.. This work is licensed under a Creative Commons Attribution 4.0 International .. License. .. http://creativecommons.org/licenses/by/4.0 .. (c) OPNFV, Arm Limited. =============================================== SRIOV CNI with PF Mode Deployment on Arm Server =============================================== Abstract ======== This document gives a brief introduction on how to deploy SRIOV CNI with PF mode for data plane. Introduction ============ .. _sriov_cni: https://github.com/hustcat/sriov-cni .. _Flannel: https://github.com/coreos/flannel .. _Multus: https://github.com/Intel-Corp/multus-cni .. _cni: https://github.com/containernetworking/cni .. _kubeadm: https://kubernetes.io/docs/setup/independent/create-cluster-kubeadm/ .. _k8s-crd: https://kubernetes.io/docs/concepts/api-extension/custom-resources/ .. _arm64: https://github.com/kubernetes/website/pull/6511 .. _files: https://github.com/kubernetes/website/pull/6511/files As we know, in some cases we need to deploy multiple network interfaces with data-plane acceration for single Pod. Typically, in production area(5G, Edge computing), one interface we used for control plane, it usually will be flannel. For data plane, sriov cni + DPDK has the best throughput and the lowest lantency. In this case, I will introduce sriov cni with PF mode firstly. SRIOV with PF mode is always used in Edge computing. Because sriov NIC is not so common on Edge computing. And also PF mode is used as vFirewall, vPorxy in data center. NIC with SR-IOV capabilities works by introducing the idea of physical functions (PFs) and virtual functions (VFs). In general, PF is used by host. Each VFs can be treated as a separate physical NIC and assigned to one container, and configured with separate MAC, VLAN and IP, etc. If we want the best networking performance for Pods, this should be the best solution. For demo purpose, I suggest that we use Kubeadm to deploy a Kubernetes cluster firstly. Then I will give out a typical deployment scenario with SRIOV data plane interface added. Use Case Architecture ===================== Kubelet is responsible for establishing the network interfaces for each pod; it does this by invoking its configured CNI plugin. When Multus is invoked, it recovers pod annotations related to Multus, in turn, then it uses these annotations to recover a Kubernetes custom resource definition (CRD), which is an object that informs which plugins to invoke and the configuration needing to be passed to them. Basic Information about Environment =================================== Cluster Info In this case, we deploy master and slave as one node. Suppose it to be: 192.168.1.2 In 192.168.1.2, 2 NIC as required. Suppose it to be: eth0, eth1, eth0 is used to be controle plane, and eth1 is used to be data plane. Deploy Kubernetes ----------------- Please see link(https://kubernetes.io/docs/setup/independent/create-cluster-kubeadm/) as reference. Rbac Added (optional) --------------------- Please make sure that rbac was added for Kubernetes cluster. here we name it as rbac.yaml: :: apiVersion: rbac.authorization.k8s.io/v1beta1 kind: ClusterRoleBinding metadata: name: fabric8-rbac subjects: - kind: ServiceAccount # Reference to upper's `metadata.name` name: default # Reference to upper's `metadata.namespace` namespace: default roleRef: kind: ClusterRole name: cluster-admin apiGroup: rbac.authorization.k8s.io command: :: kubectl create -f rbac.yaml Creat CRD --------- Please make sure that CRD was added for Kubernetes cluster. Here we name it as crdnetwork.yaml: :: apiVersion: apiextensions.k8s.io/v1beta1 kind: CustomResourceDefinition metadata: # name must match the spec fields below, and be in the form: . name: networks.kubernetes.com spec: # group name to use for REST API: /apis// group: kubernetes.com # version name to use for REST API: /apis// version: v1 # either Namespaced or Cluster scope: Namespaced names: # plural name to be used in the URL: /apis/// plural: networks # singular name to be used as an alias on the CLI and for display singular: network # kind is normally the CamelCased singular type. Your resource manifests use this. kind: Network # shortNames allow shorter string to match your resource on the CLI shortNames: - net command: :: kubectl create -f crdnetwork.yaml Create Flannel-network for Control Plane ---------------------------------------- Create flannel network as control plane. Here we name it as flannel-network.yaml: :: apiVersion: "kubernetes.com/v1" kind: Network metadata: name: flannel-conf plugin: flannel args: '[ { "masterplugin": true, "delegate": { "isDefaultGateway": true } } ]' command: :: kubectl create -f flannel-network.yaml Create Sriov-network for Data Plane ----------------------------------- Create sriov network with PF mode as data plane. Here we name it as sriov-network.yaml: :: apiVersion: "kubernetes.com/v1" kind: Network metadata: name: sriov-conf plugin: sriov args: '[ { "master": "eth1", "pfOnly": true, "ipam": { "type": "host-local", "subnet": "192.168.123.0/24", "rangeStart": "192.168.123.2", "rangeEnd": "192.168.123.10", "routes": [ { "dst": "0.0.0.0/0" } ], "gateway": "192.168.123.1" } } ]' command: :: kubectl create -f sriov-network.yaml CNI Installation ================ .. _CNI: https://github.com/containernetworking/plugins Firstly, we should deploy all CNI plugins. The build process is following: :: git clone https://github.com/containernetworking/plugins.git cd plugins ./build.sh cp bin/* /opt/cni/bin .. _Multus: https://github.com/Intel-Corp/multus-cni To deploy control plane and data plane interfaces, besides the Flannel CNI and SRIOV CNI, we need to deploy the Multus_. The build process of it is as: :: git clone https://github.com/Intel-Corp/multus-cni.git cd multus-cni ./build cp bin/multus /opt/cni/bin To use the Multus_ CNI, we should put the Multus CNI binary to /opt/cni/bin/ where the Flannel CNI and SRIOV CNIs are put. .. _SRIOV: https://github.com/hustcat/sriov-cni The build process of it is as: :: git clone https://github.com/hustcat/sriov-cni.git cd sriov-cni ./build cp bin/* /opt/cni/bin CNI Configuration ================= The following multus CNI configuration is located in /etc/cni/net.d/, here we name it as multus-cni.conf: :: { "name": "minion-cni-network", "type": "multus", "kubeconfig": "/etc/kubernetes/admin.conf", "delegates": [{ "type": "flannel", "masterplugin": true, "delegate": { "isDefaultGateway": true } }] } command: :: step1, remove all files in /etc/cni/net.d/ rm /etc/cni/net.d/* -rf step2, copy /etc/kubernetes/admin.conf into each nodes. step3, copy multus-cni.conf into /etc/cni/net.d/ step4, restart kubelet systemctl restart kubelet Configuring Pod with Control Plane and Data Plane ================================================= 1, Save the below following YAML to pod-sriov.yaml. In this case flannle-conf network object act as the primary network. :: apiVersion: v1 kind: Pod metadata: name: pod-sriov annotations: networks: '[ { "name": "flannel-conf" }, { "name": "sriov-conf" } ]' spec: # specification of the pod's contents containers: - name: pod-sriov image: "busybox" command: ["top"] stdin: true tty: true 2, Create Pod :: command: kubectl create -f pod-sriov.yaml 3, Get the details of the running pod from the master :: # kubectl get pods NAME READY STATUS RESTARTS AGE pod-sriov 1/1 Running 0 30s Verifying Pod Network ===================== :: # kubectl exec pod-sriov -- ip a 1: lo: mtu 65536 qdisc noqueue qlen 1000 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet6 ::1/128 scope host valid_lft forever preferred_lft forever 3: eth0@if124: mtu 1450 qdisc noqueue link/ether 0a:58:0a:e9:40:2a brd ff:ff:ff:ff:ff:ff inet 10.233.64.42/24 scope global eth0 valid_lft forever preferred_lft forever inet6 fe80::8e6:32ff:fed3:7645/64 scope link valid_lft forever preferred_lft forever 4: net0: mtu 1500 qdisc pfifo_fast qlen 1000 link/ether 52:54:00:d4:d2:e5 brd ff:ff:ff:ff:ff:ff inet 192.168.123.2/24 scope global net0 valid_lft forever preferred_lft forever inet6 fe80::5054:ff:fed4:d2e5/64 scope link valid_lft forever preferred_lft forever Contacts ======== Bin Lu: bin.lu@arm.com