Remove nv-peer-mem

Signed-off-by: Fred Rolland <frolland@nvidia.com>

README.md

@@ -92,9 +92,6 @@ NICClusterPolicy CRD Spec includes the following sub-states:
 and related configurations.
 - `sriovDevicePlugin`: [SR-IOV Network Device Plugin](https://github.com/k8snetworkplumbingwg/sriov-network-device-plugin)
     and related configurations.
-- `nvPeerDriver`: [NVIDIA Peer Memory client driver container](https://github.com/Mellanox/ofed-docker)
-to be deployed on RDMA & GPU supporting nodes (required for GPUDirect workloads).
-  For NVIDIA GPU driver version < 465. Check [compatibility notes](#compatibility-notes) for details
 - `ibKubernetes`: [InfiniBand Kubernetes](https://github.com/Mellanox/ib-kubernetes/) and related configurations. 
 - `secondaryNetwork`: Specifies components to deploy in order to facilitate a secondary network in Kubernetes. It consists of the following optionally deployed components:
     - [Multus-CNI](https://github.com/intel/multus-cni): Delegate CNI plugin to support secondary networks in Kubernetes
@@ -109,8 +106,7 @@ to be deployed on RDMA & GPU supporting nodes (required for GPUDirect workloads)
 
 
 ##### Example for NICClusterPolicy resource:
-In the example below we request OFED driver to be deployed together with RDMA shared device plugin
-but without NV Peer Memory driver.
+In the example below we request OFED driver to be deployed together with RDMA shared device plugin.
 
 ```
 apiVersion: mellanox.com/v1alpha1
@@ -263,8 +259,6 @@ status:
     state: ignore
   - name: state-RDMA-device-plugin
     state: ready
-  - name: state-NV-Peer
-    state: ignore
   - name: state-ib-kubernetes
     state: ignore
   - name: state-nv-ipam-cni
@@ -443,8 +437,7 @@ The following Network Adapters have been tested with NVIDIA Network Operator:
 
 ## Compatibility Notes
 * NVIDIA Network Operator is compatible with NVIDIA GPU Operator v1.5.2 and above
-* Network Operator will deploy nvPeerDriver POD on a node only if NVIDIA GPU driver version < 465.
-  Starting from v465 NVIDIA GPU driver includes a built-in nvidia_peermem module
+* Starting from v465 NVIDIA GPU driver includes a built-in nvidia_peermem module
   which is a replacement for nv_peer_mem module. NVIDIA GPU operator manages nvidia_peermem module loading.
 
 ## Deployment Example
@@ -480,7 +473,7 @@ making them available to the kernel.
 
 While this approach may seem odd. It provides a way to deliver drivers to immutable systems.
 
-[Mellanox OFED and NV Peer Memory driver container](https://github.com/Mellanox/ofed-docker)
+[Mellanox OFED container](https://github.com/Mellanox/ofed-docker)
 
 ## Upgrade
 Check [Upgrade section in Helm Chart documentation](deployment/network-operator/README.md#upgrade) for details.

api/v1alpha1/nicclusterpolicy_types.go

@@ -152,14 +152,6 @@ type DrainSpec struct {
 	DeleteEmptyDir bool `json:"deleteEmptyDir,omitempty"`
 }
 
-// NVPeerDriverSpec describes configuration options for NV Peer Memory driver
-type NVPeerDriverSpec struct {
-	// Image information for nv peer memory driver container
-	ImageSpec `json:""`
-	// GPU driver sources path - Optional
-	GPUDriverSourcePath string `json:"gpuDriverSourcePath,omitempty"`
-}
-
 // DevicePluginSpec describes configuration options for device plugin
 // 1. Image information for device plugin
 // 2. Device plugin configuration
@@ -227,7 +219,6 @@ type NicClusterPolicySpec struct {
 	NodeAffinity           *v1.NodeAffinity      `json:"nodeAffinity,omitempty"`
 	Tolerations            []v1.Toleration       `json:"tolerations,omitempty"`
 	OFEDDriver             *OFEDDriverSpec       `json:"ofedDriver,omitempty"`
-	NVPeerDriver           *NVPeerDriverSpec     `json:"nvPeerDriver,omitempty"`
 	RdmaSharedDevicePlugin *DevicePluginSpec     `json:"rdmaSharedDevicePlugin,omitempty"`
 	SriovDevicePlugin      *DevicePluginSpec     `json:"sriovDevicePlugin,omitempty"`
 	IBKubernetes           *IBKubernetesSpec     `json:"ibKubernetes,omitempty"`

config/crd/bases/mellanox.com_nicclusterpolicies.yaml

@@ -299,31 +299,6 @@ spec:
                 - repository
                 - version
                 type: object
-              nvPeerDriver:
-                description: NVPeerDriverSpec describes configuration options for
-                  NV Peer Memory driver
-                properties:
-                  gpuDriverSourcePath:
-                    description: GPU driver sources path - Optional
-                    type: string
-                  image:
-                    pattern: '[a-zA-Z0-9\-]+'
-                    type: string
-                  imagePullSecrets:
-                    items:
-                      type: string
-                    type: array
-                  repository:
-                    pattern: '[a-zA-Z0-9\.\-\/]+'
-                    type: string
-                  version:
-                    pattern: '[a-zA-Z0-9\.-]+'
-                    type: string
-                required:
-                - image
-                - repository
-                - version
-                type: object
               ofedDriver:
                 description: OFEDDriverSpec describes configuration options for OFED
                   driver

controllers/nicclusterpolicy_controller.go

@@ -109,8 +109,8 @@ func (r *NicClusterPolicyReconciler) Reconcile(ctx context.Context, req ctrl.Req
 
 	// Create a new State service catalog
 	sc := state.NewInfoCatalog()
-	if instance.Spec.OFEDDriver != nil || instance.Spec.NVPeerDriver != nil ||
-		instance.Spec.RdmaSharedDevicePlugin != nil || instance.Spec.SriovDevicePlugin != nil {
+	if instance.Spec.OFEDDriver != nil || instance.Spec.RdmaSharedDevicePlugin != nil ||
+		instance.Spec.SriovDevicePlugin != nil {
 		// Create node infoProvider and add to the service catalog
 		reqLogger.V(consts.LogLevelInfo).Info("Creating Node info provider")
 		nodeList := &corev1.NodeList{}

deployment/network-operator/README.md

@@ -380,17 +380,6 @@ imagePullSecrets:
 | `ofedDriver.readinessProbe.initialDelaySeconds` | int | 10 | Mellanox OFED readiness probe initial delay                                                                                                                               |
 | `ofedDriver.readinessProbe.periodSeconds` | int | 30 | Mellanox OFED readiness probe interval                                                                                                                                    |
 
-#### NVIDIA Peer memory driver
-
-| Name | Type | Default | description |
-| ---- | ---- | ------- | ----------- |
-| `nvPeerDriver.deploy` | bool | `false` | deploy NVIDIA Peer memory driver container |
-| `nvPeerDriver.repository` | string | `mellanox` | NVIDIA Peer memory driver image repository |
-| `nvPeerDriver.image` | string | `nv-peer-mem-driver` | NVIDIA Peer memory driver image name  |
-| `nvPeerDriver.version` | string | `1.1-0` | NVIDIA Peer memory driver version  |
-| `nvPeerDriver.imagePullSecrets` | list | `[]` | An optional list of references to secrets to use for pulling any of the NVIDIA Peer memory driver image |
-| `nvPeerDriver.gpuDriverSourcePath` | string | `/run/nvidia/driver` | GPU driver soruces root filesystem path(usually used in tandem with [gpu-operator](https://github.com/NVIDIA/gpu-operator)) |
-
 #### RDMA Device Plugin
 
 | Name | Type | Default | description |
@@ -595,7 +584,7 @@ rdmaSharedDevicePlugin:
 
 #### Example 2
 
-Network Operator deployment with the default version of OFED and NV Peer mem driver, RDMA device plugin with two RDMA
+Network Operator deployment with the default version of OFED, RDMA device plugin with two RDMA
 resources, the first mapped to `enp1` and `enp2`, the second mapped to `ib0`.
 
 __values.yaml:__
@@ -604,8 +593,6 @@ __values.yaml:__
 deployCR: true
 ofedDriver:
   deploy: true
-nvPeerDriver:
-  deploy: true
 rdmaSharedDevicePlugin:
   deploy: true
   resources:

deployment/network-operator/crds/mellanox.com_nicclusterpolicies.yaml

@@ -299,31 +299,6 @@ spec:
                 - repository
                 - version
                 type: object
-              nvPeerDriver:
-                description: NVPeerDriverSpec describes configuration options for
-                  NV Peer Memory driver
-                properties:
-                  gpuDriverSourcePath:
-                    description: GPU driver sources path - Optional
-                    type: string
-                  image:
-                    pattern: '[a-zA-Z0-9\-]+'
-                    type: string
-                  imagePullSecrets:
-                    items:
-                      type: string
-                    type: array
-                  repository:
-                    pattern: '[a-zA-Z0-9\.\-\/]+'
-                    type: string
-                  version:
-                    pattern: '[a-zA-Z0-9\.-]+'
-                    type: string
-                required:
-                - image
-                - repository
-                - version
-                type: object
               ofedDriver:
                 description: OFEDDriverSpec describes configuration options for OFED
                   driver

deployment/network-operator/templates/_helpers.tpl

@@ -101,22 +101,6 @@ imagePullSecrets helpers
 {{- $imagePullSecrets | toJson }}
 {{- end }}
 
-{{- define "network-operator.nvPeerDriver.imagePullSecrets" }}
-{{- $imagePullSecrets := list }}
-{{- if .Values.nvPeerDriver.imagePullSecrets }}
-{{- range .Values.nvPeerDriver.imagePullSecrets }}
-{{- $imagePullSecrets  = append $imagePullSecrets  . }}
-{{- end }}
-{{- else }}
-{{- if .Values.imagePullSecrets }}
-{{- range .Values.imagePullSecrets }}
-{{- $imagePullSecrets  = append $imagePullSecrets  . }}
-{{- end }}
-{{- end }}
-{{- end }}
-{{- $imagePullSecrets | toJson }}
-{{- end }}
-
 {{- define "network-operator.rdmaSharedDevicePlugin.imagePullSecrets" }}
 {{- $imagePullSecrets := list }}
 {{- if .Values.rdmaSharedDevicePlugin.imagePullSecrets }}

deployment/network-operator/templates/mellanox.com_v1alpha1_nicclusterpolicy_cr.yaml

@@ -67,14 +67,6 @@ spec:
         deleteEmptyDir: {{ .Values.ofedDriver.upgradePolicy.drain.deleteEmptyDir | default false}}
     {{- end }}
   {{- end }}
-  {{- if .Values.nvPeerDriver.deploy }}
-  nvPeerDriver:
-    image: {{ .Values.nvPeerDriver.image }}
-    repository: {{ .Values.nvPeerDriver.repository }}
-    version: {{ .Values.nvPeerDriver.version }}
-    imagePullSecrets: {{ include "network-operator.nvPeerDriver.imagePullSecrets" . }}
-    gpuDriverSourcePath: {{ .Values.nvPeerDriver.gpuDriverSourcePath }}
-  {{- end }}
   {{- if .Values.rdmaSharedDevicePlugin.deploy }}
   rdmaSharedDevicePlugin:
     # {{ required "A valid value for .Values.rdmaSharedDevicePlugin.resources is required" .Values.rdmaSharedDevicePlugin.resources }}

deployment/network-operator/values.yaml

@@ -198,14 +198,6 @@ ofedDriver:
       timeoutSeconds: 300
       deleteEmptyDir: false
 
-nvPeerDriver:
-  deploy: false
-  image: nv-peer-mem-driver
-  repository: mellanox
-  version: 1.1-0
-  # imagePullSecrets: []
-  gpuDriverSourcePath: /run/nvidia/driver
-
 rdmaSharedDevicePlugin:
   deploy: true
   image: k8s-rdma-shared-dev-plugin

example/README.md

@@ -8,7 +8,6 @@ We assume familiarity with RDMA, Kubernetes and related CNI project.
 network-operator, at this stage, deploys and configures the follwoing components:
 * [Mellanox OFED](https://www.mellanox.com/products/infiniband-drivers/linux/mlnx_ofed) driver container
 * [RDMA device plugin](https://github.com/Mellanox/k8s-rdma-shared-dev-plugin)
-* [NVIDIA peer memory client](https://github.com/Mellanox/nv_peer_memory) driver container
 * SecondaryNetwork`: Specifies components to deploy in order to facilitate a secondary network in Kubernetes. It consists of the following optionally deployed components:
     - [Multus-CNI](https://github.com/intel/multus-cni): Delegate CNI plugin to support secondary networks in Kubernetes
     - CNI plugins: Currently only [containernetworking-plugins](https://github.com/containernetworking/plugins) is supported

example/dgx-helm-values.yaml

@@ -14,7 +14,7 @@
 
 # Custom Helm Chart values example to deploy Network Operator
 # at NVIDIA DGX SYSTEMS.
-# We don't need to deploy MOFED and nv-peer-mem containers.
+# We don't need to deploy MOFED container.
 
 nfd:
   enabled: true
@@ -24,14 +24,6 @@ deployCR: true
 ofedDriver:
   deploy: false
 
-nvPeerDriver:
-  deploy: false
-  image: nv-peer-mem-driver
-  repository: mellanox
-  version: 1.1-0
-  imagePullSecrets: []
-  gpuDriverSourcePath: /run/nvidia/driver
-
 secondaryNetwork:
   deploy: true
   cniPlugins:

hack/templates/values/values.template

@@ -198,14 +198,6 @@ ofedDriver:
       timeoutSeconds: 300
       deleteEmptyDir: false
 
-nvPeerDriver:
-  deploy: false
-  image: nv-peer-mem-driver
-  repository: mellanox
-  version: 1.1-0
-  # imagePullSecrets: []
-  gpuDriverSourcePath: /run/nvidia/driver
-
 rdmaSharedDevicePlugin:
   deploy: true
   image: {{ .RdmaSharedDevicePlugin.Image }}