Autoscaler should not try to monitor or scale virtual nodes and virtual node pools

Author: lxuan94-pp

cluster-autoscaler/cloudprovider/oci/nodepools/consts/annotations.go

@@ -28,4 +28,7 @@ const (
 
 	// EphemeralStorageSize is the freeform tag key that would be used to determine the ephemeral-storage size of the node
 	EphemeralStorageSize = "cluster-autoscaler/node-ephemeral-storage"
+
+	// OciNodePoolResourceIdent is the string identifier in the ocid that indicates the resource is a virtual node pool
+	OciVirtualNodeResourceIdent = "virtualnode"
 )

cluster-autoscaler/cloudprovider/oci/nodepools/oci_cloud_provider.go

@@ -10,9 +10,11 @@ import (
 	"k8s.io/apimachinery/pkg/api/resource"
 	"k8s.io/autoscaler/cluster-autoscaler/cloudprovider"
 	ocicommon "k8s.io/autoscaler/cluster-autoscaler/cloudprovider/oci/common"
+	npconsts "k8s.io/autoscaler/cluster-autoscaler/cloudprovider/oci/nodepools/consts"
 	caerrors "k8s.io/autoscaler/cluster-autoscaler/utils/errors"
 	"k8s.io/autoscaler/cluster-autoscaler/utils/gpu"
 	klog "k8s.io/klog/v2"
+	"strings"
 )
 
 // OciCloudProvider creates a cloud provider object that is compatible with node pools
@@ -86,6 +88,10 @@ func (ocp *OciCloudProvider) HasInstance(node *apiv1.Node) (bool, error) {
 	if err != nil {
 		return true, err
 	}
+	// Properly handle virtual nodes and missing node pool IDs to prevent crashes
+	if np == nil || np.Id() == "" || strings.Contains(instance.InstanceID, npconsts.OciVirtualNodeResourceIdent) {
+		return false, cloudprovider.ErrNotImplemented
+	}
 	nodes, err := ocp.manager.GetNodePoolNodes(np)
 	if err != nil {
 		return true, err

cluster-autoscaler/cloudprovider/oci/nodepools/oci_manager.go

@@ -630,6 +630,9 @@ func (m *ociManagerImpl) GetNodePoolNodes(np NodePool) ([]cloudprovider.Instance
 
 // GetNodePoolForInstance returns NodePool to which the given instance belongs.
 func (m *ociManagerImpl) GetNodePoolForInstance(instance ocicommon.OciRef) (NodePool, error) {
+	if strings.Contains(instance.InstanceID, npconsts.OciVirtualNodeResourceIdent) {
+		return nil, nil
+	}
 	if instance.NodePoolID == "" {
 		klog.V(4).Infof("node pool id missing from reference: %+v", instance)
 

cluster-autoscaler/cloudprovider/oci/nodepools/oci_manager_test.go

@@ -97,6 +97,16 @@ func TestGetNodePoolForInstance(t *testing.T) {
 	if np.Id() != "ocid2" {
 		t.Fatalf("got unexpected ocid %q ; wanted \"ocid2\"", np.Id())
 	}
+
+	// now verify node pool can be found via lookup up by instance id in cache
+	np, err = manager.GetNodePoolForInstance(ocicommon.OciRef{InstanceID: "virtualnode"})
+	if err != nil {
+		t.Fatalf("unexpected error: %+v", err)
+	}
+
+	if np != nil {
+		t.Fatalf("got unexpected ocid %q ; wanted nil", np.Id())
+	}
 }
 
 func TestGetNodePoolNodes(t *testing.T) {