Skip to content

Latest commit

 

History

History
283 lines (216 loc) · 10.8 KB

aks-installation.md

File metadata and controls

283 lines (216 loc) · 10.8 KB

Deploying Antrea on AKS and AKS Engine

This document describes steps to deploy Antrea to an AKS cluster or an AKS Engine cluster.

Deploy Antrea to an AKS cluster

Antrea can be deployed to an AKS cluster either in networkPolicyOnly mode or in encap mode.

In networkPolicyOnly mode, Antrea enforces NetworkPolicies and implements other services for the AKS cluster, while the Azure CNI takes care of Pod IPAM and traffic routing across Nodes. For more information about networkPolicyOnly mode, refer to this design document.

In encap mode, Antrea is in charge of Pod IPAM and of all the networking functions on the Nodes. Using encap mode provides access to additional Antrea features, such as Multicast, as inter-Node Pod traffic is encapsulated, and is not handled directly by the Azure Virtual Network. Note that the caveats which apply when deploying Antrea in encap mode on EKS do not apply for AKS.

We recommend encap mode, as it will give you access to the most Antrea features.

AKS Prerequisites

Install the Azure Cloud CLI. Refer to Azure CLI installation guide

We recommend using the latest version available (use at least version 2.39.0).

Deploying Antrea in networkPolicyOnly mode

Creating the cluster

You can use any method to create an AKS cluster. The example given here is using the Azure Cloud CLI.

  1. Create an AKS cluster

    export RESOURCE_GROUP_NAME=aks-antrea-cluster
    export CLUSTER_NAME=aks-antrea-cluster
    export LOCATION=westus
    
    az group create --name $RESOURCE_GROUP_NAME --location $LOCATION
    az aks create \
        --resource-group $RESOURCE_GROUP_NAME \
        --name $CLUSTER_NAME \
        --node-count 2 \
        --network-plugin azure

    Note Do not specify network-policy option.

  2. Get AKS cluster credentials

    az aks get-credentials --name $CLUSTER_NAME --resource-group $RESOURCE_GROUP_NAME
  3. Access your cluster

    kubectl get nodes
    NAME                                STATUS   ROLES   AGE     VERSION
    aks-nodepool1-84330359-vmss000000   Ready    agent   6m21s   v1.16.10
    aks-nodepool1-84330359-vmss000001   Ready    agent   6m25s   v1.16.10

Deploying Antrea

  1. Prepare the cluster Nodes

    Deploy antrea-node-init DaemonSet to enable azure cni to operate in transparent mode.

    kubectl apply -f https://raw.githubusercontent.com/antrea-io/antrea/main/build/yamls/antrea-aks-node-init.yml
  2. Deploy Antrea

    To deploy a released version of Antrea, pick a deployment manifest from the list of releases. Note that AKS support was added in release 0.9.0, which means you cannot pick a release older than 0.9.0. For any given release <TAG> (e.g. v0.9.0), you can deploy Antrea as follows:

    kubectl apply -f https://github.com/antrea-io/antrea/releases/download/<TAG>/antrea-aks.yml

    To deploy the latest version of Antrea (built from the main branch), use the checked-in deployment yaml:

    kubectl apply -f https://raw.githubusercontent.com/antrea-io/antrea/main/build/yamls/antrea-aks.yml

    The command will deploy a single replica of Antrea controller to the AKS cluster and deploy Antrea agent to every Node. After a successful deployment you should be able to see these Pods running in your cluster:

    $ kubectl get pods --namespace kube-system  -l app=antrea
    NAME                                 READY   STATUS    RESTARTS   AGE
    antrea-agent-bpj72                   2/2     Running   0          40s
    antrea-agent-j2sjz                   2/2     Running   0          40s
    antrea-controller-6f7468cbff-5sk4t   1/1     Running   0          43s
    antrea-node-init-6twqg               1/1     Running   0          2m
    antrea-node-init-mqsqr               1/1     Running   0          2m
  3. Restart remaining Pods

    Once Antrea is up and running, restart all Pods in all Namespaces (kube-system, etc) so they can be managed by Antrea.

    kubectl delete pods -n kube-system $(kubectl get pods -n kube-system -o custom-columns=NAME:.metadata.name,HOSTNETWORK:.spec.hostNetwork --no-headers=true | grep '<none>' | awk '{ print $1 }')
    pod "coredns-544d979687-96xm9" deleted
    pod "coredns-544d979687-p7dfb" deleted
    pod "coredns-autoscaler-78959b4578-849k8" deleted
    pod "dashboard-metrics-scraper-5f44bbb8b5-5qkkx" deleted
    pod "kube-proxy-6qxdw" deleted
    pod "kube-proxy-h6d89" deleted
    pod "kubernetes-dashboard-785654f667-7twsm" deleted
    pod "metrics-server-85c57978c6-pwzcx" deleted
    pod "tunnelfront-649ff5fb55-5lxg7" deleted

Deploying Antrea in encap mode

AKS now officially supports Bring your own Container Network Interface (BYOCNI). Thanks to this, you can deploy Antrea on AKS in encap mode, and you will not lose access to any functionality. Check the AKS BYOCNI documentation for prerequisites, in particular for AKS version requirements.

Creating the cluster

You can use any method to create an AKS cluster. The example given here is using the Azure Cloud CLI.

  1. Create an AKS cluster

    export RESOURCE_GROUP_NAME=aks-antrea-cluster
    export CLUSTER_NAME=aks-antrea-cluster
    export LOCATION=westus
    
    az group create --name $RESOURCE_GROUP_NAME --location $LOCATION
    az aks create \
        --resource-group $RESOURCE_GROUP_NAME \
        --name $CLUSTER_NAME \
        --node-count 2 \
        --network-plugin none

    Notice --network-plugin none, which tells AKS not to install any CNI plugin.

  2. Get AKS cluster credentials

    az aks get-credentials --name $CLUSTER_NAME --resource-group $RESOURCE_GROUP_NAME
  3. Access your cluster

    kubectl get nodes
    NAME                                STATUS     ROLES   AGE   VERSION
    aks-nodepool1-40948307-vmss000000   NotReady   agent   18m   v1.27.7
    aks-nodepool1-40948307-vmss000001   NotReady   agent   17m   v1.27.7

    The Nodes are supposed to report a NotReady Status, since no CNI plugin is installed yet.

Deploying Antrea

You can use Helm to easily install Antrea (or any other supported installation method). Just make sure that you configure Antrea NodeIPAM:

# you may not need this:
helm repo add antrea https://charts.antrea.io
helm repo update

cat <<EOF >> values-aks.yml
nodeIPAM:
  enable: true
  clusterCIDRs: ["10.10.0.0/16"]
EOF

helm install -n kube-system -f values-aks.yml antrea antrea/antrea

For more information about how to configure Antrea Node IPAM, please refer to Antrea Node IPAM guide.

After a while, make sure that all your Nodes report a Ready Status and that all your Pods are running correctly. Some Pods, and in particular the metrics-server Pods, may restart once after installing Antrea; this is not an issue.

After a successful installation, Pods should look like this:

NAMESPACE     NAME                                  READY   STATUS    RESTARTS   AGE
kube-system   antrea-agent-bpskv                    2/2     Running   0          7m34s
kube-system   antrea-agent-pfqrn                    2/2     Running   0          7m34s
kube-system   antrea-controller-555b8c799d-wk8zz    1/1     Running   0          7m34s
kube-system   cloud-node-manager-2nszz              1/1     Running   0          31m
kube-system   cloud-node-manager-wj68q              1/1     Running   0          31m
kube-system   coredns-789789675-2nwd7               1/1     Running   0          6m48s
kube-system   coredns-789789675-lbkfn               1/1     Running   0          31m
kube-system   coredns-autoscaler-649b947bbd-j5wqc   1/1     Running   0          31m
kube-system   csi-azuredisk-node-4bnnl              3/3     Running   0          31m
kube-system   csi-azuredisk-node-52nwd              3/3     Running   0          31m
kube-system   csi-azurefile-node-2h66l              3/3     Running   0          31m
kube-system   csi-azurefile-node-dhrf2              3/3     Running   0          31m
kube-system   konnectivity-agent-5fc7989878-6nhwl   1/1     Running   0          31m
kube-system   konnectivity-agent-5fc7989878-t2n6h   1/1     Running   0          30m
kube-system   kube-proxy-96c9p                      1/1     Running   0          31m
kube-system   kube-proxy-x8g8s                      1/1     Running   0          31m
kube-system   metrics-server-5955767688-2hjvn       2/2     Running   0          3m45s
kube-system   metrics-server-5955767688-vmcq7       2/2     Running   0          3m45s

Deploy Antrea to an AKS Engine cluster

Antrea is an integrated CNI of AKS Engine, and can be installed in networkPolicyOnly mode or encap mode to an AKS Engine cluster as part of the AKS Engine cluster deployment. To learn basics of AKS Engine cluster deployment, please refer to AKS Engine Quickstart Guide.

Deploying Antrea in networkPolicyOnly mode

To configure Antrea to enforce NetworkPolicies for the AKS Engine cluster, "networkPolicy": "antrea" needs to be set in kubernetesConfig of the AKS Engine cluster definition (Azure CNI will be used as the networkPlugin):

  "apiVersion": "vlabs",
  "properties": {
    "orchestratorProfile": {
      "kubernetesConfig": {
        "networkPolicy": "antrea"
      }
    }
  }

You can use the deployment template examples/networkpolicy/kubernetes-antrea.json to deploy an AKS Engine cluster with Antrea in networkPolicyOnly mode:

$ aks-engine deploy --dns-prefix <dns-prefix> \
    --resource-group <reource-group> \
    --location westus2 \
    --api-model examples/networkpolicy/kubernetes-antrea.json \
    --auto-suffix

Deploying Antrea in encap mode

To deploy Antrea in encap mode for an AKS Engine cluster, both "networkPlugin": "antrea" and "networkPolicy": "antrea" need to be set in kubernetesConfig of the AKS Engine cluster definition:

  "apiVersion": "vlabs",
  "properties": {
    "orchestratorProfile": {
      "kubernetesConfig": {
        "networkPlugin": "antrea",
        "networkPolicy": "antrea"
      }
    }
  }

You can add "networkPlugin": "antrea" to the deployment template examples/networkpolicy/kubernetes-antrea.json, and use the template to deploy an AKS Engine cluster with Antrea in encap mode.