chore: handle correctly the kata enabled

Author: mchmarny

distros/kubernetes/nvsentinel/charts/syslog-health-monitor/templates/configmap.yaml

@@ -33,7 +33,7 @@ data:
   # 1. Kata detection happens at the deployment level via node labels
   # 2. Each DaemonSet variant (regular/kata) handles journal access differently:
   #    - Regular: mounts /var/log (file-based logs)
-  #    - Kata: mounts /run/log/journal and /var/log/journal (systemd journal)
+  #    - Kata: mounts /run/systemd/journal and /var/log/journal (systemd journal)
   # 3. The kata variant accesses the systemd journal directly, so it can read
   #    containerd logs without needing special tags or filters in the ConfigMap
   log_check_definitions.yaml: |

health-monitors/syslog-health-monitor/main.go

@@ -27,6 +27,7 @@ import (
 
 	"github.com/nvidia/nvsentinel/commons/pkg/logger"
 	"github.com/nvidia/nvsentinel/commons/pkg/server"
+	"github.com/nvidia/nvsentinel/commons/pkg/stringutil"
 	pb "github.com/nvidia/nvsentinel/data-models/pkg/protos"
 	fd "github.com/nvidia/nvsentinel/health-monitors/syslog-health-monitor/pkg/syslog-monitor"
 	"golang.org/x/sync/errgroup"
@@ -131,14 +132,27 @@ func run() error {
 		return fmt.Errorf("no checks defined in the config file")
 	}
 
-	// Add kata-specific journal tags if running in Kata mode
-	if *kataEnabled == "true" {
-		slog.Info("Kata mode enabled, adding containerd service filter to journal checks")
+	// Handle kata-specific configuration
+	if stringutil.IsTruthyValue(*kataEnabled) {
+		slog.Info("Kata mode enabled, adding containerd service filter and removing SysLogsSXIDError check")
 
+		// Add containerd service filter to all checks for kata nodes
 		for i := range config.Checks {
-			// Add "-u containerd.service" tag to filter for containerd logs in systemd journal
-			config.Checks[i].Tags = append(config.Checks[i].Tags, "-u", "containerd.service")
+			if config.Checks[i].Tags == nil {
+				config.Checks[i].Tags = []string{"-u", "containerd.service"}
+			} else {
+				config.Checks[i].Tags = append(config.Checks[i].Tags, "-u", "containerd.service")
+			}
+		}
+
+		// Remove SysLogsSXIDError check for kata nodes (not supported in kata environment)
+		filteredChecks := make([]fd.CheckDefinition, 0, len(config.Checks))
+		for _, check := range config.Checks {
+			if check.Name != "SysLogsSXIDError" {
+				filteredChecks = append(filteredChecks, check)
+			}
 		}
+		config.Checks = filteredChecks
 	}
 
 	slog.Info("Creating syslog monitor", "checksCount", len(config.Checks))

labeler-module/pkg/labeler/labeler.go

@@ -309,6 +309,7 @@ func (l *Labeler) getKataLabelForNode(node *v1.Node) string {
 
 // isKataEnabled checks if a node has Kata Containers enabled by examining node labels.
 // Checks the configured kata labels (either custom override or default) for truthy values.
+// Returns true if ANY of the configured labels has a truthy value (OR logic).
 // Truthy values are: "true", "enabled", "1", "yes" (case-insensitive).
 func isKataEnabled(node *v1.Node, kataLabels []string) bool {
 	for _, label := range kataLabels {
@@ -475,13 +476,8 @@ func (l *Labeler) updateKataLabel(nodeName, expectedKataLabel string) error {
 			return nil
 		}
 
-		if expectedKataLabel == "" {
-			delete(node.Labels, KataEnabledLabel)
-			slog.Info("Removing Kata enabled label from node", "node", nodeName)
-		} else {
-			node.Labels[KataEnabledLabel] = expectedKataLabel
-			slog.Info("Setting Kata enabled label on node", "node", nodeName, "kata", expectedKataLabel)
-		}
+		node.Labels[KataEnabledLabel] = expectedKataLabel
+		slog.Info("Setting Kata enabled label on node", "node", nodeName, "kata", expectedKataLabel)
 
 		_, err = l.clientset.CoreV1().Nodes().Update(l.ctx, node, metav1.UpdateOptions{})