fix: autodetect Volcano installation (#3956)

Signed-off-by: Julien Mancuso <jmancuso@nvidia.com>

deploy/cloud/operator/cmd/main.go

@@ -295,6 +295,9 @@ func main() {
 		mgrOpts.Cache.DefaultNamespaces = map[string]cache.Config{
 			restrictedNamespace: {},
 		}
+		setupLog.Info("Restricted namespace configured, launching in restricted mode", "namespace", restrictedNamespace)
+	} else {
+		setupLog.Info("No restricted namespace configured, launching in cluster-wide mode")
 	}
 	mgr, err := ctrl.NewManager(ctrl.GetConfigOrDie(), mgrOpts)
 	if err != nil {
@@ -308,13 +311,22 @@ func main() {
 	ctrlConfig.Grove.Enabled = groveEnabled
 	setupLog.Info("Detecting LWS availability...")
 	lwsEnabled := commonController.DetectLWSAvailability(mainCtx, mgr)
-	ctrlConfig.LWS.Enabled = lwsEnabled
-
+	setupLog.Info("Detecting Volcano availability...")
+	volcanoEnabled := commonController.DetectVolcanoAvailability(mainCtx, mgr)
+	// LWS for multinode deployment usage depends on both LWS and Volcano availability
+	ctrlConfig.LWS.Enabled = lwsEnabled && volcanoEnabled
 	// Detect Kai-scheduler availability using discovery client
 	setupLog.Info("Detecting Kai-scheduler availability...")
 	kaiSchedulerEnabled := commonController.DetectKaiSchedulerAvailability(mainCtx, mgr)
 	ctrlConfig.KaiScheduler.Enabled = kaiSchedulerEnabled
 
+	setupLog.Info("Detected orchestrators availability",
+		"grove", groveEnabled,
+		"lws", lwsEnabled,
+		"volcano", volcanoEnabled,
+		"kai-scheduler", kaiSchedulerEnabled,
+	)
+
 	// Create etcd client
 	cli, err := clientv3.New(clientv3.Config{
 		Endpoints:            []string{etcdAddr},

deploy/cloud/operator/internal/controller_common/predicate.go

@@ -101,6 +101,12 @@ func DetectLWSAvailability(ctx context.Context, mgr ctrl.Manager) bool {
 	return detectAPIGroupAvailability(ctx, mgr, "leaderworkerset.x-k8s.io")
 }
 
+// detectVolcanoAvailability checks if Volcano is available by checking if the Volcano API group is registered
+// This approach uses the discovery client which is simpler and more reliable
+func DetectVolcanoAvailability(ctx context.Context, mgr ctrl.Manager) bool {
+	return detectAPIGroupAvailability(ctx, mgr, "scheduling.volcano.sh")
+}
+
 // DetectKaiSchedulerAvailability checks if Kai-scheduler is available by checking if the scheduling.run.ai API group is registered
 // This approach uses the discovery client which is simpler and more reliable
 func DetectKaiSchedulerAvailability(ctx context.Context, mgr ctrl.Manager) bool {

docs/kubernetes/installation_guide.md

@@ -131,14 +131,24 @@ Found existing namespace-restricted Dynamo operators in namespaces: ...
 ```
 
 > [!TIP]
-> For multinode deployments, you need to enable Grove and KAI Scheduler.
-> You might chose to install them manually or through the dynamo-platform helm install command.
-> When using the dynamo-platform helm install command, Grove and KAI Scheduler are NOT installed by default. You can enable their installation by setting the following flags in the helm install command:
-
-```bash
---set "grove.enabled=true"
---set "kai-scheduler.enabled=true"
-```
+> For multinode deployments, you need to install multinode orchestration components:
+>
+> **Option 1 (Recommended): Grove + KAI Scheduler**
+> - Grove and KAI Scheduler can be installed manually or through the dynamo-platform helm install command.
+> - When using the dynamo-platform helm install command, Grove and KAI Scheduler are NOT installed by default. You can enable their installation by setting the following flags:
+>
+> ```bash
+> --set "grove.enabled=true"
+> --set "kai-scheduler.enabled=true"
+> ```
+>
+> **Option 2: LeaderWorkerSet (LWS) + Volcano**
+> - If using LWS for multinode deployments, you must also install Volcano (required dependency):
+>   - [LWS Installation](https://github.com/kubernetes-sigs/lws#installation)
+>   - [Volcano Installation](https://volcano.sh/en/docs/installation/) (required for gang scheduling with LWS)
+> - These must be installed manually before deploying multinode workloads with LWS.
+>
+> See the [Multinode Deployment Guide](./deployment/multinode-deployment.md) for details on orchestrator selection.
 
 > [!TIP]
 > By default, Model Express Server is not used.