Introduce a new nodeVolumeDetachTimeout field and ExcludeNodeVolumeDetachingAnnotation annotation

to be able to skip the wait for volume detachment. Move out shouldWaitForNodeVolumes from the node
drain logic to its own condition to keep a cleaner separation between them.

Author: furkatgofurov7

api/v1alpha3/conversion.go

@@ -98,6 +98,7 @@ func (src *Machine) ConvertTo(dstRaw conversion.Hub) error {
 	}
 
 	dst.Spec.NodeDeletionTimeout = restored.Spec.NodeDeletionTimeout
+	dst.Spec.NodeVolumeDetachTimeout = restored.Spec.NodeVolumeDetachTimeout
 	dst.Status.NodeInfo = restored.Status.NodeInfo
 	return nil
 }
@@ -141,6 +142,7 @@ func (src *MachineSet) ConvertTo(dstRaw conversion.Hub) error {
 		return err
 	}
 	dst.Spec.Template.Spec.NodeDeletionTimeout = restored.Spec.Template.Spec.NodeDeletionTimeout
+	dst.Spec.Template.Spec.NodeVolumeDetachTimeout = restored.Spec.Template.Spec.NodeVolumeDetachTimeout
 	dst.Status.Conditions = restored.Status.Conditions
 	return nil
 }
@@ -195,6 +197,7 @@ func (src *MachineDeployment) ConvertTo(dstRaw conversion.Hub) error {
 	}
 
 	dst.Spec.Template.Spec.NodeDeletionTimeout = restored.Spec.Template.Spec.NodeDeletionTimeout
+	dst.Spec.Template.Spec.NodeVolumeDetachTimeout = restored.Spec.Template.Spec.NodeVolumeDetachTimeout
 	dst.Status.Conditions = restored.Status.Conditions
 	return nil
 }

api/v1alpha3/zz_generated.conversion.go

@@ -159,6 +159,7 @@ func (src *Machine) ConvertTo(dstRaw conversion.Hub) error {
 	}
 
 	dst.Spec.NodeDeletionTimeout = restored.Spec.NodeDeletionTimeout
+	dst.Spec.NodeVolumeDetachTimeout = restored.Spec.NodeVolumeDetachTimeout
 	return nil
 }
 
@@ -203,6 +204,7 @@ func (src *MachineSet) ConvertTo(dstRaw conversion.Hub) error {
 	}
 
 	dst.Spec.Template.Spec.NodeDeletionTimeout = restored.Spec.Template.Spec.NodeDeletionTimeout
+	dst.Spec.Template.Spec.NodeVolumeDetachTimeout = restored.Spec.Template.Spec.NodeVolumeDetachTimeout
 	return nil
 }
 
@@ -243,6 +245,7 @@ func (src *MachineDeployment) ConvertTo(dstRaw conversion.Hub) error {
 	}
 
 	dst.Spec.Template.Spec.NodeDeletionTimeout = restored.Spec.Template.Spec.NodeDeletionTimeout
+	dst.Spec.Template.Spec.NodeVolumeDetachTimeout = restored.Spec.Template.Spec.NodeVolumeDetachTimeout
 	return nil
 }
 

api/v1alpha4/conversion.go

@@ -33,6 +33,9 @@ const (
 	// ExcludeNodeDrainingAnnotation annotation explicitly skips node draining if set.
 	ExcludeNodeDrainingAnnotation = "machine.cluster.x-k8s.io/exclude-node-draining"
 
+	// ExcludeWaitForNodeVolumeDetachAnnotation annotation explicitly skips the waiting for node volume detaching if set.
+	ExcludeWaitForNodeVolumeDetachAnnotation = "machine.cluster.x-k8s.io/exclude-wait-for-node-volume-detach"
+
 	// MachineSetLabelName is the label set on machines if they're controlled by MachineSet.
 	MachineSetLabelName = "cluster.x-k8s.io/set-name"
 
@@ -97,6 +100,11 @@ type MachineSpec struct {
 	// +optional
 	NodeDrainTimeout *metav1.Duration `json:"nodeDrainTimeout,omitempty"`
 
+	// NodeVolumeDetachTimeout is the total amount of time that the controller will spend on waiting for all volumes
+	// to be detached. The default value is 0, meaning that the volumes can be detached without any time limitations.
+	// +optional
+	NodeVolumeDetachTimeout *metav1.Duration `json:"nodeVolumeDetachTimeout,omitempty"`
+
 	// NodeDeletionTimeout defines how long the controller will attempt to delete the Node that the Machine
 	// hosts after the Machine is marked for deletion. A duration of 0 will retry deletion indefinitely.
 	// Defaults to 10 seconds.

api/v1alpha4/zz_generated.conversion.go

@@ -42,6 +42,7 @@ func (src *KubeadmControlPlane) ConvertTo(dstRaw conversion.Hub) error {
 	dst.Spec.MachineTemplate.NodeDeletionTimeout = restored.Spec.MachineTemplate.NodeDeletionTimeout
 	dst.Spec.KubeadmConfigSpec.Files = restored.Spec.KubeadmConfigSpec.Files
 	dst.Spec.KubeadmConfigSpec.Users = restored.Spec.KubeadmConfigSpec.Users
+	dst.Spec.MachineTemplate.NodeVolumeDetachTimeout = restored.Spec.MachineTemplate.NodeVolumeDetachTimeout
 	dst.Status.Version = restored.Status.Version
 
 	if restored.Spec.KubeadmConfigSpec.Users != nil {

api/v1beta1/machine_types.go

@@ -67,6 +67,7 @@ func (src *KubeadmControlPlane) ConvertTo(dstRaw conversion.Hub) error {
 	}
 
 	dst.Spec.MachineTemplate.NodeDeletionTimeout = restored.Spec.MachineTemplate.NodeDeletionTimeout
+	dst.Spec.MachineTemplate.NodeVolumeDetachTimeout = restored.Spec.MachineTemplate.NodeVolumeDetachTimeout
 
 	return nil
 }
@@ -138,6 +139,7 @@ func (src *KubeadmControlPlaneTemplate) ConvertTo(dstRaw conversion.Hub) error {
 		dst.Spec.Template.Spec.MachineTemplate = restored.Spec.Template.Spec.MachineTemplate
 	} else if restored.Spec.Template.Spec.MachineTemplate != nil {
 		dst.Spec.Template.Spec.MachineTemplate.NodeDeletionTimeout = restored.Spec.Template.Spec.MachineTemplate.NodeDeletionTimeout
+		dst.Spec.Template.Spec.MachineTemplate.NodeVolumeDetachTimeout = restored.Spec.Template.Spec.MachineTemplate.NodeVolumeDetachTimeout
 	}
 
 	return nil

api/v1beta1/zz_generated.deepcopy.go

@@ -102,6 +102,11 @@ type KubeadmControlPlaneMachineTemplate struct {
 	// +optional
 	NodeDrainTimeout *metav1.Duration `json:"nodeDrainTimeout,omitempty"`
 
+	// NodeVolumeDetachTimeout is the total amount of time that the controller will spend on waiting for all volumes
+	// to be detached. The default value is 0, meaning that the volumes can be detached without any time limitations.
+	// +optional
+	NodeVolumeDetachTimeout *metav1.Duration `json:"nodeVolumeDetachTimeout,omitempty"`
+
 	// NodeDeletionTimeout defines how long the machine controller will attempt to delete the Node that the Machine
 	// hosts after the Machine is marked for deletion. A duration of 0 will retry deletion indefinitely.
 	// If no value is provided, the default value for this property of the Machine resource will be used.

api/v1beta1/zz_generated.openapi.go

@@ -159,6 +159,7 @@ func (in *KubeadmControlPlane) ValidateUpdate(old runtime.Object) error {
 		{spec, "machineTemplate", "infrastructureRef", "name"},
 		{spec, "machineTemplate", "infrastructureRef", "kind"},
 		{spec, "machineTemplate", "nodeDrainTimeout"},
+		{spec, "machineTemplate", "nodeVolumeDetachTimeout"},
 		{spec, "machineTemplate", "nodeDeletionTimeout"},
 		{spec, "replicas"},
 		{spec, "version"},

config/crd/bases/cluster.x-k8s.io_machinedeployments.yaml

@@ -261,8 +261,9 @@ func TestKubeadmControlPlaneValidateUpdate(t *testing.T) {
 					Namespace:  "foo",
 					Name:       "infraTemplate",
 				},
-				NodeDrainTimeout:    &metav1.Duration{Duration: time.Second},
-				NodeDeletionTimeout: &metav1.Duration{Duration: time.Second},
+				NodeDrainTimeout:        &metav1.Duration{Duration: time.Second},
+				NodeVolumeDetachTimeout: &metav1.Duration{Duration: time.Second},
+				NodeDeletionTimeout:     &metav1.Duration{Duration: time.Second},
 			},
 			Replicas: pointer.Int32Ptr(1),
 			RolloutStrategy: &RolloutStrategy{
@@ -388,6 +389,7 @@ func TestKubeadmControlPlaneValidateUpdate(t *testing.T) {
 	validUpdate.Spec.MachineTemplate.InfrastructureRef.APIVersion = "test/v1alpha2"
 	validUpdate.Spec.MachineTemplate.InfrastructureRef.Name = "orange"
 	validUpdate.Spec.MachineTemplate.NodeDrainTimeout = &metav1.Duration{Duration: 10 * time.Second}
+	validUpdate.Spec.MachineTemplate.NodeVolumeDetachTimeout = &metav1.Duration{Duration: 10 * time.Second}
 	validUpdate.Spec.MachineTemplate.NodeDeletionTimeout = &metav1.Duration{Duration: 10 * time.Second}
 	validUpdate.Spec.Replicas = pointer.Int32Ptr(5)
 	now := metav1.NewTime(time.Now())

config/crd/bases/cluster.x-k8s.io_machinepools.yaml

@@ -100,6 +100,11 @@ type KubeadmControlPlaneTemplateMachineTemplate struct {
 	// +optional
 	NodeDrainTimeout *metav1.Duration `json:"nodeDrainTimeout,omitempty"`
 
+	// NodeVolumeDetachTimeout is the total amount of time that the controller will spend on waiting for all volumes
+	// to be detached. The default value is 0, meaning that the volumes can be detached without any time limitations.
+	// +optional
+	NodeVolumeDetachTimeout *metav1.Duration `json:"nodeVolumeDetachTimeout,omitempty"`
+
 	// NodeDeletionTimeout defines how long the machine controller will attempt to delete the Node that the Machine
 	// hosts after the Machine is marked for deletion. A duration of 0 will retry deletion indefinitely.
 	// If no value is provided, the default value for this property of the Machine resource will be used.

config/crd/bases/cluster.x-k8s.io_machines.yaml

@@ -98,6 +98,10 @@ documentation][scale].
   that the controller will spend on draining a control plane node.
   The default value is 0, meaning that the node can be drained without any time limitations.
 
+* `machineTemplate.nodeVolumeDetachTimeout` - is a *metav1.Duration defining how long the controller
+  will spend on waiting for all volumes to be detached.
+  The default value is 0, meaning that the volume can be detached without any time limitations.
+
 * `machineTemplate.nodeDeletionTimeout` - is a *metav1.Duration defining how long the controller
   will attempt to delete the Node that is hosted by a Machine after the Machine is marked for
   deletion. A duration of 0 will retry deletion indefinitely. It defaults to 10 seconds on the

config/crd/bases/cluster.x-k8s.io_machinesets.yaml

@@ -71,6 +71,7 @@ func (src *MachinePool) ConvertTo(dstRaw conversion.Hub) error {
 		return err
 	}
 	dst.Spec.Template.Spec.NodeDeletionTimeout = restored.Spec.Template.Spec.NodeDeletionTimeout
+	dst.Spec.Template.Spec.NodeVolumeDetachTimeout = restored.Spec.Template.Spec.NodeVolumeDetachTimeout
 	return nil
 }
 

controlplane/kubeadm/api/v1alpha3/conversion.go

@@ -36,6 +36,7 @@ func (src *MachinePool) ConvertTo(dstRaw conversion.Hub) error {
 		return err
 	}
 	dst.Spec.Template.Spec.NodeDeletionTimeout = restored.Spec.Template.Spec.NodeDeletionTimeout
+	dst.Spec.Template.Spec.NodeVolumeDetachTimeout = restored.Spec.Template.Spec.NodeVolumeDetachTimeout
 	return nil
 }