fix(exoscale): enhance UX around SKS nodepools

.gitignore

@@ -16,6 +16,7 @@
 # VSCode project files
 **/.vscode
 *.code-workspace
+__debug_bin*
 
 # Emacs save files
 *~

cluster-autoscaler/cloudprovider/exoscale/exoscale_cloud_provider_test.go

@@ -197,6 +197,21 @@ func (ts *cloudProviderTestSuite) TestExoscaleCloudProvider_NodeGroupForNode_Ins
 			nil,
 		)
 
+	ts.p.manager.client.(*exoscaleClientMock).
+		On("ListSKSClusters", ts.p.manager.ctx, ts.p.manager.zone).
+		Return(
+			[]*egoscale.SKSCluster{{
+				ID:   &testSKSClusterID,
+				Name: &testSKSClusterName,
+				Nodepools: []*egoscale.SKSNodepool{{
+					ID:             &testSKSNodepoolID,
+					InstancePoolID: &testInstancePoolID,
+					Name:           &testSKSNodepoolName,
+				}},
+			}},
+			nil,
+		)
+
 	nodeGroup, err := ts.p.NodeGroupForNode(&apiv1.Node{
 		Spec: apiv1.NodeSpec{
 			ProviderID: toProviderID(testInstanceID),
@@ -280,7 +295,7 @@ func (ts *cloudProviderTestSuite) TestExoscaleCloudProvider_NodeGroupForNode_SKS
 	})
 	ts.Require().NoError(err)
 	ts.Require().NotNil(nodeGroup)
-	ts.Require().Equal(testInstancePoolID, nodeGroup.Id())
+	ts.Require().Equal(testSKSNodepoolID, nodeGroup.Id())
 	ts.Require().IsType(&sksNodepoolNodeGroup{}, nodeGroup)
 }
 
@@ -359,6 +374,21 @@ func (ts *cloudProviderTestSuite) TestExoscaleCloudProvider_NodeGroups() {
 			nil,
 		)
 
+	ts.p.manager.client.(*exoscaleClientMock).
+		On("ListSKSClusters", ts.p.manager.ctx, ts.p.manager.zone).
+		Return(
+			[]*egoscale.SKSCluster{{
+				ID:   &testSKSClusterID,
+				Name: &testSKSClusterName,
+				Nodepools: []*egoscale.SKSNodepool{{
+					ID:             &testSKSNodepoolID,
+					InstancePoolID: &sksNodepoolInstancePoolID,
+					Name:           &testSKSNodepoolName,
+				}},
+			}},
+			nil,
+		)
+
 	instancePoolNodeGroup, err := ts.p.NodeGroupForNode(&apiv1.Node{
 		Spec: apiv1.NodeSpec{
 			ProviderID: toProviderID(instancePoolInstanceID),
@@ -432,7 +462,11 @@ func (ts *cloudProviderTestSuite) TestExoscaleCloudProvider_NodeGroups() {
 
 	// ---------------------------------------------------------------
 
-	ts.Require().Len(ts.p.NodeGroups(), 2)
+	nodeGroups := ts.p.NodeGroups()
+	ts.Require().Len(nodeGroups, 2)
+	if len(nodeGroups) >= 2 {
+		ts.Require().Equal(nodeGroups[1].Id(), testSKSNodepoolID)
+	}
 }
 
 func TestSuiteExoscaleCloudProvider(t *testing.T) {

cluster-autoscaler/cloudprovider/exoscale/exoscale_manager.go

@@ -96,14 +96,46 @@ func newManager(discoveryOpts cloudprovider.NodeGroupDiscoveryOptions) (*Manager
 // based on the `--scan-interval`. By default it's 10 seconds.
 func (m *Manager) Refresh() error {
 	var nodeGroups []cloudprovider.NodeGroup
+
+	// load clusters, it's required for SKS node groups check
+	sksClusters, err := m.client.ListSKSClusters(m.ctx, m.zone)
+	if err != nil {
+		errorf("unable to list SKS clusters: %v", err)
+		return err
+	}
+
 	for _, ng := range m.nodeGroups {
-		if _, err := m.client.GetInstancePool(m.ctx, m.zone, ng.Id()); err != nil {
-			if errors.Is(err, exoapi.ErrNotFound) {
+		// Check SKS Nodepool existence first
+		found := false
+		for _, c := range sksClusters {
+			for _, np := range c.Nodepools {
+				if *np.ID == ng.Id() {
+					if _, err := m.client.GetInstancePool(m.ctx, m.zone, *np.InstancePoolID); err != nil {
+						if !errors.Is(err, exoapi.ErrNotFound) {
+							errorf("unable to retrieve SKS Instance Pool %s: %v", ng.Id(), err)
+							return err
+						}
+					} else {
+						found = true
+						break
+					}
+				}
+			}
+		}
+
+		if !found {
+			// If SKS Nodepool is not found, check the Instance Pool
+			// it was the previous behavior which was less convenient for end user UX
+			if _, err := m.client.GetInstancePool(m.ctx, m.zone, ng.Id()); err != nil {
+				if !errors.Is(err, exoapi.ErrNotFound) {
+					errorf("unable to retrieve SKS Instance Pool %s: %v", ng.Id(), err)
+					return err
+				}
+
+				// Neither SKS Nodepool nor Instance Pool found, remove it from cache
 				debugf("removing node group %s from manager cache", ng.Id())
 				continue
 			}
-			errorf("unable to retrieve Instance Pool %s: %v", ng.Id(), err)
-			return err
 		}
 
 		nodeGroups = append(nodeGroups, ng)

cluster-autoscaler/cloudprovider/exoscale/exoscale_node_group_instance_pool.go

@@ -212,9 +212,9 @@ func (n *instancePoolNodeGroup) GetOptions(_ config.NodeGroupAutoscalingOptions)
 
 func (n *instancePoolNodeGroup) waitUntilRunning(ctx context.Context) error {
 	return pollCmd(ctx, func() (bool, error) {
-		instancePool, err := n.m.client.GetInstancePool(ctx, n.m.zone, n.Id())
+		instancePool, err := n.m.client.GetInstancePool(ctx, n.m.zone, *n.instancePool.ID)
 		if err != nil {
-			errorf("unable to retrieve Instance Pool %s: %s", n.Id(), err)
+			errorf("unable to retrieve Instance Pool %s: %s", *n.instancePool.ID, err)
 			return false, err
 		}
 

cluster-autoscaler/cloudprovider/exoscale/exoscale_node_group_sks_nodepool.go

@@ -151,7 +151,7 @@ func (n *sksNodepoolNodeGroup) DecreaseTargetSize(_ int) error {
 
 // Id returns an unique identifier of the node group.
 func (n *sksNodepoolNodeGroup) Id() string {
-	return *n.sksNodepool.InstancePoolID
+	return *n.sksNodepool.ID
 }
 
 // Debug returns a string containing all information regarding this node group.
@@ -229,9 +229,9 @@ func (n *sksNodepoolNodeGroup) GetOptions(_ config.NodeGroupAutoscalingOptions)
 
 func (n *sksNodepoolNodeGroup) waitUntilRunning(ctx context.Context) error {
 	return pollCmd(ctx, func() (bool, error) {
-		instancePool, err := n.m.client.GetInstancePool(ctx, n.m.zone, n.Id())
+		instancePool, err := n.m.client.GetInstancePool(ctx, n.m.zone, *n.sksNodepool.InstancePoolID)
 		if err != nil {
-			errorf("unable to retrieve Instance Pool %s: %s", n.Id(), err)
+			errorf("unable to retrieve Instance Pool %s: %s", *n.sksNodepool.InstancePoolID, err)
 			return false, err
 		}
 

cluster-autoscaler/cloudprovider/exoscale/exoscale_node_group_sks_nodepool_test.go

@@ -206,7 +206,7 @@ func (ts *cloudProviderTestSuite) TestSKSNodepoolNodeGroup_Id() {
 		maxSize: int(testComputeInstanceQuotaLimit),
 	}
 
-	ts.Require().Equal(testInstancePoolID, nodeGroup.Id())
+	ts.Require().Equal(testSKSNodepoolID, nodeGroup.Id())
 }
 
 func (ts *cloudProviderTestSuite) TestSKSNodepoolNodeGroup_Nodes() {