fix(provider/dra): match nodes by annotated instance ID

CHANGELOG.md

@@ -34,6 +34,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 - The NFD engine now rejects an empty generated object set when cleanup is enabled, preserving the last published topology instead of deleting every Topograph-managed NFD object after an empty provider result or over-narrow node selection.
 - The NFD engine now groups nodes with `nvidia.com/gpu.clique` by that authoritative accelerator value instead of omitting their accelerator attribute.
 - The NFD engine now publishes the `system.name/nodename` attribute required by NFD to populate `NodeFeatureGroup.status.nodes`, including for simulated KWOK nodes where no NFD worker executes.
+- The DRA provider now matches nodes using the `topograph.nvidia.com/instance` annotation instead of assuming the instance ID equals the Kubernetes node name.
 - `kwok-nodes` now maps generated instance IDs back to model hostnames when naming Kubernetes nodes and writing the Topograph instance annotation.
 - The node-observer now discovers the optional node-data-broker through `NODE_DATA_BROKER_NAME` and `NODE_DATA_BROKER_NAMESPACE` and gates topology generation on the broker DaemonSet's desired and ready replica counts. Helm injects the variables only when `nodeDataBroker.enabled=true`, so disabling the broker cannot leave the observer waiting for nonexistent pods.
 - The node-observer reports an actionable error and defers topology generation when an enabled node-data-broker DaemonSet has zero desired replicas.

pkg/providers/dra/provider.go

@@ -14,6 +14,7 @@ import (
 	"k8s.io/apimachinery/pkg/labels"
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/rest"
+	"k8s.io/klog/v2"
 
 	"github.com/NVIDIA/topograph/internal/config"
 	"github.com/NVIDIA/topograph/internal/httperr"
@@ -30,7 +31,7 @@ const (
 
 type Provider struct {
 	config *rest.Config
-	client *kubernetes.Clientset
+	client kubernetes.Interface
 	params *Params
 }
 
@@ -109,8 +110,14 @@ func (p *Provider) GenerateTopologyConfig(ctx context.Context, _ *int, instances
 		}
 
 		i2n := instances[indx].Instances
-		if host, ok := i2n[node.Name]; ok {
-			domainMap.AddHost(clusterID, node.Name, host)
+		instanceID, ok := node.Annotations[topology.KeyNodeInstance]
+		if !ok || instanceID == "" {
+			klog.Warningf("missing or empty %q annotation in node %s", topology.KeyNodeInstance, node.Name)
+			continue
+		}
+
+		if host, ok := i2n[instanceID]; ok {
+			domainMap.AddHost(clusterID, instanceID, host)
 		}
 	}
 

pkg/providers/dra/provider_test.go

@@ -6,10 +6,14 @@
 package dra
 
 import (
+	"context"
 	"testing"
 
+	"github.com/NVIDIA/topograph/pkg/topology"
 	"github.com/stretchr/testify/require"
+	corev1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/client-go/kubernetes/fake"
 )
 
 func TestGetParameters(t *testing.T) {
@@ -53,3 +57,33 @@ func TestGetParameters(t *testing.T) {
 		})
 	}
 }
+
+func TestGenerateTopologyConfigUsesAnnotatedInstanceID(t *testing.T) {
+	node := &corev1.Node{ObjectMeta: metav1.ObjectMeta{
+		Name: "k8s-node-1",
+		Labels: map[string]string{
+			DomainLabel: "clique-1",
+		},
+		Annotations: map[string]string{
+			topology.KeyNodeInstance: "instance-123",
+			topology.KeyNodeRegion:   "local",
+		},
+	}}
+	provider := &Provider{
+		client: fake.NewSimpleClientset(node),
+		params: &Params{},
+	}
+	instances := []topology.ComputeInstances{{
+		Region: "local",
+		Instances: map[string]string{
+			"instance-123": "scheduler-node-1",
+		},
+	}}
+
+	graph, httpErr := provider.GenerateTopologyConfig(context.Background(), nil, instances)
+
+	require.Nil(t, httpErr)
+	expectedDomains := topology.NewDomainMap()
+	expectedDomains.AddHost("clique-1", "instance-123", "scheduler-node-1")
+	require.Equal(t, expectedDomains, graph.Domains)
+}