Merge pull request #174 from NearNodeFlash/release-v0.1.1

Release v0.1.1

Dockerfile

@@ -18,7 +18,7 @@
 # These ARGs must be before the first FROM. This allows them to be valid for
 # use in FROM instructions.
 ARG NNFMFU_TAG_BASE=ghcr.io/nearnodeflash/nnf-mfu
-ARG NNFMFU_VERSION=0.0.3
+ARG NNFMFU_VERSION=0.1.0
 
 # Build the manager binary
 FROM golang:1.19-alpine as builder

Makefile

@@ -55,7 +55,7 @@ IMAGE_TAG_BASE ?= ghcr.io/nearnodeflash/nnf-dm
 
 # The NNF-MFU container image to use in NNFContainerProfile resources.
 NNFMFU_TAG_BASE ?= ghcr.io/nearnodeflash/nnf-mfu
-NNFMFU_VERSION ?= 0.0.3
+NNFMFU_VERSION ?= 0.1.0
 
 DOCKER_BUILDARGS=--build-arg NNFMFU_TAG_BASE=$(NNFMFU_TAG_BASE) --build-arg NNFMFU_VERSION=$(NNFMFU_VERSION)
 

config/manager/kustomization.yaml

@@ -16,7 +16,7 @@ kind: Kustomization
 images:
 - name: controller
   newName: ghcr.io/nearnodeflash/nnf-dm
-  newTag: 0.1.0
+  newTag: 0.1.1
 - name: nnf-mfu
   newName: ghcr.io/nearnodeflash/nnf-mfu
-  newTag: 0.0.3
+  newTag: 0.1.0

config/manager/manager.yaml

@@ -64,7 +64,7 @@ spec:
       maxUnavailable: '25%'
   template:
     spec:
-      serviceAccountName: nnf-dm-controller-manager # not sure why 'controller-manager' isn't patched to include the 'nnf-dm-' prefix
+      serviceAccountName: nnf-dm-node-controller
       tolerations:
         - key: 'cray.nnf.node'
           operator: 'Equal'

config/rbac/daemon_role.yaml

@@ -12,6 +12,13 @@ rules:
   - get
   - list
   - watch
+- apiGroups:
+  - dataworkflowservices.github.io
+  resources:
+  - workflows
+  verbs:
+  - get
+  - list
 - apiGroups:
   - dataworkflowservices.github.io
   resources:

config/rbac/role.yaml

@@ -162,6 +162,14 @@ rules:
   - get
   - patch
   - update
+- apiGroups:
+  - nnf.cray.hpe.com
+  resources:
+  - nnfnodestorages
+  verbs:
+  - get
+  - list
+  - watch
 - apiGroups:
   - nnf.cray.hpe.com
   resources:

config/rbac/role_binding.yaml

@@ -9,4 +9,17 @@ roleRef:
 subjects:
 - kind: ServiceAccount
   name: controller-manager
-  namespace: system
+  namespace: system
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRoleBinding
+metadata:
+  name: node-manager-rolebinding
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: ClusterRole
+  name: manager-role
+subjects:
+- kind: ServiceAccount
+  name: node-controller
+  namespace: system

config/rbac/service_account.yaml

@@ -15,4 +15,20 @@ metadata:
   annotations:
     kubernetes.io/service-account.name: controller-manager
     kubernetes.io/service-account.namespace: system
-type: kubernetes.io/service-account-token
+type: kubernetes.io/service-account-token
+---
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: node-controller
+  namespace: system
+---
+apiVersion: v1
+kind: Secret
+metadata:
+  name: node-controller
+  namespace: system
+  annotations:
+    kubernetes.io/service-account.name: node-controller
+    kubernetes.io/service-account.namespace: system
+type: kubernetes.io/service-account-token

daemons/compute/client-go/main.go

@@ -181,20 +181,23 @@ func main() {
 	// and couldn't recover the data movement request). The NNF Data Movement Workflow ensures that these requests are
 	// deleted.
 	if !*skipDelete {
-		// Use List to cleanup and delete requests
-		for _, uid := range listResponse.GetUids() {
+		// Delete each DM this client created
+		for _, resp := range responses {
+			uid := resp.GetUid()
 			log.Printf("Deleting request: %v", uid)
 
 			deleteResponse, err := deleteRequest(ctx, c, *workflow, *namespace, uid)
 			if err != nil {
 				log.Fatalf("could not delete data movement request: %s", err)
 			}
 
-			if deleteResponse.Status != pb.DataMovementDeleteResponse_SUCCESS {
-				log.Fatalf("data movement delete failed: %+v", deleteResponse)
+			if deleteResponse.Status == pb.DataMovementDeleteResponse_NOT_FOUND {
+				log.Printf("Data movement request deleted (not found): %v %v", uid, deleteResponse.String())
+			} else if deleteResponse.Status != pb.DataMovementDeleteResponse_SUCCESS {
+				log.Fatalf("data movement delete failed: %v %+v", uid, deleteResponse)
+			} else {
+				log.Printf("Data movement request deleted: %v %v", uid, deleteResponse.String())
 			}
-
-			log.Printf("Data movement request deleted: %v %v", uid, deleteResponse.String())
 		}
 
 		// Print out the list again to verify deletes
@@ -288,7 +291,8 @@ func deleteRequest(ctx context.Context, client pb.DataMoverClient, workflow stri
 		Uid: uid,
 	})
 
-	if err != nil {
+	// It's already been deleted if it's not found
+	if err != nil && rsp.Status != pb.DataMovementDeleteResponse_NOT_FOUND {
 		return nil, err
 	}
 

daemons/compute/server/servers/server_default.go

@@ -295,6 +295,25 @@ func (*defaultServer) Version(context.Context, *emptypb.Empty) (*pb.DataMovement
 
 func (s *defaultServer) Create(ctx context.Context, req *pb.DataMovementCreateRequest) (*pb.DataMovementCreateResponse, error) {
 
+	// Ensure workflow exists and is in PreRun:Ready
+	workflow := &dwsv1alpha2.Workflow{ObjectMeta: metav1.ObjectMeta{
+		Name:      req.Workflow.Name,
+		Namespace: req.Workflow.Namespace,
+	}}
+	if err := s.client.Get(ctx, client.ObjectKeyFromObject(workflow), workflow); err != nil {
+		return &pb.DataMovementCreateResponse{
+			Status:  pb.DataMovementCreateResponse_FAILED,
+			Message: "Could find matching workflow: " + err.Error(),
+		}, nil
+	}
+
+	if workflow.Status.State != dwsv1alpha2.StatePreRun || workflow.Status.Status != "Completed" {
+		return &pb.DataMovementCreateResponse{
+			Status:  pb.DataMovementCreateResponse_FAILED,
+			Message: fmt.Sprintf("Workflow must be in '%s' state and 'Completed' status", dwsv1alpha2.StatePreRun),
+		}, nil
+	}
+
 	computeClientMount, computeMountInfo, err := s.findComputeMountInfo(ctx, req)
 	if err != nil {
 		return &pb.DataMovementCreateResponse{
@@ -313,6 +332,9 @@ func (s *defaultServer) Create(ctx context.Context, req *pb.DataMovementCreateRe
 		dm, err = s.createNnfNodeDataMovement(ctx, req, computeMountInfo)
 		dmFunc = "createNnfNodeDataMovement()"
 	default:
+		// xfs is not supported since it can only be mounted in one location at a time. It is
+		// already mounted on the compute node when copy offload occurs (PreRun/PostRun), therefore
+		// it cannot be mounted on the rabbit to perform data movement.
 		return &pb.DataMovementCreateResponse{
 			Status:  pb.DataMovementCreateResponse_INVALID,
 			Message: fmt.Sprintf("filesystem not supported: '%s'", computeMountInfo.Type),
@@ -338,16 +360,9 @@ func (s *defaultServer) Create(ctx context.Context, req *pb.DataMovementCreateRe
 	dm.Spec.UserId = userId
 	dm.Spec.GroupId = groupId
 
-	// We don't have the actual NnfDataMovement parent available, but we know the name
-	// and the namespace because they will match the workflow's name and namespace.
-	parentDm := &nnfv1alpha1.NnfDataMovement{
-		ObjectMeta: metav1.ObjectMeta{
-			Name:      req.Workflow.Name,
-			Namespace: req.Workflow.Namespace,
-		},
-	}
-
-	dwsv1alpha2.AddOwnerLabels(dm, parentDm)
+	// Add appropriate workflow labels so this is cleaned up
+	dwsv1alpha2.AddWorkflowLabels(dm, workflow)
+	dwsv1alpha2.AddOwnerLabels(dm, workflow)
 
 	// Label the NnfDataMovement with a teardown state of "post_run" so the NNF workflow
 	// controller can identify compute initiated data movements.
@@ -470,7 +485,8 @@ func (s *defaultServer) createNnfNodeDataMovement(ctx context.Context, req *pb.D
 		},
 		Spec: nnfv1alpha1.NnfDataMovementSpec{
 			Source: &nnfv1alpha1.NnfDataMovementSpecSourceDestination{
-				Path: source,
+				Path:             source,
+				StorageReference: computeMountInfo.Device.DeviceReference.ObjectReference,
 			},
 			Destination: &nnfv1alpha1.NnfDataMovementSpecSourceDestination{
 				Path: req.Destination,

go.mod

@@ -3,8 +3,8 @@ module github.com/NearNodeFlash/nnf-dm
 go 1.19
 
 require (
-	github.com/NearNodeFlash/lustre-fs-operator v0.0.1-0.20240220213720-51597bca637d
-	github.com/NearNodeFlash/nnf-sos v0.0.1-0.20240221214302-e7989177289a
+	github.com/NearNodeFlash/lustre-fs-operator v0.0.1-0.20240326175906-15cfe803227d
+	github.com/NearNodeFlash/nnf-sos v0.0.1-0.20240501135550-002d992157a9
 	github.com/onsi/ginkgo/v2 v2.11.0
 	github.com/onsi/gomega v1.27.10
 	github.com/prometheus/client_golang v1.16.0
@@ -22,12 +22,12 @@ require (
 )
 
 require (
-	github.com/NearNodeFlash/nnf-ec v0.0.0-20231010162453-a8168bb6a52f // indirect
+	github.com/NearNodeFlash/nnf-ec v0.0.1-0.20240318141758-e8ded5e13eb8 // indirect
 	github.com/beorn7/perks v1.0.1 // indirect
 	github.com/cespare/xxhash/v2 v2.2.0 // indirect
 	github.com/davecgh/go-spew v1.1.1 // indirect
 	github.com/fsnotify/fsnotify v1.6.0 // indirect
-	github.com/go-logr/logr v1.2.4 // indirect
+	github.com/go-logr/logr v1.2.4
 	github.com/go-logr/zapr v1.2.4 // indirect
 	github.com/go-openapi/jsonpointer v0.19.6 // indirect
 	github.com/go-openapi/jsonreference v0.20.2 // indirect
@@ -73,7 +73,7 @@ require (
 )
 
 require (
-	github.com/DataWorkflowServices/dws v0.0.1-0.20240221183421-1a123a9274b6
+	github.com/DataWorkflowServices/dws v0.0.1-0.20240423152131-d92c9aadede8
 	go.openly.dev/pointy v1.3.0
 )
 

go.sum

@@ -1,11 +1,11 @@
-github.com/DataWorkflowServices/dws v0.0.1-0.20240221183421-1a123a9274b6 h1:LYKIIoawsuo+1ByvQaIpgl8vZc2KrE0q7AE7t0YumrI=
-github.com/DataWorkflowServices/dws v0.0.1-0.20240221183421-1a123a9274b6/go.mod h1:vSTBLWbsFjMYxx+sjMDyZpMXLY9m5Bp73cjnmAL30WU=
-github.com/NearNodeFlash/lustre-fs-operator v0.0.1-0.20240220213720-51597bca637d h1:AP1TgQlneYZT/AxkYFyvJp1j86+7MTYOoo3I1Zw3L2E=
-github.com/NearNodeFlash/lustre-fs-operator v0.0.1-0.20240220213720-51597bca637d/go.mod h1:qBcz9p8sXm1qhDf8WUmhxTlD1NCMEjoAD7NoHbQvMiI=
-github.com/NearNodeFlash/nnf-ec v0.0.0-20231010162453-a8168bb6a52f h1:aWtSSQLLk9mUZj94mowirQeVw9saf80gVe10X0rZe8o=
-github.com/NearNodeFlash/nnf-ec v0.0.0-20231010162453-a8168bb6a52f/go.mod h1:oxdwMqfttOF9dabJhqrWlirCnMk8/8eyLMwl+hducjk=
-github.com/NearNodeFlash/nnf-sos v0.0.1-0.20240221214302-e7989177289a h1:m3lPHiWObITk+zp3GAuOeawkaMr+U4aKlHcblYnsP58=
-github.com/NearNodeFlash/nnf-sos v0.0.1-0.20240221214302-e7989177289a/go.mod h1:4kJuGEwS46EYIt24NmNweaMryYp8M6RabNc8Nd+GIBE=
+github.com/DataWorkflowServices/dws v0.0.1-0.20240423152131-d92c9aadede8 h1:Ic1rErr2VFbIfGYx8loePH7HRqPMQEW5ZIDiOP2CMLk=
+github.com/DataWorkflowServices/dws v0.0.1-0.20240423152131-d92c9aadede8/go.mod h1:vSTBLWbsFjMYxx+sjMDyZpMXLY9m5Bp73cjnmAL30WU=
+github.com/NearNodeFlash/lustre-fs-operator v0.0.1-0.20240326175906-15cfe803227d h1:gqAZOwvFrsgsAQYLqSnMBU4aALOH7+QpqYruhhI6MZU=
+github.com/NearNodeFlash/lustre-fs-operator v0.0.1-0.20240326175906-15cfe803227d/go.mod h1:qBcz9p8sXm1qhDf8WUmhxTlD1NCMEjoAD7NoHbQvMiI=
+github.com/NearNodeFlash/nnf-ec v0.0.1-0.20240318141758-e8ded5e13eb8 h1:Ja+ZQVGl/+buQXGqKFsM5mDwX4ReiI5UkputLsaiRSo=
+github.com/NearNodeFlash/nnf-ec v0.0.1-0.20240318141758-e8ded5e13eb8/go.mod h1:oxdwMqfttOF9dabJhqrWlirCnMk8/8eyLMwl+hducjk=
+github.com/NearNodeFlash/nnf-sos v0.0.1-0.20240501135550-002d992157a9 h1:IS42BQNDBmdFDreU+LpMcjj/Cai4iqDEmSKgadUc9Q0=
+github.com/NearNodeFlash/nnf-sos v0.0.1-0.20240501135550-002d992157a9/go.mod h1:/q/cLHgca7wBy0QAbcj80ANWuWjmNLrLD39xbEwi0tQ=
 github.com/benbjohnson/clock v1.1.0/go.mod h1:J11/hYXuz8f4ySSvYwY0FKfm+ezbsZBKZxNJlLklBHA=
 github.com/benbjohnson/clock v1.3.0 h1:ip6w0uFQkncKQ979AypyG0ER7mqUSBdKLOgAle/AT8A=
 github.com/beorn7/perks v1.0.1 h1:VlbKKnNfV8bJzeqoa4cOKqO6bYr3WgKZxO8Z16+hsOM=