1. Documentation
×

Using DPDK and RDMA

Using a virtual function in DPDK mode with an Intel NIC

Prerequisites

  • Install the OpenShift CLI (oc).

  • Install the SR-IOV Network Operator.

  • Log in as a user with cluster-admin privileges.

Procedure

  1. Create the following SriovNetworkNodePolicy object, and then save the YAML in the intel-dpdk-node-policy.yaml file.

    apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetworkNodePolicy
metadata:
  name: intel-dpdk-node-policy
  namespace: openshift-sriov-network-operator
spec:
  resourceName: intelnics
  nodeSelector:
    feature.node.kubernetes.io/network-sriov.capable: "true"
  priority: <priority>
  numVfs: <num>
  nicSelector:
    vendor: "8086"
    deviceID: "158b"
    pfNames: ["<pf_name>", ...]
    rootDevices: ["<pci_bus_id>", "..."]
  deviceType: vfio-pci (1)
    1 Specify the driver type for the virtual functions to vfio-pci.

    See the Configuring SR-IOV network devices section for a detailed explanation on each option in SriovNetworkNodePolicy.

    When applying the configuration specified in a SriovNetworkNodePolicy object, the SR-IOV Operator may drain the nodes, and in some cases, reboot nodes. It may take several minutes for a configuration change to apply. Ensure that there are enough available nodes in your cluster to handle the evicted workload beforehand.

    After the configuration update is applied, all the pods in openshift-sriov-network-operator namespace will change to a Running status.

  2. Create the SriovNetworkNodePolicy object by running the following command:

    $ oc create -f intel-dpdk-node-policy.yaml

  3. Create the following SriovNetwork object, and then save the YAML in the intel-dpdk-network.yaml file.

    apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetwork
metadata:
  name: intel-dpdk-network
  namespace: openshift-sriov-network-operator
spec:
  networkNamespace: <target_namespace>
  ipam: |-
# ... (1)
  vlan: <vlan>
  resourceName: intelnics
    1 Specify a configuration object for the ipam CNI plugin as a YAML block scalar. The plugin manages IP address assignment for the attachment definition.

    See the "Configuring SR-IOV additional network" section for a detailed explanation on each option in SriovNetwork.

    An optional library, app-netutil, provides several API methods for gathering network information about a container’s parent pod.

  4. Create the SriovNetwork object by running the following command:

    $ oc create -f intel-dpdk-network.yaml

  5. Create the following Pod spec, and then save the YAML in the intel-dpdk-pod.yaml file.

    apiVersion: v1
kind: Pod
metadata:
  name: dpdk-app
  namespace: <target_namespace> (1)
  annotations:
    k8s.v1.cni.cncf.io/networks: intel-dpdk-network
spec:
  containers:
  - name: testpmd
    image: <DPDK_image> (2)
    securityContext:
      runAsUser: 0
      capabilities:
        add: ["IPC_LOCK","SYS_RESOURCE","NET_RAW"] (3)
    volumeMounts:
    - mountPath: /mnt/huge (4)
      name: hugepage
    resources:
      limits:
        openshift.io/intelnics: "1" (5)
        memory: "1Gi"
        cpu: "4" (6)
        hugepages-1Gi: "4Gi" (7)
      requests:
        openshift.io/intelnics: "1"
        memory: "1Gi"
        cpu: "4"
        hugepages-1Gi: "4Gi"
    command: ["sleep", "infinity"]
  volumes:
  - name: hugepage
    emptyDir:
      medium: HugePages
    1 Specify the same target_namespace where the SriovNetwork object intel-dpdk-network is created. If you would like to create the pod in a different namespace, change target_namespace in both the Pod spec and the SriovNetwork object.
    2 Specify the DPDK image which includes your application and the DPDK library used by application.
    3 Specify additional capabilities required by the application inside the container for hugepage allocation, system resource allocation, and network interface access.
    4 Mount a hugepage volume to the DPDK pod under /mnt/huge. The hugepage volume is backed by the emptyDir volume type with the medium being Hugepages.
    5 Optional: Specify the number of DPDK devices allocated to DPDK pod. This resource request and limit, if not explicitly specified, will be automatically added by the SR-IOV network resource injector. The SR-IOV network resource injector is an admission controller component managed by the SR-IOV Operator. It is enabled by default and can be disabled by setting enableInjector option to false in the default SriovOperatorConfig CR.
    6 Specify the number of CPUs. The DPDK pod usually requires exclusive CPUs to be allocated from the kubelet. This is achieved by setting CPU Manager policy to static and creating a pod with Guaranteed QoS.
    7 Specify hugepage size hugepages-1Gi or hugepages-2Mi and the quantity of hugepages that will be allocated to the DPDK pod. Configure 2Mi and 1Gi hugepages separately. Configuring 1Gi hugepage requires adding kernel arguments to Nodes. For example, adding kernel arguments default_hugepagesz=1GB, hugepagesz=1G and hugepages=16 will result in 16*1Gi hugepages be allocated during system boot.

  6. Create the DPDK pod by running the following command:

    $ oc create -f intel-dpdk-pod.yaml

Using a virtual function in DPDK mode with a Mellanox NIC

You can create a network node policy and create a Data Plane Development Kit (DPDK) pod using a virtual function in DPDK mode with a Mellanox NIC.

Prerequisites

  • You have installed the OpenShift CLI (oc).

  • You have installed the Single Root I/O Virtualization (SR-IOV) Network Operator.

  • You have logged in as a user with cluster-admin privileges.

Procedure

  1. Save the following SriovNetworkNodePolicy YAML configuration to an mlx-dpdk-node-policy.yaml file:

    apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetworkNodePolicy
metadata:
  name: mlx-dpdk-node-policy
  namespace: openshift-sriov-network-operator
spec:
  resourceName: mlxnics
  nodeSelector:
    feature.node.kubernetes.io/network-sriov.capable: "true"
  priority: <priority>
  numVfs: <num>
  nicSelector:
    vendor: "15b3"
    deviceID: "1015" (1)
    pfNames: ["<pf_name>", ...]
    rootDevices: ["<pci_bus_id>", "..."]
  deviceType: netdevice (2)
  isRdma: true (3)
    1 Specify the device hex code of the SR-IOV network device.
    2 Specify the driver type for the virtual functions to netdevice. A Mellanox SR-IOV Virtual Function (VF) can work in DPDK mode without using the vfio-pci device type. The VF device appears as a kernel network interface inside a container.
    3 Enable Remote Direct Memory Access (RDMA) mode. This is required for Mellanox cards to work in DPDK mode.

    See Configuring an SR-IOV network device for a detailed explanation of each option in the SriovNetworkNodePolicy object.

    When applying the configuration specified in an SriovNetworkNodePolicy object, the SR-IOV Operator might drain the nodes, and in some cases, reboot nodes. It might take several minutes for a configuration change to apply. Ensure that there are enough available nodes in your cluster to handle the evicted workload beforehand.

    After the configuration update is applied, all the pods in the openshift-sriov-network-operator namespace will change to a Running status.

  2. Create the SriovNetworkNodePolicy object by running the following command:

    $ oc create -f mlx-dpdk-node-policy.yaml

  3. Save the following SriovNetwork YAML configuration to an mlx-dpdk-network.yaml file:

    apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetwork
metadata:
  name: mlx-dpdk-network
  namespace: openshift-sriov-network-operator
spec:
  networkNamespace: <target_namespace>
  ipam: |- (1)
...
  vlan: <vlan>
  resourceName: mlxnics
    1 Specify a configuration object for the IP Address Management (IPAM) Container Network Interface (CNI) plugin as a YAML block scalar. The plugin manages IP address assignment for the attachment definition.

    See Configuring an SR-IOV network device for a detailed explanation on each option in the SriovNetwork object.

    The app-netutil option library provides several API methods for gathering network information about the parent pod of a container.

  4. Create the SriovNetwork object by running the following command:

    $ oc create -f mlx-dpdk-network.yaml

  5. Save the following Pod YAML configuration to an mlx-dpdk-pod.yaml file:

    apiVersion: v1
kind: Pod
metadata:
  name: dpdk-app
  namespace: <target_namespace> (1)
  annotations:
    k8s.v1.cni.cncf.io/networks: mlx-dpdk-network
spec:
  containers:
  - name: testpmd
    image: <DPDK_image> (2)
    securityContext:
      runAsUser: 0
      capabilities:
        add: ["IPC_LOCK","SYS_RESOURCE","NET_RAW"] (3)
    volumeMounts:
    - mountPath: /mnt/huge (4)
      name: hugepage
    resources:
      limits:
        openshift.io/mlxnics: "1" (5)
        memory: "1Gi"
        cpu: "4" (6)
        hugepages-1Gi: "4Gi" (7)
      requests:
        openshift.io/mlxnics: "1"
        memory: "1Gi"
        cpu: "4"
        hugepages-1Gi: "4Gi"
    command: ["sleep", "infinity"]
  volumes:
  - name: hugepage
    emptyDir:
      medium: HugePages
    1 Specify the same target_namespace where SriovNetwork object mlx-dpdk-network is created. To create the pod in a different namespace, change target_namespace in both the Pod spec and SriovNetwork object.
    2 Specify the DPDK image which includes your application and the DPDK library used by the application.
    3 Specify additional capabilities required by the application inside the container for hugepage allocation, system resource allocation, and network interface access.
    4 Mount the hugepage volume to the DPDK pod under /mnt/huge. The hugepage volume is backed by the emptyDir volume type with the medium being Hugepages.
    5 Optional: Specify the number of DPDK devices allocated for the DPDK pod. If not explicitly specified, this resource request and limit is automatically added by the SR-IOV network resource injector. The SR-IOV network resource injector is an admission controller component managed by SR-IOV Operator. It is enabled by default and can be disabled by setting the enableInjector option to false in the default SriovOperatorConfig CR.
    6 Specify the number of CPUs. The DPDK pod usually requires that exclusive CPUs be allocated from the kubelet. To do this, set the CPU Manager policy to static and create a pod with Guaranteed Quality of Service (QoS).
    7 Specify hugepage size hugepages-1Gi or hugepages-2Mi and the quantity of hugepages that will be allocated to the DPDK pod. Configure 2Mi and 1Gi hugepages separately. Configuring 1Gi hugepages requires adding kernel arguments to Nodes.

  6. Create the DPDK pod by running the following command:

    $ oc create -f mlx-dpdk-pod.yaml

Using the TAP CNI to run a rootless DPDK workload with kernel access

DPDK applications can use virtio-user as an exception path to inject certain types of packets, such as log messages, into the kernel for processing. For more information about this feature, see Virtio_user as Exception Path.

In OpenShift Container Platform version 4.14 and later, you can use non-privileged pods to run DPDK applications alongside the tap CNI plugin. To enable this functionality, you need to mount the vhost-net device by setting the needVhostNet parameter to true within the SriovNetworkNodePolicy object.

DPDK and TAP plugin
Figure 1. DPDK and TAP example configuration
Prerequisites

  • You have installed the OpenShift CLI (oc).

  • You have installed the SR-IOV Network Operator.

  • You are logged in as a user with cluster-admin privileges.

  • Ensure that setsebools container_use_devices=on is set as root on all nodes.

    Use the Machine Config Operator to set this SELinux boolean.
Procedure

  1. Create a file, such as test-namespace.yaml, with content like the following example:

    apiVersion: v1
kind: Namespace
metadata:
  name: test-namespace
  labels:
    pod-security.kubernetes.io/enforce: privileged
    pod-security.kubernetes.io/audit: privileged
    pod-security.kubernetes.io/warn: privileged
    security.openshift.io/scc.podSecurityLabelSync: "false"

  2. Create the new Namespace object by running the following command:

    $ oc apply -f test-namespace.yaml

  3. Create a file, such as sriov-node-network-policy.yaml, with content like the following example::

    apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetworkNodePolicy
metadata:
 name: sriovnic
 namespace: openshift-sriov-network-operator
spec:
 deviceType: netdevice (1)
 isRdma: true (2)
 needVhostNet: true (3)
 nicSelector:
   vendor: "15b3" (4)
   deviceID: "101b" (5)
   rootDevices: ["00:05.0"]
 numVfs: 10
 priority: 99
 resourceName: sriovnic
 nodeSelector:
    feature.node.kubernetes.io/network-sriov.capable: "true"
    1 This indicates that the profile is tailored specifically for Mellanox Network Interface Controllers (NICs).
    2 Setting isRdma to true is only required for a Mellanox NIC.
    3 This mounts the /dev/net/tun and /dev/vhost-net devices into the container so the application can create a tap device and connect the tap device to the DPDK workload.
    4 The vendor hexadecimal code of the SR-IOV network device. The value 15b3 is associated with a Mellanox NIC.
    5 The device hexadecimal code of the SR-IOV network device.

  4. Create the SriovNetworkNodePolicy object by running the following command:

    $ oc create -f sriov-node-network-policy.yaml

  5. Create the following SriovNetwork object, and then save the YAML in the sriov-network-attachment.yaml file:

    apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetwork
metadata:
 name: sriov-network
 namespace: openshift-sriov-network-operator
spec:
 networkNamespace: test-namespace
 resourceName: sriovnic
 spoofChk: "off"
 trust: "on"

    See the "Configuring SR-IOV additional network" section for a detailed explanation on each option in SriovNetwork.

    An optional library, app-netutil, provides several API methods for gathering network information about a container’s parent pod.

  6. Create the SriovNetwork object by running the following command:

    $ oc create -f sriov-network-attachment.yaml

  7. Create a file, such as tap-example.yaml, that defines a network attachment definition, with content like the following example:

    apiVersion: "k8s.cni.cncf.io/v1"
kind: NetworkAttachmentDefinition
metadata:
 name: tap-one
 namespace: test-namespace (1)
spec:
 config: '{
   "cniVersion": "0.4.0",
   "name": "tap",
   "plugins": [
     {
        "type": "tap",
        "multiQueue": true,
        "selinuxcontext": "system_u:system_r:container_t:s0"
     },
     {
       "type":"tuning",
       "capabilities":{
         "mac":true
       }
     }
   ]
 }'
    1 Specify the same target_namespace where the SriovNetwork object is created.

  8. Create the NetworkAttachmentDefinition object by running the following command:

    $ oc apply -f tap-example.yaml

  9. Create a file, such as dpdk-pod-rootless.yaml, with content like the following example:

    apiVersion: v1
kind: Pod
metadata:
  name: dpdk-app
  namespace: test-namespace (1)
  annotations:
    k8s.v1.cni.cncf.io/networks: '[
      {"name": "sriov-network", "namespace": "test-namespace"},
      {"name": "tap-one", "interface": "ext0", "namespace": "test-namespace"}]'
spec:
  nodeSelector:
    kubernetes.io/hostname: "worker-0"
  securityContext:
      fsGroup: 1001 (2)
      runAsGroup: 1001 (3)
      seccompProfile:
        type: RuntimeDefault
  containers:
  - name: testpmd
    image: <DPDK_image> (4)
    securityContext:
      capabilities:
        drop: ["ALL"] (5)
        add: (6)
          - IPC_LOCK
          - NET_RAW #for mlx only (7)
      runAsUser: 1001 (8)
      privileged: false (9)
      allowPrivilegeEscalation: true (10)
      runAsNonRoot: true (11)
    volumeMounts:
    - mountPath: /mnt/huge (12)
      name: hugepages
    resources:
      limits:
        openshift.io/sriovnic: "1" (13)
        memory: "1Gi"
        cpu: "4" (14)
        hugepages-1Gi: "4Gi" (15)
      requests:
        openshift.io/sriovnic: "1"
        memory: "1Gi"
        cpu: "4"
        hugepages-1Gi: "4Gi"
    command: ["sleep", "infinity"]
  runtimeClassName: performance-cnf-performanceprofile (16)
  volumes:
  - name: hugepages
    emptyDir:
      medium: HugePages
    1 Specify the same target_namespace in which the SriovNetwork object is created. If you want to create the pod in a different namespace, change target_namespace in both the Pod spec and the SriovNetwork object.
    2 Sets the group ownership of volume-mounted directories and files created in those volumes.
    3 Specify the primary group ID used for running the container.
    4 Specify the DPDK image that contains your application and the DPDK library used by application.
    5 Removing all capabilities (ALL) from the container’s securityContext means that the container has no special privileges beyond what is necessary for normal operation.
    6 Specify additional capabilities required by the application inside the container for hugepage allocation, system resource allocation, and network interface access. These capabilities must also be set in the binary file by using the setcap command.
    7 Mellanox network interface controller (NIC) requires the NET_RAW capability.
    8 Specify the user ID used for running the container.
    9 This setting indicates that the container or containers within the pod should not be granted privileged access to the host system.
    10 This setting allows a container to escalate its privileges beyond the initial non-root privileges it might have been assigned.
    11 This setting ensures that the container runs with a non-root user. This helps enforce the principle of least privilege, limiting the potential impact of compromising the container and reducing the attack surface.
    12 Mount a hugepage volume to the DPDK pod under /mnt/huge. The hugepage volume is backed by the emptyDir volume type with the medium being Hugepages.
    13 Optional: Specify the number of DPDK devices allocated for the DPDK pod. If not explicitly specified, this resource request and limit is automatically added by the SR-IOV network resource injector. The SR-IOV network resource injector is an admission controller component managed by SR-IOV Operator. It is enabled by default and can be disabled by setting the enableInjector option to false in the default SriovOperatorConfig CR.
    14 Specify the number of CPUs. The DPDK pod usually requires exclusive CPUs to be allocated from the kubelet. This is achieved by setting CPU Manager policy to static and creating a pod with Guaranteed QoS.
    15 Specify hugepage size hugepages-1Gi or hugepages-2Mi and the quantity of hugepages that will be allocated to the DPDK pod. Configure 2Mi and 1Gi hugepages separately. Configuring 1Gi hugepage requires adding kernel arguments to Nodes. For example, adding kernel arguments default_hugepagesz=1GB, hugepagesz=1G and hugepages=16 will result in 16*1Gi hugepages be allocated during system boot.
    16 If your performance profile is not named cnf-performance profile, replace that string with the correct performance profile name.

  10. Create the DPDK pod by running the following command:

    $ oc create -f dpdk-pod-rootless.yaml
Additional resources

Overview of achieving a specific DPDK line rate

To achieve a specific Data Plane Development Kit (DPDK) line rate, deploy a Node Tuning Operator and configure Single Root I/O Virtualization (SR-IOV). You must also tune the DPDK settings for the following resources:

  • Isolated CPUs

  • Hugepages

  • The topology scheduler

In previous versions of OKD, the Performance Addon Operator was used to implement automatic tuning to achieve low latency performance for OKD applications. In OKD 4.11 and later, this functionality is part of the Node Tuning Operator.
DPDK test environment

The following diagram shows the components of a traffic-testing environment:

DPDK test environment

  • Traffic generator: An application that can generate high-volume packet traffic.

  • SR-IOV-supporting NIC: A network interface card compatible with SR-IOV. The card runs a number of virtual functions on a physical interface.

  • Physical Function (PF): A PCI Express (PCIe) function of a network adapter that supports the SR-IOV interface.

  • Virtual Function (VF): A lightweight PCIe function on a network adapter that supports SR-IOV. The VF is associated with the PCIe PF on the network adapter. The VF represents a virtualized instance of the network adapter.

  • Switch: A network switch. Nodes can also be connected back-to-back.

  • testpmd: An example application included with DPDK. The testpmd application can be used to test the DPDK in a packet-forwarding mode. The testpmd application is also an example of how to build a fully-fledged application using the DPDK Software Development Kit (SDK).

  • worker 0 and worker 1: OKD nodes.

Using SR-IOV and the Node Tuning Operator to achieve a DPDK line rate

You can use the Node Tuning Operator to configure isolated CPUs, hugepages, and a topology scheduler. You can then use the Node Tuning Operator with Single Root I/O Virtualization (SR-IOV) to achieve a specific Data Plane Development Kit (DPDK) line rate.

Prerequisites

  • You have installed the OpenShift CLI (oc).

  • You have installed the SR-IOV Network Operator.

  • You have logged in as a user with cluster-admin privileges.

  • You have deployed a standalone Node Tuning Operator.

    In previous versions of OKD, the Performance Addon Operator was used to implement automatic tuning to achieve low latency performance for OpenShift applications. In OKD 4.11 and later, this functionality is part of the Node Tuning Operator.
Procedure

  1. Create a PerformanceProfile object based on the following example:

    apiVersion: performance.openshift.io/v2
kind: PerformanceProfile
metadata:
  name: performance
spec:
  globallyDisableIrqLoadBalancing: true
  cpu:
    isolated: 21-51,73-103 (1)
    reserved: 0-20,52-72 (2)
  hugepages:
    defaultHugepagesSize: 1G (3)
    pages:
      - count: 32
        size: 1G
  net:
    userLevelNetworking: true
  numa:
    topologyPolicy: "single-numa-node"
  nodeSelector:
    node-role.kubernetes.io/worker-cnf: ""
    1 If hyperthreading is enabled on the system, allocate the relevant symbolic links to the isolated and reserved CPU groups. If the system contains multiple non-uniform memory access nodes (NUMAs), allocate CPUs from both NUMAs to both groups. You can also use the Performance Profile Creator for this task. For more information, see Creating a performance profile.
    2 You can also specify a list of devices that will have their queues set to the reserved CPU count. For more information, see Reducing NIC queues using the Node Tuning Operator.
    3 Allocate the number and size of hugepages needed. You can specify the NUMA configuration for the hugepages. By default, the system allocates an even number to every NUMA node on the system. If needed, you can request the use of a realtime kernel for the nodes. See Provisioning a worker with real-time capabilities for more information.

  2. Save the yaml file as mlx-dpdk-perfprofile-policy.yaml.

  3. Apply the performance profile using the following command:

    $ oc create -f mlx-dpdk-perfprofile-policy.yaml

Example SR-IOV Network Operator for virtual functions

You can use the Single Root I/O Virtualization (SR-IOV) Network Operator to allocate and configure Virtual Functions (VFs) from SR-IOV-supporting Physical Function NICs on the nodes.

For more information on deploying the Operator, see Installing the SR-IOV Network Operator. For more information on configuring an SR-IOV network device, see Configuring an SR-IOV network device.

There are some differences between running Data Plane Development Kit (DPDK) workloads on Intel VFs and Mellanox VFs. This section provides object configuration examples for both VF types. The following is an example of an sriovNetworkNodePolicy object used to run DPDK applications on Intel NICs:

apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetworkNodePolicy
metadata:
  name: dpdk-nic-1
  namespace: openshift-sriov-network-operator
spec:
  deviceType: vfio-pci (1)
  needVhostNet: true (2)
  nicSelector:
    pfNames: ["ens3f0"]
  nodeSelector:
    node-role.kubernetes.io/worker-cnf: ""
  numVfs: 10
  priority: 99
  resourceName: dpdk_nic_1
---
apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetworkNodePolicy
metadata:
  name: dpdk-nic-1
  namespace: openshift-sriov-network-operator
spec:
  deviceType: vfio-pci
  needVhostNet: true
  nicSelector:
    pfNames: ["ens3f1"]
  nodeSelector:
  node-role.kubernetes.io/worker-cnf: ""
  numVfs: 10
  priority: 99
  resourceName: dpdk_nic_2
1 For Intel NICs, deviceType must be vfio-pci.
2 If kernel communication with DPDK workloads is required, add needVhostNet: true. This mounts the /dev/net/tun and /dev/vhost-net devices into the container so the application can create a tap device and connect the tap device to the DPDK workload.

The following is an example of an sriovNetworkNodePolicy object for Mellanox NICs:

apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetworkNodePolicy
metadata:
  name: dpdk-nic-1
  namespace: openshift-sriov-network-operator
spec:
  deviceType: netdevice (1)
  isRdma: true (2)
  nicSelector:
    rootDevices:
      - "0000:5e:00.1"
  nodeSelector:
    node-role.kubernetes.io/worker-cnf: ""
  numVfs: 5
  priority: 99
  resourceName: dpdk_nic_1
---
apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetworkNodePolicy
metadata:
  name: dpdk-nic-2
  namespace: openshift-sriov-network-operator
spec:
  deviceType: netdevice
  isRdma: true
  nicSelector:
    rootDevices:
      - "0000:5e:00.0"
  nodeSelector:
    node-role.kubernetes.io/worker-cnf: ""
  numVfs: 5
  priority: 99
  resourceName: dpdk_nic_2
1 For Mellanox devices the deviceType must be netdevice.
2 For Mellanox devices isRdma must be true. Mellanox cards are connected to DPDK applications using Flow Bifurcation. This mechanism splits traffic between Linux user space and kernel space, and can enhance line rate processing capability.

Example SR-IOV network operator

The following is an example definition of an sriovNetwork object. In this case, Intel and Mellanox configurations are identical:

apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetwork
metadata:
  name: dpdk-network-1
  namespace: openshift-sriov-network-operator
spec:
  ipam: '{"type": "host-local","ranges": [[{"subnet": "10.0.1.0/24"}]],"dataDir":
   "/run/my-orchestrator/container-ipam-state-1"}' (1)
  networkNamespace: dpdk-test (2)
  spoofChk: "off"
  trust: "on"
  resourceName: dpdk_nic_1 (3)
---
apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetwork
metadata:
  name: dpdk-network-2
  namespace: openshift-sriov-network-operator
spec:
  ipam: '{"type": "host-local","ranges": [[{"subnet": "10.0.2.0/24"}]],"dataDir":
   "/run/my-orchestrator/container-ipam-state-1"}'
  networkNamespace: dpdk-test
  spoofChk: "off"
  trust: "on"
  resourceName: dpdk_nic_2
1 You can use a different IP Address Management (IPAM) implementation, such as Whereabouts. For more information, see Dynamic IP address assignment configuration with Whereabouts.
2 You must request the networkNamespace where the network attachment definition will be created. You must create the sriovNetwork CR under the openshift-sriov-network-operator namespace.
3 The resourceName value must match that of the resourceName created under the sriovNetworkNodePolicy.

Example DPDK base workload

The following is an example of a Data Plane Development Kit (DPDK) container:

apiVersion: v1
kind: Namespace
metadata:
  name: dpdk-test
---
apiVersion: v1
kind: Pod
metadata:
  annotations:
    k8s.v1.cni.cncf.io/networks: '[ (1)
     {
      "name": "dpdk-network-1",
      "namespace": "dpdk-test"
     },
     {
      "name": "dpdk-network-2",
      "namespace": "dpdk-test"
     }
   ]'
    irq-load-balancing.crio.io: "disable" (2)
    cpu-load-balancing.crio.io: "disable"
    cpu-quota.crio.io: "disable"
  labels:
    app: dpdk
  name: testpmd
  namespace: dpdk-test
spec:
  runtimeClassName: performance-performance (3)
  containers:
    - command:
        - /bin/bash
        - -c
        - sleep INF
      image: registry.redhat.io/openshift4/dpdk-base-rhel8
      imagePullPolicy: Always
      name: dpdk
      resources: (4)
        limits:
          cpu: "16"
          hugepages-1Gi: 8Gi
          memory: 2Gi
        requests:
          cpu: "16"
          hugepages-1Gi: 8Gi
          memory: 2Gi
      securityContext:
        capabilities:
          add:
            - IPC_LOCK
            - SYS_RESOURCE
            - NET_RAW
            - NET_ADMIN
        runAsUser: 0
      volumeMounts:
        - mountPath: /mnt/huge
          name: hugepages
  terminationGracePeriodSeconds: 5
  volumes:
    - emptyDir:
        medium: HugePages
      name: hugepages
1 Request the SR-IOV networks you need. Resources for the devices will be injected automatically.
2 Disable the CPU and IRQ load balancing base. See Disabling interrupt processing for individual pods for more information.
3 Set the runtimeClass to performance-performance. Do not set the runtimeClass to HostNetwork or privileged.
4 Request an equal number of resources for requests and limits to start the pod with Guaranteed Quality of Service (QoS).

Do not start the pod with SLEEP and then exec into the pod to start the testpmd or the DPDK workload. This can add additional interrupts as the exec process is not pinned to any CPU.

Example testpmd script

The following is an example script for running testpmd:

#!/bin/bash
set -ex
export CPU=$(cat /sys/fs/cgroup/cpuset/cpuset.cpus)
echo ${CPU}

dpdk-testpmd -l ${CPU} -a ${PCIDEVICE_OPENSHIFT_IO_DPDK_NIC_1} -a ${PCIDEVICE_OPENSHIFT_IO_DPDK_NIC_2} -n 4 -- -i --nb-cores=15 --rxd=4096 --txd=4096 --rxq=7 --txq=7 --forward-mode=mac --eth-peer=0,50:00:00:00:00:01 --eth-peer=1,50:00:00:00:00:02

This example uses two different sriovNetwork CRs. The environment variable contains the Virtual Function (VF) PCI address that was allocated for the pod. If you use the same network in the pod definition, you must split the pciAddress. It is important to configure the correct MAC addresses of the traffic generator. This example uses custom MAC addresses.

Using a virtual function in RDMA mode with a Mellanox NIC

RDMA over Converged Ethernet (RoCE) is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.

For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.

RDMA over Converged Ethernet (RoCE) is the only supported mode when using RDMA on OKD.

Prerequisites

  • Install the OpenShift CLI (oc).

  • Install the SR-IOV Network Operator.

  • Log in as a user with cluster-admin privileges.

Procedure

  1. Create the following SriovNetworkNodePolicy object, and then save the YAML in the mlx-rdma-node-policy.yaml file.

    apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetworkNodePolicy
metadata:
  name: mlx-rdma-node-policy
  namespace: openshift-sriov-network-operator
spec:
  resourceName: mlxnics
  nodeSelector:
    feature.node.kubernetes.io/network-sriov.capable: "true"
  priority: <priority>
  numVfs: <num>
  nicSelector:
    vendor: "15b3"
    deviceID: "1015" (1)
    pfNames: ["<pf_name>", ...]
    rootDevices: ["<pci_bus_id>", "..."]
  deviceType: netdevice (2)
  isRdma: true (3)
    1 Specify the device hex code of the SR-IOV network device.
    2 Specify the driver type for the virtual functions to netdevice.
    3 Enable RDMA mode.

    See the Configuring SR-IOV network devices section for a detailed explanation on each option in SriovNetworkNodePolicy.

    When applying the configuration specified in a SriovNetworkNodePolicy object, the SR-IOV Operator may drain the nodes, and in some cases, reboot nodes. It may take several minutes for a configuration change to apply. Ensure that there are enough available nodes in your cluster to handle the evicted workload beforehand.

    After the configuration update is applied, all the pods in the openshift-sriov-network-operator namespace will change to a Running status.

  2. Create the SriovNetworkNodePolicy object by running the following command:

    $ oc create -f mlx-rdma-node-policy.yaml

  3. Create the following SriovNetwork object, and then save the YAML in the mlx-rdma-network.yaml file.

    apiVersion: sriovnetwork.openshift.io/v1
kind: SriovNetwork
metadata:
  name: mlx-rdma-network
  namespace: openshift-sriov-network-operator
spec:
  networkNamespace: <target_namespace>
  ipam: |- (1)
# ...
  vlan: <vlan>
  resourceName: mlxnics
    1 Specify a configuration object for the ipam CNI plugin as a YAML block scalar. The plugin manages IP address assignment for the attachment definition.

    See the "Configuring SR-IOV additional network" section for a detailed explanation on each option in SriovNetwork.

    An optional library, app-netutil, provides several API methods for gathering network information about a container’s parent pod.

  4. Create the SriovNetworkNodePolicy object by running the following command:

    $ oc create -f mlx-rdma-network.yaml

  5. Create the following Pod spec, and then save the YAML in the mlx-rdma-pod.yaml file.

    apiVersion: v1
kind: Pod
metadata:
  name: rdma-app
  namespace: <target_namespace> (1)
  annotations:
    k8s.v1.cni.cncf.io/networks: mlx-rdma-network
spec:
  containers:
  - name: testpmd
    image: <RDMA_image> (2)
    securityContext:
      runAsUser: 0
      capabilities:
        add: ["IPC_LOCK","SYS_RESOURCE","NET_RAW"] (3)
    volumeMounts:
    - mountPath: /mnt/huge (4)
      name: hugepage
    resources:
      limits:
        memory: "1Gi"
        cpu: "4" (5)
        hugepages-1Gi: "4Gi" (6)
      requests:
        memory: "1Gi"
        cpu: "4"
        hugepages-1Gi: "4Gi"
    command: ["sleep", "infinity"]
  volumes:
  - name: hugepage
    emptyDir:
      medium: HugePages
    1 Specify the same target_namespace where SriovNetwork object mlx-rdma-network is created. If you would like to create the pod in a different namespace, change target_namespace in both Pod spec and SriovNetwork object.
    2 Specify the RDMA image which includes your application and RDMA library used by application.
    3 Specify additional capabilities required by the application inside the container for hugepage allocation, system resource allocation, and network interface access.
    4 Mount the hugepage volume to RDMA pod under /mnt/huge. The hugepage volume is backed by the emptyDir volume type with the medium being Hugepages.
    5 Specify number of CPUs. The RDMA pod usually requires exclusive CPUs be allocated from the kubelet. This is achieved by setting CPU Manager policy to static and create pod with Guaranteed QoS.
    6 Specify hugepage size hugepages-1Gi or hugepages-2Mi and the quantity of hugepages that will be allocated to the RDMA pod. Configure 2Mi and 1Gi hugepages separately. Configuring 1Gi hugepage requires adding kernel arguments to Nodes.

  6. Create the RDMA pod by running the following command:

    $ oc create -f mlx-rdma-pod.yaml

A test pod template for clusters that use OVS-DPDK on OpenStack

The following testpmd pod demonstrates container creation with huge pages, reserved CPUs, and the SR-IOV port.

An example testpmd pod
apiVersion: v1
kind: Pod
metadata:
  name: testpmd-dpdk
  namespace: mynamespace
  annotations:
    cpu-load-balancing.crio.io: "disable"
    cpu-quota.crio.io: "disable"
# ...
spec:
  containers:
  - name: testpmd
    command: ["sleep", "99999"]
    image: registry.redhat.io/openshift4/dpdk-base-rhel8:v4.9
    securityContext:
      capabilities:
        add: ["IPC_LOCK","SYS_ADMIN"]
      privileged: true
      runAsUser: 0
    resources:
      requests:
        memory: 1000Mi
        hugepages-1Gi: 1Gi
        cpu: '2'
        openshift.io/dpdk1: 1 (1)
      limits:
        hugepages-1Gi: 1Gi
        cpu: '2'
        memory: 1000Mi
        openshift.io/dpdk1: 1
    volumeMounts:
      - mountPath: /mnt/huge
        name: hugepage
        readOnly: False
  runtimeClassName: performance-cnf-performanceprofile (2)
  volumes:
  - name: hugepage
    emptyDir:
      medium: HugePages
1 The name dpdk1 in this example is a user-created SriovNetworkNodePolicy resource. You can substitute this name for that of a resource that you create.
2 If your performance profile is not named cnf-performance profile, replace that string with the correct performance profile name.

A test pod template for clusters that use OVS hardware offloading on OpenStack

The following testpmd pod demonstrates Open vSwitch (OVS) hardware offloading on OpenStack.

An example testpmd pod
apiVersion: v1
kind: Pod
metadata:
  name: testpmd-sriov
  namespace: mynamespace
  annotations:
    k8s.v1.cni.cncf.io/networks: hwoffload1
spec:
  runtimeClassName: performance-cnf-performanceprofile (1)
  containers:
  - name: testpmd
    command: ["sleep", "99999"]
    image: registry.redhat.io/openshift4/dpdk-base-rhel8:v4.9
    securityContext:
      capabilities:
        add: ["IPC_LOCK","SYS_ADMIN"]
      privileged: true
      runAsUser: 0
    resources:
      requests:
        memory: 1000Mi
        hugepages-1Gi: 1Gi
        cpu: '2'
      limits:
        hugepages-1Gi: 1Gi
        cpu: '2'
        memory: 1000Mi
    volumeMounts:
      - mountPath: /mnt/huge
        name: hugepage
        readOnly: False
  volumes:
  - name: hugepage
    emptyDir:
      medium: HugePages
1 If your performance profile is not named cnf-performance profile, replace that string with the correct performance profile name.

Additional resources