fix: use SNO check to disable node-removal controller

node-removal controller wouldn't be necessary at the end of the lifecycle of SNO nodes as the entire cluster would go down

Signed-off-by: jakobmoellerdev <[email protected]>

main.go

@@ -90,7 +90,13 @@ func main() {
 	}
 	setupLog.Info("Watching namespace", "Namespace", operatorNamespace)
 
-	leaderElectionResolver, err := cluster.NewLeaderElectionResolver(ctrl.GetConfigOrDie(), scheme, enableLeaderElection, operatorNamespace)
+	setupClient, err := client.New(ctrl.GetConfigOrDie(), client.Options{Scheme: scheme})
+	if err != nil {
+		setupLog.Error(err, "unable to initialize setup client for pre-manager startup checks")
+		os.Exit(1)
+	}
+	snoCheck := cluster.NewMasterSNOCheck(setupClient)
+	leaderElectionResolver, err := cluster.NewLeaderElectionResolver(snoCheck, enableLeaderElection, operatorNamespace)
 	if err != nil {
 		setupLog.Error(err, "unable to setup leader election")
 		os.Exit(1)
@@ -130,12 +136,15 @@ func main() {
 		setupLog.Error(err, "unable to create controller", "controller", "LVMCluster")
 		os.Exit(1)
 	}
-	// TODO only run when not in SNO
-	if err = (&controllers.NodeRemovalController{
-		Client: mgr.GetClient(),
-	}).SetupWithManager(mgr); err != nil {
-		setupLog.Error(err, "unable to create controller", "controller", "NodeRemovalControlelr")
-		os.Exit(1)
+
+	if !snoCheck.IsSNO(context.Background()) {
+		setupLog.Info("starting node-removal controller to observe node removal in MultiNode")
+		if err = (&controllers.NodeRemovalController{
+			Client: mgr.GetClient(),
+		}).SetupWithManager(mgr); err != nil {
+			setupLog.Error(err, "unable to create controller", "controller", "NodeRemovalControlelr")
+			os.Exit(1)
+		}
 	}
 
 	if err = mgr.GetFieldIndexer().IndexField(context.Background(), &lvmv1alpha1.LVMVolumeGroupNodeStatus{}, "metadata.name", func(object client.Object) []string {

pkg/cluster/leaderelection.go

@@ -2,14 +2,8 @@ package cluster
 
 import (
 	"context"
-	"fmt"
 	configv1 "github.com/openshift/api/config/v1"
 	"github.com/openshift/library-go/pkg/config/leaderelection"
-	corev1 "k8s.io/api/core/v1"
-	"k8s.io/apimachinery/pkg/runtime"
-	"k8s.io/client-go/rest"
-	"os"
-	"sigs.k8s.io/controller-runtime/pkg/client"
 	log "sigs.k8s.io/controller-runtime/pkg/log"
 )
 
@@ -25,45 +19,32 @@ type LeaderElectionResolver interface {
 // on the amount of master nodes discovered in the cluster. If there is exactly one control-plane/master node,
 // the returned LeaderElectionResolver settings are optimized for SNO deployments.
 func NewLeaderElectionResolver(
-	config *rest.Config,
-	scheme *runtime.Scheme,
+	snoCheck SNOCheck,
 	enableLeaderElection bool,
 	operatorNamespace string,
 ) (LeaderElectionResolver, error) {
-	leaderElectionClient, err := client.New(config, client.Options{Scheme: scheme})
-	if err != nil {
-		return nil, fmt.Errorf("cannot create leader election client: %w", err)
-	}
-
 	defaultElectionConfig := leaderelection.LeaderElectionDefaulting(configv1.LeaderElection{
 		Disable: !enableLeaderElection,
 	}, operatorNamespace, "1136b8a6.topolvm.io")
 
 	return &nodeLookupSNOLeaderElection{
-		clnt:                  leaderElectionClient,
+		snoCheck:              snoCheck,
 		defaultElectionConfig: defaultElectionConfig,
 	}, nil
 }
 
 type nodeLookupSNOLeaderElection struct {
-	clnt                  client.Client
+	snoCheck              SNOCheck
 	defaultElectionConfig configv1.LeaderElection
 }
 
 func (le *nodeLookupSNOLeaderElection) Resolve(ctx context.Context) (configv1.LeaderElection, error) {
 	logger := log.FromContext(ctx)
-	nodes := &corev1.NodeList{}
-	if err := le.clnt.List(context.Background(), nodes, client.MatchingLabels{
-		ControlPlaneIDLabel: "",
-	}); err != nil {
-		logger.Error(err, "unable to retrieve nodes for SNO check with lease configuration")
-		os.Exit(1)
-	}
-	if len(nodes.Items) != 1 {
+	if !le.snoCheck.IsSNO(ctx) {
+		logger.Info("Using default Multi-Node leader election settings optimized for high-availability")
 		return le.defaultElectionConfig, nil
 	}
-	logger.Info("Overwriting defaults with SNO leader election config as only a single node was discovered",
-		"node", nodes.Items[0].GetName())
+	logger.Info("Overwriting defaults with SNO leader election config as only a single node was discovered")
 	config := leaderelection.LeaderElectionSNOConfig(le.defaultElectionConfig)
 	logger.Info("leader election config setup succeeded",
 		"retry-period", config.RetryPeriod,

pkg/cluster/leaderelection_test.go

@@ -84,7 +84,7 @@ func Test_nodeLookupSNOLeaderElection_Resolve(t *testing.T) {
 		t.Run(tt.name, func(t *testing.T) {
 			clnt := fake.NewClientBuilder().WithObjects(tt.nodes...).Build()
 			le := &nodeLookupSNOLeaderElection{
-				clnt: clnt,
+				snoCheck: NewMasterSNOCheck(clnt),
 				defaultElectionConfig: leaderelection.LeaderElectionDefaulting(configv1.LeaderElection{},
 					"test", "test-leader-id"),
 			}

pkg/cluster/sno.go

@@ -0,0 +1,33 @@
+package cluster
+
+import (
+	"context"
+	corev1 "k8s.io/api/core/v1"
+	"os"
+	"sigs.k8s.io/controller-runtime/pkg/client"
+	"sigs.k8s.io/controller-runtime/pkg/log"
+)
+
+type SNOCheck interface {
+	IsSNO(ctx context.Context) bool
+}
+
+func NewMasterSNOCheck(clnt client.Client) SNOCheck {
+	return &masterSNOCheck{clnt: clnt}
+}
+
+type masterSNOCheck struct {
+	clnt client.Client
+}
+
+func (chk *masterSNOCheck) IsSNO(ctx context.Context) bool {
+	logger := log.FromContext(ctx)
+	nodes := &corev1.NodeList{}
+	if err := chk.clnt.List(context.Background(), nodes, client.MatchingLabels{
+		ControlPlaneIDLabel: "",
+	}); err != nil {
+		logger.Error(err, "unable to retrieve nodes for SNO check with lease configuration")
+		os.Exit(1)
+	}
+	return nodes.Items != nil && len(nodes.Items) == 1
+}