restrict network migration mode to techpreview

config/v1/feature_gates.go

@@ -282,6 +282,16 @@ var (
 		OwningProduct:       ocpSpecific,
 	}
 
+	FeatureGateNetworkLiveMigration = FeatureGateName("NetworkLiveMigration")
+	sdnLiveMigration                = FeatureGateDescription{
+		FeatureGateAttributes: FeatureGateAttributes{
+			Name: FeatureGateNetworkLiveMigration,
+		},
+		OwningJiraComponent: "Networking/ovn-kubernetes",
+		ResponsiblePerson:   "pliu",
+		OwningProduct:       ocpSpecific,
+	}
+
 	FeatureGateAutomatedEtcdBackup = FeatureGateName("AutomatedEtcdBackup")
 	automatedEtcdBackup            = FeatureGateDescription{
 		FeatureGateAttributes: FeatureGateAttributes{

config/v1/types_feature.go

@@ -185,6 +185,7 @@ var FeatureSets = map[FeatureSet]*FeatureGateEnabledDisabled{
 		with(automatedEtcdBackup).
 		without(machineAPIOperatorDisableMachineHealthCheckController).
 		with(adminNetworkPolicy).
+		with(sdnLiveMigration).
 		toFeatures(defaultFeatures),
 	LatencySensitive: newDefaultFeatures().
 		toFeatures(defaultFeatures),

openapi/openapi.json

@@ -26994,6 +26994,11 @@
           "description": "features contains the features migration configuration. Set this to migrate feature configuration when changing the cluster default network provider. if unset, the default operation is to migrate all the configuration of supported features.",
           "$ref": "#/definitions/com.github.openshift.api.operator.v1.FeaturesMigration"
         },
+        "mode": {
+          "description": "mode indicates the mode of network migration. The supported values are \"Live\", \"Offline\" and omitted. A \"Live\" migration operation will not cause service interruption by migrating the CNI of each node one by one. The cluster network will work as normal during the network migration. An \"Offline\" migration operation will cause service interruption. During an \"Offline\" migration, two rounds of node reboots are required. The cluster network will be malfunctioning during the network migration. When omitted, this means no opinion and the platform is left to choose a reasonable default which is subject to change over time. The current default value is \"Offline\".",
+          "type": "string",
+          "default": ""
+        },
         "mtu": {
           "description": "mtu contains the MTU migration configuration. Set this to allow changing the MTU values for the default network. If unset, the operation of changing the MTU for the default network will be rejected.",
           "$ref": "#/definitions/com.github.openshift.api.operator.v1.MTUMigration"

operator/v1/0000_70_cluster-network-operator_01.crd.yaml → operator/v1/0000_70_cluster-network-operator_01-Default.crd.yaml

@@ -5,6 +5,7 @@ metadata:
     api-approved.openshift.io: https://github.com/openshift/api/pull/475
     include.release.openshift.io/self-managed-high-availability: "true"
     include.release.openshift.io/single-node-developer: "true"
+    release.openshift.io/feature-set: Default
   name: networks.operator.openshift.io
 spec:
   group: operator.openshift.io

operator/v1/custom.network.testsuite.yaml

@@ -0,0 +1,21 @@
+apiVersion: apiextensions.k8s.io/v1 # Hack because controller-gen complains if we don't have this
+name: "[Stable] Network"
+crd: 0000_70_cluster-network-operator_01-Custom.crd.yaml
+tests:
+  onCreate:
+    - name: Should be able to create migration mode
+      initial: |
+        apiVersion: operator.openshift.io/v1
+        kind: Network
+        spec: 
+          migration:
+            mode: Live
+      expected: |
+        apiVersion: operator.openshift.io/v1
+        kind: Network
+        spec:
+          disableNetworkDiagnostics: false
+          logLevel: Normal
+          operatorLogLevel: Normal
+          migration:
+            mode: Live

operator/v1/stable.network.testsuite.yaml

@@ -1,6 +1,6 @@
 apiVersion: apiextensions.k8s.io/v1 # Hack because controller-gen complains if we don't have this
 name: "[Stable] Network"
-crd: 0000_70_cluster-network-operator_01.crd.yaml
+crd: 0000_70_cluster-network-operator_01-Default.crd.yaml
 tests:
   onCreate:
   - name: Should be able to create a minimal Network
@@ -15,3 +15,17 @@ tests:
         disableNetworkDiagnostics: false
         logLevel: Normal
         operatorLogLevel: Normal
+  - name: Should not be able to create migration mode
+    initial: |
+      apiVersion: operator.openshift.io/v1
+      kind: Network
+      spec: 
+        migration:
+          mode: Live
+    expected: |
+      apiVersion: operator.openshift.io/v1
+      kind: Network
+      spec:
+        disableNetworkDiagnostics: false
+        logLevel: Normal
+        operatorLogLevel: Normal

operator/v1/techpreview.network.testsuite.yaml

@@ -0,0 +1,21 @@
+apiVersion: apiextensions.k8s.io/v1 # Hack because controller-gen complains if we don't have this
+name: "[Stable] Network"
+crd: 0000_70_cluster-network-operator_01-TechPreviewNoUpgrade.crd.yaml
+tests:
+  onCreate:
+  - name: Should be able to create migration mode
+    initial: |
+      apiVersion: operator.openshift.io/v1
+      kind: Network
+      spec: 
+        migration:
+          mode: Live
+    expected: |
+      apiVersion: operator.openshift.io/v1
+      kind: Network
+      spec:
+        disableNetworkDiagnostics: false
+        logLevel: Normal
+        operatorLogLevel: Normal
+        migration:
+          mode: Live

operator/v1/types_network.go

@@ -116,7 +116,20 @@ type NetworkSpec struct {
 	Migration *NetworkMigration `json:"migration,omitempty"`
 }
 
+// NetworkMigrationMode is an enumeration of the possible mode of the network migration
+// Valid values are "Live" or "Offline".
+// +kubebuilder:validation:Enum:="";Live;Offline
+type NetworkMigrationMode string
+
+const (
+	// A "Live" migration operation will not cause service interruption by migrating the CNI of each node one by one. The cluster network will work as normal during the network migration.
+	LiveNetworkMigrationMode NetworkMigrationMode = "Live"
+	// An "Offline" migration operation will cause service interruption. During an "Offline" migration, two rounds of node reboots are required. The cluster network will be malfunctioning during the network migration.
+	OfflineNetworkMigrationMode NetworkMigrationMode = "Offline"
+)
+
 // NetworkMigration represents the cluster network configuration.
+// +openshift:validation:FeatureSetAwareXValidation:featureSet=CustomNoUpgrade;TechPreviewNoUpgrade,rule="!has(self.mtu) || !has(self.mode) || self.mode == ”",message="migration mode may not be configured at the same time as an mtu migration"
 type NetworkMigration struct {
 	// networkType is the target type of network migration. Set this to the
 	// target network type to allow changing the default network. If unset, the
@@ -137,6 +150,16 @@ type NetworkMigration struct {
 	// supported features.
 	// +optional
 	Features *FeaturesMigration `json:"features,omitempty"`
+
+	// mode indicates the mode of network migration.
+	// The supported values are "Live", "Offline" and omitted.
+	// A "Live" migration operation will not cause service interruption by migrating the CNI of each node one by one. The cluster network will work as normal during the network migration.
+	// An "Offline" migration operation will cause service interruption. During an "Offline" migration, two rounds of node reboots are required. The cluster network will be malfunctioning during the network migration.
+	// When omitted, this means no opinion and the platform is left to choose a reasonable default which is subject to change over time.
+	// The current default value is "Offline".
+	// +openshift:enable:FeatureSets=CustomNoUpgrade;TechPreviewNoUpgrade
+	// +optional
+	Mode NetworkMigrationMode `json:"mode"`
 }
 
 type FeaturesMigration struct {