType, clusterSummary.Spec.ClusterNamespace,
 		clusterSummary.Spec.ClusterName)

controllers/events.go

@@ -0,0 +1,169 @@
+/*
+Copyright 2026. projectsveltos.io. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package controllers
+
+import (
+	"errors"
+	"fmt"
+
+	corev1 "k8s.io/api/core/v1"
+	"k8s.io/apimachinery/pkg/runtime"
+	"k8s.io/client-go/tools/events"
+
+	configv1beta1 "github.com/projectsveltos/addon-controller/api/v1beta1"
+	libsveltosv1beta1 "github.com/projectsveltos/libsveltos/api/v1beta1"
+	"github.com/projectsveltos/libsveltos/lib/deployer"
+	"github.com/projectsveltos/libsveltos/lib/eventreasons"
+)
+
+// recordEventf records an event on regarding, about the related object (can be nil).
+// It does nothing when no recorder is set.
+// The note is capped, so an oversized message is not rejected by the API server.
+func recordEventf(recorder events.EventRecorder, regarding, related runtime.Object, eventType, reason, action,
+	format string, args ...any) {
+
+	if recorder == nil {
+		return
+	}
+
+	note := eventreasons.TruncateNote(fmt.Sprintf(format, args...))
+	recorder.Eventf(regarding, related, eventType, reason, action, "%s", note)
+}
+
+// featureAction returns the action of an event about a feature: the verb followed by the feature, for
+// instance DeployingHelm. The events recorder merges events with the same reason, action and regarding
+// object and keeps only the first note, so an action which is the same for every feature would make
+// the failures of two features of the same ClusterSummary a single event.
+func featureAction(action string, featureID libsveltosv1beta1.FeatureID) string {
+	return action + string(featureID)
+}
+
+// getActionForFeatureStatus returns the action an event refers to given the status the feature had before:
+// a feature being removed is Removing, anything else is Deploying.
+func getActionForFeatureStatus(previousStatus *libsveltosv1beta1.FeatureStatus) string {
+	if previousStatus != nil &&
+		(*previousStatus == libsveltosv1beta1.FeatureStatusRemoving ||
+			*previousStatus == libsveltosv1beta1.FeatureStatusAgentRemoving) {
+
+		return eventreasons.ActionRemoving
+	}
+
+	return eventreasons.ActionDeploying
+}
+
+var (
+	// errPullModeLicenseRequired is the failure of a ClusterSummary targeting a SveltosCluster in pull mode which
+	// is not covered by a license.
+	errPullModeLicenseRequired = errors.New("license is required to manage SveltosCluster in pull mode")
+)
+
+// hasOwnEvent returns true for the errors which are already reported by their own event: a failure which
+// will not be retried (FeatureFailedNonRetriable) and a missing license (LicenseInvalid).
+// FeatureFailed is not recorded for those, or the same failure would be reported twice.
+func hasOwnEvent(err error) bool {
+	var nonRetriableError *configv1beta1.NonRetriableError
+	var templateError *configv1beta1.TemplateInstantiationError
+
+	return errors.As(err, &nonRetriableError) ||
+		errors.As(err, &templateError) ||
+		errors.Is(err, errPullModeLicenseRequired)
+}
+
+// isConflictError returns true when err is, or wraps, a conflict with another profile.
+func isConflictError(err error) bool {
+	var conflictError *deployer.ConflictError
+
+	return errors.As(err, &conflictError)
+}
+
+// recordFailedStatus records the event for a feature which has just been set to Failed with failure: a
+// ProfileConflict for a conflict with another profile, a FeatureFailed for any other failure which is
+// retried. The failures which have their own event (hasOwnEvent) are not recorded here.
+// previousStatus and previousMessage are what the ClusterSummary recorded before this update.
+func (r *ClusterSummaryReconciler) recordFailedStatus(clusterSummary *configv1beta1.ClusterSummary,
+	featureID libsveltosv1beta1.FeatureID, previousStatus *libsveltosv1beta1.FeatureStatus, previousMessage *string,
+	failure error) {
+
+	switch {
+	case isConflictError(failure):
+		r.recordProfileConflict(clusterSummary, previousMessage, failure)
+	case !hasOwnEvent(failure):
+		r.recordFeatureFailure(clusterSummary, featureID, getActionForFeatureStatus(previousStatus),
+			previousMessage, failure)
+	}
+}
+
+// recordProfileConflict records a Warning ProfileConflict event: a resource this ClusterSummary wants to deploy
+// is already deployed by another profile. Nothing is recorded when the ClusterSummary already failed with the
+// same message: a retry which hits the same conflict is not news.
+func (r *ClusterSummaryReconciler) recordProfileConflict(clusterSummary *configv1beta1.ClusterSummary,
+	previousMessage *string, conflict error) {
+
+	message := conflict.Error()
+	if previousMessage != nil && *previousMessage == message {
+		return
+	}
+
+	recordEventf(r.eventRecorder, clusterSummary, nil, corev1.EventTypeWarning, eventreasons.ProfileConflict,
+		eventreasons.ActionDeploying, "Conflict detected for cluster %s %s/%s: %s",
+		clusterSummary.Spec.ClusterType, clusterSummary.Spec.ClusterNamespace,
+		clusterSummary.Spec.ClusterName, message)
+}
+
+// getFeatureFailure returns the status and the failure message the ClusterSummary currently records for
+// the feature. Both are nil when the feature has no entry yet.
+// Call it before updating the feature status: the failure message of the previous attempt is what tells
+// whether a failure is new.
+func getFeatureFailure(clusterSummary *configv1beta1.ClusterSummary, featureID libsveltosv1beta1.FeatureID,
+) (status *libsveltosv1beta1.FeatureStatus, message *string) {
+
+	featureSummary := getFeatureSummaryForFeatureID(clusterSummary, featureID)
+	if featureSummary == nil {
+		return nil, nil
+	}
+
+	currentStatus := featureSummary.Status
+	status = &currentStatus
+	if featureSummary.FailureMessage != nil {
+		currentMessage := *featureSummary.FailureMessage
+		message = &currentMessage
+	}
+
+	return status, message
+}
+
+// recordFeatureFailure records a Warning FeatureFailed event when a feature fails to be deployed, or removed,
+// and the operation is going to be retried. Nothing is recorded when the feature already failed with the same
+// message (previousMessage): a retry which fails the same way is not news. A different message is recorded.
+// The failure message is cleared when the feature succeeds, so a failure after a success is recorded again.
+func (r *ClusterSummaryReconciler) recordFeatureFailure(clusterSummary *configv1beta1.ClusterSummary,
+	featureID libsveltosv1beta1.FeatureID, action string, previousMessage *string, failure error) {
+
+	if failure == nil {
+		return
+	}
+
+	message := failure.Error()
+	if previousMessage != nil && *previousMessage == message {
+		return
+	}
+
+	recordEventf(r.eventRecorder, clusterSummary, nil, corev1.EventTypeWarning, eventreasons.FeatureFailed,
+		featureAction(action, featureID), "Feature %s for cluster %s %s/%s failed: %s",
+		featureID, clusterSummary.Spec.ClusterType, clusterSummary.Spec.ClusterNamespace,
+		clusterSummary.Spec.ClusterName, message)
+}