Fixing KMM deployment for clusters under RedHat SRE (ROSA)

Openshift cluster that are managed by Redhat SRE, do not allow(restrict)
user from changing ignitions etc'. It means that the MachineConfig CRD
is not deployed on the cluster, which causes the BMC controller to
constantly fail, since it is watching for that object.
The fix is to check the presence of the MachineConfig CRD, and to
kick-off the BMC controller only in case it is present

Author: yevgeny-shnaidman

cmd/manager/main.go

@@ -48,6 +48,8 @@ import (
 	"github.com/rh-ecosystem-edge/kernel-module-management/internal/networkpolicy"
 	"github.com/rh-ecosystem-edge/kernel-module-management/internal/nmc"
 	"github.com/rh-ecosystem-edge/kernel-module-management/internal/syncronizedmap"
+	"k8s.io/apimachinery/pkg/api/meta"
+	runtimescheme "k8s.io/apimachinery/pkg/runtime/schema"
 
 	"k8s.io/apimachinery/pkg/runtime"
 	"k8s.io/apimachinery/pkg/types"
@@ -199,8 +201,17 @@ func main() {
 		cmd.FatalError(setupLogger, err, "unable to create controller", "name", controllers.MICReconcilerName)
 	}
 
-	if err = controllers.NewBMCReconciler(client, mcfgAPI, scheme).SetupWithManager(mgr); err != nil {
-		cmd.FatalError(setupLogger, err, "unable to create controller", "name", controllers.BMCReconcilerName)
+	bmcSupported, err := isBMCSupported(mgr.GetRESTMapper())
+	if err != nil {
+		cmd.FatalError(setupLogger, err, "failed to check presence of CRDs needed for BMC controller")
+	}
+
+	if bmcSupported {
+		if err = controllers.NewBMCReconciler(client, mcfgAPI, scheme).SetupWithManager(mgr); err != nil {
+			cmd.FatalError(setupLogger, err, "unable to create controller", "name", controllers.BMCReconcilerName)
+		}
+	} else {
+		setupLogger.Info("BMC controller is not running on the cluster, since MachineConfig CRD is not provided by MCO")
 	}
 
 	if managed {
@@ -282,3 +293,22 @@ func GetBoolEnv(s string) (bool, error) {
 
 	return managed, nil
 }
+
+func isBMCSupported(mapper meta.RESTMapper) (bool, error) {
+	gk := runtimescheme.GroupKind{
+		Group: "machineconfiguration.openshift.io",
+		Kind:  "MachineConfig",
+	}
+
+	_, err := mapper.RESTMapping(gk, "v1")
+
+	if err != nil {
+		if meta.IsNoMatchError(err) {
+			return false, nil
+		}
+		// Any other error (network, etc.) should be treated as a failure
+		return false, fmt.Errorf("failed to get the MC CRD due to internal error: %v", err)
+	}
+
+	return true, nil
+}