Implement option to disable node taints

cmd/agent/main.go

@@ -22,7 +22,7 @@ func NewAgentCommand() *cobra.Command {
 	var drbdStatusInterval, reconcileInterval, resyncInterval, operationsTimeout, failOverTimeout time.Duration
 	var deletionGraceSeconds int64
 	var nodeName, healthzBindAddress, satellitePodLabel string
-	var failOverUnsafePods bool
+	var failOverUnsafePods, disableNodeTaints bool
 
 	cmd := &cobra.Command{
 		Use:     "node-agent",
@@ -52,6 +52,7 @@ func NewAgentCommand() *cobra.Command {
 				NodeName:             nodeName,
 				FailOverUnsafePods:   failOverUnsafePods,
 				SatellitePodSelector: selector,
+				DisableNodeTaints:    disableNodeTaints,
 			})
 			if err != nil {
 				return err
@@ -90,6 +91,7 @@ func NewAgentCommand() *cobra.Command {
 	cmd.Flags().StringVar(&healthzBindAddress, "health-bind-address", ":8000", "the address to bind to for the /healthz endpoint")
 	cmd.Flags().StringVar(&satellitePodLabel, "satellite-pod-label", "app.kubernetes.io/component=linstor-satellite", "the connection name reported by DRBD is one of the Pods with this label")
 	cmd.Flags().BoolVar(&failOverUnsafePods, "fail-over-unsafe-pods", false, "fail over Pods that use storage with unknown fail over properties")
+	cmd.Flags().BoolVar(&disableNodeTaints, "disable-node-taints", false, "prevent nodes from being tainted")
 	return cmd
 }
 

pkg/agent/agent.go

@@ -54,6 +54,8 @@ type Options struct {
 	// SatellitePodSelector selects the Pods that should be considered LINSTOR Satellites.
 	// If the DRBD connection name matches on of these Pods, the Kubernetes Node name is taken from these Pods.
 	SatellitePodSelector labels.Selector
+	// DisableNodeTaints prevents the nodes in a cluster from being tainted.
+	DisableNodeTaints bool
 }
 
 // Timeout returns the operations timeout.
@@ -567,8 +569,8 @@ func hasPersistentVolumeClaimRef(pv *corev1.PersistentVolume) bool {
 // TaintNode adds the specific taint to the node.
 //
 // Returns false, nil if the taint was already present.
-func TaintNode(ctx context.Context, client kubernetes.Interface, node *corev1.Node, taint corev1.Taint) (bool, error) {
-	if node == nil {
+func TaintNode(ctx context.Context, client kubernetes.Interface, node *corev1.Node, taint corev1.Taint, disableNodeTaints bool) (bool, error) {
+	if disableNodeTaints || node == nil {
 		return false, nil
 	}
 

pkg/agent/reconcile_failover.go

@@ -148,7 +148,7 @@ func (f *failoverReconciler) evictPods(ctx context.Context, res *DrbdResourceSta
 		Key:       metadata.NodeLostQuorumTaint,
 		Effect:    corev1.TaintEffectNoSchedule,
 		TimeAdded: &taintTime,
-	})
+	}, f.opt.DisableNodeTaints)
 	if err != nil {
 		return err
 	}

pkg/agent/reconcile_force_io_error_pods.go

@@ -65,7 +65,7 @@ func (f *forceIoErrorReconciler) RunForResource(ctx context.Context, req *Reconc
 		Key:       metadata.NodeForceIoErrorTaint,
 		Effect:    corev1.TaintEffectNoSchedule,
 		TimeAdded: &taintTime,
-	})
+	}, f.opt.DisableNodeTaints)
 	if err != nil {
 		return err
 	}