Getting Started
Antrea is super easy to install. All the Antrea components are containerized and can be installed using the Kubernetes deployment manifest.
Ensuring requirements are satisfied
When using kubeadm
to create the Kubernetes cluster, passing --pod-network-cidr=<CIDR Range for Pods>
to kubeadm init
will enable NodeIpamController
. Clusters created with kubeadm will always have CNI
plugins enabled. Refer to Creating a cluster with kubeadm for more information about setting up a Kubernetes cluster with kubeadm
.
When the cluster is deployed by other means then:
To enable
NodeIpamController
,kube-controller-manager
should be started with the following flags:--cluster-cidr=<CIDR Range for Pods>
--allocate-node-cidrs=true
To enable
CNI
network plugins,kubelet
should be started with the--network-plugin=cni
flag.To enable masquerading of traffic for Service cluster IP via iptables,
kube-proxy
should be started with the--cluster-cidr=<CIDR Range for Pods>
flag.
As for OVS, when using the built-in kernel module, kernel version >= 4.6 is required. On the other hand, when building it from OVS sources, OVS version >= 2.6.0 is required.
Red Hat Enterprise Linux and CentOS 7.x use kernel 3.10, but as changes to OVS kernel modules are regularly backported to these kernel versions, they should work with Antrea, starting with version 7.4.
In case a node does not have a supported OVS module installed, you can install it following the instructions at: Installing Open vSwitch.
Some experimental features disabled by default may have additional requirements, please refer to the Feature Gates documentation to determine whether it applies to you.
Antrea will work out-of-the-box on most popular Operating Systems. Known issues encountered when running Antrea on specific OSes are documented here.
Installation
To deploy a released version of Antrea, pick a deployment manifest from the list of releases. For any given release <TAG>
(e.g. v0.1.0
), you can deploy Antrea as follows:
kubectl apply -f https://github.com/vmware-tanzu/antrea/releases/download/<TAG>/antrea.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/vmware-tanzu/antrea/main/build/yamls/antrea.yml
Antrea supports some experimental features that can be enabled or disabled, please refer to the Feature Gates documentation for more information.
Windows support
If you want to add Windows Nodes to your cluster, please refer to these installation instructions.
ARM support
Starting with v0.13, Antrea supports arm64 and arm/v7 Nodes. Please refer to these installation instructions.
Deploying Antrea on a Cluster with Existing CNI
The instructions above only apply when deploying Antrea in a new cluster. If you need to migrate your existing cluster from another CNI plugin to Antrea, you will need to do the following:
- Delete previous CNI, including all resources (K8s objects, iptables rules, interfaces, …) created by that CNI.
- Deploy Antrea.
- Restart all Pods in the CNI network in order for Antrea to set-up networking for them. This does not apply to Pods which use the Node’s network namespace (i.e. Pods configured with
hostNetwork: true
). You may usekubectl drain
to drain each Node or reboot all your Nodes.
While this is in-progress, networking will be disrupted in your cluster. After deleting the previous CNI, existing Pods may not be reachable anymore.
For example, when migrating from Flannel to Antrea, you will need to do the following:
- Delete Flannel with
kubectl delete -f <path to your Flannel YAML manifest>
. - Delete Flannel bridge and tunnel interface with
ip link delete flannel.1 && ip link delete flannel cni0
on each Node. - Ensure requirements are satisfied.
- Deploy Antrea.
- Drain and uncordon Nodes one-by-one. For each Node, run
kubectl drain --ignore-daemonsets <node name> && kubectl uncordon <node name>
. The--ignore-daemonsets
flag will ignore DaemonSet-managed Pods, including the Antrea Agent Pods. If you have any other DaemonSet-managed Pods (besides the Antrea ones and system ones such as kube-proxy), they will be ignored and will not be drained from the Node. Refer to the Kubernetes documentation for more information. Alternatively, you can also restart all the Pods yourself, or simply reboot your Nodes.
To build the image locally, you can follow the instructions in the Contributor Guide.
Antrea components can also be run manually as processes for development purposes. See Manual Installation for information.
Deploying Antrea in Kind
To deploy Antrea in a Kind cluster, please refer to this guide.
Deploying Antrea in EKS and GKE
Antrea can be deployed in NetworkPolicy only mode to an EKS cluster or a GKE cluster, and enforce NetworkPolicies for the cluster.
- To deploy Antrea in an EKS cluster, please refer to the EKS installation guide.
- To deploy Antrea in a GKE cluster, please refer to the GKE installation guide.
Deploying Antrea with Custom Certificates
By default, Antrea generates the certificates needed for itself to run. To provide your own certificates, please refer to Securing Control Plane.
Antctl: Installation and Usage
To use antctl, the Antrea command-line tool, please refer to this guide.
Features
Antrea Network Policy
Besides Kubernetes NetworkPolicy, Antrea also implements its own Network Policy CRDs, which provide advanced features including: policy priority, tiering, deny action, external entity, and policy statistics. For more information on usage of Antrea Network Policies, refer to the Antrea Network Policy document.
IPsec Encryption
Antrea supports encrypting GRE tunnel traffic with IPsec. To deploy Antrea with IPsec encryption enabled, please refer to this guide.
Network Flow Visibility
Antrea supports exporting network flow information using IPFIX, and provides a reference cookbook on how to visualize the exported network flows using Elastic Stack and Kibana dashboards. For more information, refer to the network flow visibility document.
NoEncap and Hybrid Traffic Modes
Besides the default Encap
mode, in which Pod traffic across Nodes will be encapsulated and sent over tunnels, Antrea also supports NoEncap
and Hybrid
traffic modes. In NoEncap
mode, Antrea does not encapsulate Pod traffic, but relies on the Node network to route the traffic across Nodes. In Hybrid
mode, Antrea encapsulates Pod traffic when the source Node and the destination Node are in different subnets, but does not encapsulate when the source and the destination Nodes are in the same subnet. Refer to this guide to learn how to configure Antrea with NoEncap
or Hybrid
mode.
Octant UI
Antrea ships with an Octant UI plugin which can show runtime information of Antrea components and perform Antrea Traceflow operations. Refer to this guide to learn how to install Octant and the Antrea plugin.
OVS Hardware Offload
Antrea can offload OVS flow processing to the NICs that support OVS kernel hardware offload using TC. The hardware offload can improve OVS performance significantly. For more information on how to configure OVS offload, refer to the OVS hardware offload guide.
Prometheus Metrics
Antrea supports exporting metrics to Prometheus. For more information, refer to the Prometheus integration document.
Traceflow
Traceflow is a very useful network diagnosis feature in Antrea. It can trace and report the forwarding path of a specified packet in the Antrea network. For usage of this feature, refer to the Traceflow user guide.