drain: adding 'USE_EXTERNAL_DRAINER' provoding an option to enable/disable SRIOV OP drain controller, in favor of using maintenance OP to drive node drain aspects

Signed-off-by: Ido Heyvi <iheyvi@nvidia.com>

Author: heyvister1

README.md

@@ -468,7 +468,7 @@ nodes in parallel from the pool the operator can drain in parallel. maxUnavailab
 
 > **NOTE**: If a node is not part of any pool it will have a default configuration of maxUnavailable 1
 
-> **NOTE**: Node draining can be delegated to an external drain-controller by setting `USE_EXTERNAL_DRAINER=true` (e.g. using [NVIDIA maintenance-operator](https://github.com/Mellanox/maintenance-operator)) (PR #952). This means that internal drain-controller continues to work on nodeState objects which were not annotated with `sriovnetwork.openshift.io/use-external-drainer`. In addition, `SriovNetworkPoolConfig` will not take any effect during drain procedure, since the maintenance operator will be in charge of [parallel node operations](https://github.com/Mellanox/maintenance-operator/blob/main/api/v1alpha1/maintenanceoperatorconfig_types.go#L38-L46).
+> **NOTE**: Internal drain controller can be disabled by exposing the following `USE_EXTERNAL_DRAINER` env variable. This means that drain operations will be done externally, for example by utilizing [NVIDIA maintenance OP](https://github.com/Mellanox/maintenance-operator). In addition, `SriovNetworkPoolConfig` will not take any effect during drain procedure, since the maintenance operator will be in charge of [parallel node operations](https://github.com/Mellanox/maintenance-operator/blob/main/api/v1alpha1/maintenanceoperatorconfig_types.go#L38-L46).
 
 #### RDMA Mode Configuration
 

deployment/sriov-network-operator-chart/templates/operator.yaml

@@ -87,10 +87,8 @@ spec:
               value: {{ .Values.operator.metricsExporter.certificates.secretName }}
             - name: METRICS_EXPORTER_KUBE_RBAC_PROXY_IMAGE
               value: {{ .Values.images.metricsExporterKubeRbacProxy }}
-            {{- if .Values.operator.externalDrainer.enabled }}
             - name: USE_EXTERNAL_DRAINER
               value: {{ .Values.operator.externalDrainer.enabled | quote }}
-            {{- end }}
             {{- if .Values.operator.metricsExporter.prometheusOperator.enabled }}
             - name: METRICS_EXPORTER_PROMETHEUS_OPERATOR_ENABLED
               value: {{ .Values.operator.metricsExporter.prometheusOperator.enabled | quote}}

doc/design/parallel-node-config.md

@@ -60,7 +60,7 @@ Node annotation, `sriovnetwork.openshift.io/state` and SriovNetworkNodeState ann
 
 *NOTE:* In the future we are going to drop the node annotation and only use the SriovNetworkNodeState
 
-*NOTE:* Node draining can be delegated to an external drain-controller by setting `USE_EXTERNAL_DRAINER=true` (e.g. using [NVIDIA maintenance-operator](https://github.com/Mellanox/maintenance-operator)) (PR #952). In addition, `SriovNetworkPoolConfig` will not take any effect during drain procedure, since the maintenance operator will be in charge of parallel node operations.
+*NOTE:* Internal drain controller can be disabled by exposing the following `USE_EXTERNAL_DRAINER` env variable. This means that drain operations will be done externally, for example by utilizing [NVIDIA maintenance OP](https://github.com/Mellanox/maintenance-operator). In addition, `SriovNetworkPoolConfig` will not take any effect during drain procedure, since the maintenance operator will be in charge of parallel node operations.
 
 Draining procedure:
 

main.go

@@ -272,12 +272,10 @@ func main() {
 }
 
 func setupDrainController(mgr ctrl.Manager, restConfig *rest.Config,
-	orch orchestrator.Interface, scheme *runtime.Scheme) error {
+	platformsHelper platforms.Interface, scheme *runtime.Scheme) error {
 	if vars.UseExternalDrainer {
-		// even though UseExternalDrainer is set, we are keeping internal drain controller for handling
-		// use-cases when there are existing nodes under drain_required state, which won't be handled
-		// externally, since UseExternalDrainer was set only after they were scheduled for draining.
-		setupLog.Info("'UseExternalDrainer' is set, draining will be done externally")
+		setupLog.Info("internal drain controller is disabled, draining will be done externally by the maintenance operator")
+		return nil
 	}
 
 	// we need a client that doesn't use the local cache for the objects
@@ -296,11 +294,10 @@ func setupDrainController(mgr ctrl.Manager, restConfig *rest.Config,
 		return err
 	}
 
-	drainController, err := controllers.NewDrainReconcileController(
-		drainKClient,
+	drainController, err := controllers.NewDrainReconcileController(drainKClient,
 		mgr.GetScheme(),
 		mgr.GetEventRecorderFor("SR-IOV operator"),
-		orch)
+		platformsHelper)
 	if err != nil {
 		setupLog.Error(err, "unable to create controller", "controller", "DrainReconcile")
 		return err

pkg/vars/vars.go

@@ -99,9 +99,6 @@ var (
 	// FeatureGates interface to interact with feature gates
 	FeatureGate featuregate.FeatureGate
 
-	// ErrOperationNotSupportedByPlatform is returned when a platform operation is not supported by the platform implementation.
-	ErrOperationNotSupportedByPlatform = errors.New("operation not supported by the platform")
-
 	// UseExternalDrainer controls if SRIOV operator will use an external drainer
 	// for draining nodes or its internal drain controller (default)
 	UseExternalDrainer bool
@@ -130,12 +127,3 @@ func init() {
 
 	UseExternalDrainer = os.Getenv("USE_EXTERNAL_DRAINER") == "true"
 }
-
-func GetPlatformType(providerID string) consts.PlatformTypes {
-	for key, pType := range PlatformsMap {
-		if strings.Contains(strings.ToLower(providerID), strings.ToLower(key)) {
-			return pType
-		}
-	}
-	return consts.Baremetal
-}