MTV-1536 | Use CDI for disk transfer in cold migration

Issues: [1] Allow migration of "unknow" guests Right now when we want to migrate an unknown and unsupported operating system which is unsupported by the virt-v2v [3]. [2] Unifying the process and potential speedup Right now we are using two different methods for the disk transfer. This brings additional engineering for maintaining two paths. It's harder to debug two different flows. The virt-v2v transfers the disks in the sequence whereas using the CDI we can start multiple disk imports in parallel. This can improve the migration speeds. Fix: MTV is already using the CNV CDI for the warm and remote migration. We just need to adjust the code to remove the virt-v2v transfer and rely on the CNV CDI to do it for us. Drawbacks: - CNV CDI *requires* the VDDK, which was till now highly recommended. - CNV CDI is not maintained inside the MTV and there might be problems escalating and backporting the patches as CNV has a different release cycle. - Because we will be migrating all disks in parallel we need to optimise our migration scheduler as we don't want to take too much of the hosts/network resources. I have already done some optimisations in [4,5,6]. Notes: This change removes the usage of virt-v2v and we will only use the virt-v2v-in-place. Ref: [1] https://issues.redhat.com/browse/MTV-1536 [2] https://issues.redhat.com/browse/MTV-1581 [3] https://access.redhat.com/articles/1351473 [4] kubev2v#1088 [5] kubev2v#1087 [6] kubev2v#1086 Signed-off-by: Martin Necas <[email protected]>
mnecas · Oct 18, 2024 · 75ac8a7 · 75ac8a7
1 parent d045987
commit 75ac8a7
Show file tree

Hide file tree

Showing 9 changed files with 81 additions and 263 deletions.
diff --git a/pkg/apis/forklift/v1beta1/BUILD.bazel b/pkg/apis/forklift/v1beta1/BUILD.bazel
@@ -23,7 +23,6 @@ go_library(
         "//pkg/apis/forklift/v1beta1/provider",
         "//pkg/apis/forklift/v1beta1/ref",
         "//pkg/lib/condition",
-        "//pkg/lib/error",
         "//vendor/k8s.io/api/core/v1:core",
         "//vendor/k8s.io/apimachinery/pkg/apis/meta/v1:meta",
         "//vendor/k8s.io/apimachinery/pkg/runtime",

diff --git a/pkg/apis/forklift/v1beta1/plan.go b/pkg/apis/forklift/v1beta1/plan.go
@@ -21,7 +21,6 @@ import (
 	"github.com/konveyor/forklift-controller/pkg/apis/forklift/v1beta1/provider"
 	"github.com/konveyor/forklift-controller/pkg/apis/forklift/v1beta1/ref"
 	libcnd "github.com/konveyor/forklift-controller/pkg/lib/condition"
-	liberr "github.com/konveyor/forklift-controller/pkg/lib/error"
 	core "k8s.io/api/core/v1"
 	meta "k8s.io/apimachinery/pkg/apis/meta/v1"
 )
@@ -93,30 +92,6 @@ type Plan struct {
 	Referenced `json:"-"`
 }
 
-// If the plan calls for the vm to be cold migrated to the local cluster, we can
-// just use virt-v2v directly to convert the vm while copying data over. In other
-// cases, we use CDI to transfer disks to the destination cluster and then use
-// virt-v2v-in-place to convert these disks after cutover.
-func (p *Plan) VSphereColdLocal() (bool, error) {
-	source := p.Referenced.Provider.Source
-	if source == nil {
-		return false, liberr.New("Cannot analyze plan, source provider is missing.")
-	}
-	destination := p.Referenced.Provider.Destination
-	if destination == nil {
-		return false, liberr.New("Cannot analyze plan, destination provider is missing.")
-	}
-
-	switch source.Type() {
-	case VSphere:
-		return !p.Spec.Warm && destination.IsHost(), nil
-	case Ova:
-		return true, nil
-	default:
-		return false, nil
-	}
-}
-
 // +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object
 type PlanList struct {
 	meta.TypeMeta `json:",inline"`
@@ -139,3 +114,7 @@ func (r *Plan) IsSourceProviderOvirt() bool {
 func (r *Plan) IsSourceProviderOCP() bool {
 	return r.Provider.Source.Type() == OpenShift
 }
+
+func (r *Plan) IsSourceProviderOVA() bool {
+	return r.Provider.Source.Type() == Ova
+}
diff --git a/pkg/controller/plan/adapter/vsphere/builder.go b/pkg/controller/plan/adapter/vsphere/builder.go
@@ -434,28 +434,16 @@ func (r *Builder) DataVolumes(vmRef ref.Ref, secret *core.Secret, _ *core.Config
 			if disk.Datastore.ID == ds.ID {
 				storageClass := mapped.Destination.StorageClass
 				var dvSource cdi.DataVolumeSource
-				coldLocal, vErr := r.Context.Plan.VSphereColdLocal()
-				if vErr != nil {
-					err = vErr
-					return
-				}
-				if coldLocal {
-					// Let virt-v2v do the copying
-					dvSource = cdi.DataVolumeSource{
-						Blank: &cdi.DataVolumeBlankImage{},
-					}
-				} else {
-					// Let CDI do the copying
-					dvSource = cdi.DataVolumeSource{
-						VDDK: &cdi.DataVolumeSourceVDDK{
-							BackingFile:  r.baseVolume(disk.File),
-							UUID:         vm.UUID,
-							URL:          url,
-							SecretRef:    secret.Name,
-							Thumbprint:   thumbprint,
-							InitImageURL: r.Source.Provider.Spec.Settings[api.VDDK],
-						},
-					}
+				// Let CDI do the copying
+				dvSource = cdi.DataVolumeSource{
+					VDDK: &cdi.DataVolumeSourceVDDK{
+						BackingFile:  r.baseVolume(disk.File),
+						UUID:         vm.UUID,
+						URL:          url,
+						SecretRef:    secret.Name,
+						Thumbprint:   thumbprint,
+						InitImageURL: r.Source.Provider.Spec.Settings[api.VDDK],
+					},
 				}
 				dvSpec := cdi.DataVolumeSpec{
 					Source: &dvSource,

diff --git a/pkg/controller/plan/adapter/vsphere/client.go b/pkg/controller/plan/adapter/vsphere/client.go
@@ -277,13 +277,6 @@ func (r *Client) getChangeIds(vmRef ref.Ref, snapshotId string, hosts util.Hosts
 }
 
 func (r *Client) getClient(vm *model.VM, hosts util.HostsFunc) (client *vim25.Client, err error) {
-	if coldLocal, vErr := r.Plan.VSphereColdLocal(); vErr == nil && coldLocal {
-		// when virt-v2v runs the migration, forklift-controller should interact only
-		// with the component that serves the SDK endpoint of the provider
-		client = r.client.Client
-		return
-	}
-
 	if r.Source.Provider.Spec.Settings[v1beta1.SDK] == v1beta1.ESXI {
 		// when migrating from ESXi host, we use the client of the SDK endpoint of the provider,
 		// there's no need in a different client (the ESXi host is the only component involved in the migration)

diff --git a/pkg/controller/plan/kubevirt.go b/pkg/controller/plan/kubevirt.go
@@ -824,18 +824,12 @@ func (r *KubeVirt) getListOptionsNamespaced() (listOptions *client.ListOptions)
 
 // Ensure the guest conversion (virt-v2v) pod exists on the destination.
 func (r *KubeVirt) EnsureGuestConversionPod(vm *plan.VMStatus, vmCr *VirtualMachine, pvcs []*core.PersistentVolumeClaim) (err error) {
-	labels := r.vmLabels(vm.Ref)
-	v2vSecret, err := r.ensureSecret(vm.Ref, r.secretDataSetterForCDI(vm.Ref), labels)
-	if err != nil {
-		return
-	}
-
 	configMap, err := r.ensureLibvirtConfigMap(vm.Ref, vmCr, pvcs)
 	if err != nil {
 		return
 	}
 
-	newPod, err := r.guestConversionPod(vm, vmCr.Spec.Template.Spec.Volumes, configMap, pvcs, v2vSecret)
+	newPod, err := r.guestConversionPod(vm, vmCr.Spec.Template.Spec.Volumes, configMap, pvcs)
 	if err != nil {
 		return
 	}
@@ -1667,7 +1661,7 @@ func (r *KubeVirt) findTemplate(vm *plan.VMStatus) (tmpl *template.Template, err
 	return
 }
 
-func (r *KubeVirt) guestConversionPod(vm *plan.VMStatus, vmVolumes []cnv.Volume, configMap *core.ConfigMap, pvcs []*core.PersistentVolumeClaim, v2vSecret *core.Secret) (pod *core.Pod, err error) {
+func (r *KubeVirt) guestConversionPod(vm *plan.VMStatus, vmVolumes []cnv.Volume, configMap *core.ConfigMap, pvcs []*core.PersistentVolumeClaim) (pod *core.Pod, err error) {
 	volumes, volumeMounts, volumeDevices, err := r.podVolumeMounts(vmVolumes, configMap, pvcs, vm)
 	if err != nil {
 		return
@@ -1684,34 +1678,6 @@ func (r *KubeVirt) guestConversionPod(vm *plan.VMStatus, vmVolumes []cnv.Volume,
 	user := qemuUser
 	nonRoot := true
 	allowPrivilageEscalation := false
-	// virt-v2v image
-	coldLocal, vErr := r.Context.Plan.VSphereColdLocal()
-	if vErr != nil {
-		err = vErr
-		return
-	}
-	if coldLocal {
-		// mount the secret for the password and CA certificate
-		volumes = append(volumes, core.Volume{
-			Name: "secret-volume",
-			VolumeSource: core.VolumeSource{
-				Secret: &core.SecretVolumeSource{
-					SecretName: v2vSecret.Name,
-				},
-			},
-		})
-		volumeMounts = append(volumeMounts, core.VolumeMount{
-			Name:      "secret-volume",
-			ReadOnly:  true,
-			MountPath: "/etc/secret",
-		})
-	} else {
-		environment = append(environment,
-			core.EnvVar{
-				Name:  "V2V_inPlace",
-				Value: "1",
-			})
-	}
 	// VDDK image
 	var initContainers []core.Container
 	if vddkImage, found := r.Source.Provider.Spec.Settings[api.VDDK]; found {
@@ -1808,19 +1774,9 @@ func (r *KubeVirt) guestConversionPod(vm *plan.VMStatus, vmVolumes []cnv.Volume,
 					Name:            "virt-v2v",
 					Env:             environment,
 					ImagePullPolicy: core.PullAlways,
-					EnvFrom: []core.EnvFromSource{
-						{
-							Prefix: "V2V_",
-							SecretRef: &core.SecretEnvSource{
-								LocalObjectReference: core.LocalObjectReference{
-									Name: v2vSecret.Name,
-								},
-							},
-						},
-					},
-					Image:         Settings.Migration.VirtV2vImage,
-					VolumeMounts:  volumeMounts,
-					VolumeDevices: volumeDevices,
+					Image:           Settings.Migration.VirtV2vImage,
+					VolumeMounts:    volumeMounts,
+					VolumeDevices:   volumeDevices,
 					Ports: []core.ContainerPort{
 						{
 							Name:          "metrics",

diff --git a/pkg/controller/plan/migration.go b/pkg/controller/plan/migration.go
@@ -47,7 +47,7 @@ var (
 	HasPostHook             libitr.Flag = 0x02
 	RequiresConversion      libitr.Flag = 0x04
 	CDIDiskCopy             libitr.Flag = 0x08
-	VirtV2vDiskCopy         libitr.Flag = 0x10
+	OvaImageMigration       libitr.Flag = 0x10
 	OpenstackImageMigration libitr.Flag = 0x20
 )
 
@@ -61,7 +61,6 @@ const (
 	CreateDataVolumes        = "CreateDataVolumes"
 	CreateVM                 = "CreateVM"
 	CopyDisks                = "CopyDisks"
-	AllocateDisks            = "AllocateDisks"
 	CopyingPaused            = "CopyingPaused"
 	AddCheckpoint            = "AddCheckpoint"
 	AddFinalCheckpoint       = "AddFinalCheckpoint"
@@ -84,7 +83,6 @@ const (
 const (
 	Initialize      = "Initialize"
 	Cutover         = "Cutover"
-	DiskAllocation  = "DiskAllocation"
 	DiskTransfer    = "DiskTransfer"
 	ImageConversion = "ImageConversion"
 	DiskTransferV2v = "DiskTransferV2v"
@@ -108,10 +106,9 @@ var (
 			{Name: WaitForPowerOff},
 			{Name: CreateDataVolumes},
 			{Name: CopyDisks, All: CDIDiskCopy},
-			{Name: AllocateDisks, All: VirtV2vDiskCopy},
 			{Name: CreateGuestConversionPod, All: RequiresConversion},
 			{Name: ConvertGuest, All: RequiresConversion},
-			{Name: CopyDisksVirtV2V, All: RequiresConversion},
+			{Name: CopyDisksVirtV2V, All: OvaImageMigration},
 			{Name: ConvertOpenstackSnapshot, All: OpenstackImageMigration},
 			{Name: CreateVM},
 			{Name: PostHook, All: HasPostHook},
@@ -643,8 +640,6 @@ func (r *Migration) step(vm *plan.VMStatus) (step string) {
 	switch vm.Phase {
 	case Started, CreateInitialSnapshot, WaitForInitialSnapshot, CreateDataVolumes:
 		step = Initialize
-	case AllocateDisks:
-		step = DiskAllocation
 	case CopyDisks, CopyingPaused, CreateSnapshot, WaitForSnapshot, AddCheckpoint, ConvertOpenstackSnapshot:
 		step = DiskTransfer
 	case CreateFinalSnapshot, WaitForFinalSnapshot, AddFinalCheckpoint, Finalize:
@@ -882,7 +877,7 @@ func (r *Migration) execute(vm *plan.VMStatus) (err error) {
 		step.MarkCompleted()
 		step.Phase = Completed
 		vm.Phase = r.next(vm.Phase)
-	case AllocateDisks, CopyDisks:
+	case CopyDisks:
 		step, found := vm.FindStep(r.step(vm))
 		if !found {
 			vm.AddError(fmt.Sprintf("Step '%s' not found", r.step(vm)))
@@ -1255,7 +1250,7 @@ func (r *Migration) buildPipeline(vm *plan.VM) (pipeline []*plan.Step, err error
 						Phase:       Pending,
 					},
 				})
-		case AllocateDisks, CopyDisks, CopyDisksVirtV2V, ConvertOpenstackSnapshot:
+		case CopyDisks, CopyDisksVirtV2V, ConvertOpenstackSnapshot:
 			tasks, pErr := r.builder.Tasks(vm.Ref)
 			if pErr != nil {
 				err = liberr.Wrap(pErr)
@@ -1270,9 +1265,6 @@ func (r *Migration) buildPipeline(vm *plan.VM) (pipeline []*plan.Step, err error
 			case CopyDisks:
 				task_name = DiskTransfer
 				task_description = "Transfer disks."
-			case AllocateDisks:
-				task_name = DiskAllocation
-				task_description = "Allocate disks."
 			case CopyDisksVirtV2V:
 				task_name = DiskTransferV2v
 				task_description = "Copy disks."
@@ -1658,11 +1650,7 @@ func (r *Migration) updateConversionProgress(vm *plan.VMStatus, step *plan.Step)
 			break
 		}
 
-		coldLocal, err := r.Context.Plan.VSphereColdLocal()
-		switch {
-		case err != nil:
-			return liberr.Wrap(err)
-		case coldLocal:
+		if r.Context.Plan.IsSourceProviderOVA() {
 			if err := r.updateConversionProgressV2vMonitor(pod, step); err != nil {
 				// Just log it. Missing progress is not fatal.
 				log.Error(err, "Failed to update conversion progress")
@@ -1853,23 +1841,17 @@ type Predicate struct {
 
 // Evaluate predicate flags.
 func (r *Predicate) Evaluate(flag libitr.Flag) (allowed bool, err error) {
-	coldLocal, vErr := r.context.Plan.VSphereColdLocal()
-	if vErr != nil {
-		err = vErr
-		return
-	}
-
 	switch flag {
 	case HasPreHook:
 		_, allowed = r.vm.FindHook(PreHook)
 	case HasPostHook:
 		_, allowed = r.vm.FindHook(PostHook)
 	case RequiresConversion:
 		allowed = r.context.Source.Provider.RequiresConversion()
+	case OvaImageMigration:
+		allowed = r.context.Plan.IsSourceProviderOVA()
 	case CDIDiskCopy:
-		allowed = !coldLocal
-	case VirtV2vDiskCopy:
-		allowed = coldLocal
+		allowed = !r.context.Plan.IsSourceProviderOVA()
 	case OpenstackImageMigration:
 		allowed = r.context.Plan.IsSourceProviderOpenstack()
 	}

diff --git a/pkg/controller/plan/scheduler/vsphere/scheduler.go b/pkg/controller/plan/scheduler/vsphere/scheduler.go
@@ -197,27 +197,15 @@ func (r *Scheduler) buildPending() (err error) {
 }
 
 func (r *Scheduler) cost(vm *model.VM, vmStatus *plan.VMStatus) int {
-	coldLocal, _ := r.Plan.VSphereColdLocal()
-	if coldLocal {
-		switch vmStatus.Phase {
-		case CreateVM, PostHook, Completed:
-			// In these phases we already have the disk transferred and are left only to create the VM
-			// By setting the cost to 0 other VMs can start migrating
-			return 0
-		default:
-			return 1
-		}
-	} else {
-		switch vmStatus.Phase {
-		case CreateVM, PostHook, Completed, CopyingPaused, ConvertGuest, CreateGuestConversionPod:
-			// The warm/remote migrations this is done on already transferred disks,
-			// and we can start other VM migrations at these point.
-			// By setting the cost to 0 other VMs can start migrating
-			return 0
-		default:
-			// CDI transfers the disks in parallel by different pods
-			return len(vm.Disks) - r.finishedDisks(vmStatus)
-		}
+	switch vmStatus.Phase {
+	case CreateVM, PostHook, Completed, CopyingPaused, ConvertGuest, CreateGuestConversionPod:
+		// The warm/remote migrations this is done on already transferred disks,
+		// and we can start other VM migrations at these point.
+		// By setting the cost to 0 other VMs can start migrating
+		return 0
+	default:
+		// CDI transfers the disks in parallel by different pods
+		return len(vm.Disks) - r.finishedDisks(vmStatus)
 	}
 }
 

diff --git a/pkg/forklift-api/webhooks/validating-webhook/admitters/plan-admitter.go b/pkg/forklift-api/webhooks/validating-webhook/admitters/plan-admitter.go
@@ -108,17 +108,6 @@ func (admitter *PlanAdmitter) validateLUKS() error {
 		log.Error(err, "Provider type (non-VSphere & non-OVA) does not support LUKS")
 		return err
 	}
-
-	coldLocal, vErr := admitter.plan.VSphereColdLocal()
-	if vErr != nil {
-		log.Error(vErr, "Could not analyze plan, failing")
-		return vErr
-	}
-	if !coldLocal {
-		err := liberr.New("migration of encrypted disks is not supported for warm migrations or migrations to remote providers")
-		log.Error(err, "Warm migration does not support LUKS")
-		return err
-	}
 	return nil
 }