nodes directive is the only required section in the
cluster.yml file. It's used by RKE to specify cluster node(s), ssh credentials used to access the node(s) and which roles these nodes will be in the Kubernetes cluster.
Node Configuration Example
The following example shows node configuration in an example
- address: 22.214.171.124
- address: 126.96.36.199
-----BEGIN RSA PRIVATE KEY-----
-----END RSA PRIVATE KEY-----
- address: 188.8.131.52
- address: 184.108.40.206
taints: # Available as of v0.3.0
- key: test-key
- address: example.com
You can specify the list of roles that you want the node to be as part of the Kubernetes cluster. Three roles are supported:
worker. Node roles are not mutually exclusive. It's possible to assign any combination of roles to any node. It's also possible to change a node's role using the upgrade process.
Before v0.1.8, workloads/pods might have run on any nodes with
controlplane roles, but as of v0.1.8, they will only be deployed to any
With this role, the
etcd container will be run on these nodes. Etcd keeps the state of your cluster and is the most important component in your cluster, single source of truth of your cluster. Although you can run etcd on just one node, it typically takes 3, 5 or more nodes to create an HA configuration. Etcd is a distributed reliable key-value store which stores all Kubernetes state. Taint set on nodes with the etcd role is shown below:
|Taint Key||Taint Value||Taint Effect|
With this role, the stateless components that are used to deploy Kubernetes will run on these nodes. These components are used to run the API server, scheduler, and controllers. Taint set on nodes with the controlplane role is shown below:
|Taint Key||Taint Value||Taint Effect|
With this role, any workloads or pods that are deployed will land on these nodes.
Within each node, there are multiple directives that can be used.
address directive will be used to set the hostname or IP address of the node. RKE must be able to connect to this address.
internal_address provides the ability to have nodes with multiple addresses set a specific address to use for inter-host communication on a private network. If the
internal_address is not set, the
address is used for inter-host communication. The
internal_address directive will set the address used for inter-host communication of the Kubernetes components, e.g. kube-apiserver and etcd. To change the interface used for the vxlan traffic of the Canal or Flannel network plug-ins please refer to the Network Plug-ins Documentation.
Overriding the Hostname
hostname_override is used to be able to provide a friendly name for RKE to use when registering the node in Kubernetes. This hostname doesn't need to be a routable address, but it must be a valid Kubernetes resource name. If the
hostname_override isn't set, then the
address directive is used when registering the node in Kubernetes.
When cloud providers are configured, you may need to override the hostname in order to use the cloud provider correctly. There is an exception for the AWS cloud provider, where the
hostname_override field will be explicitly ignored.
In each node, you specify which
port to be used when connecting to this node. The default port is
For each node, you specify the
user to be used when connecting to this node. This user must be a member of the Docker group or allowed to write to the node's Docker socket.
SSH Key Path
For each node, you specify the path, i.e.
ssh_key_path, for the SSH private key to be used when connecting to this node. The default key path for each node is
If you have a private key that can be used across all nodes, you can set the SSH key path at the cluster level. The SSH key path set in each node will always take precedence.
Instead of setting the path to the SSH key, you can alternatively specify the actual key, i.e.
ssh_key, to be used to connect to the node.
SSH Certificate Path
For each node, you can specify the path, i.e.
ssh_cert_path, for the signed SSH certificate to be used when connecting to this node.
Instead of setting the path to the signed SSH certificate, you can alternatively specify the actual certificate, i.e.
ssh_cert, to be used to connect to the node.
If the Docker socket is different than the default, you can set the
docker_socket. The default is
You have the ability to add an arbitrary map of labels for each node. It can be used when using the ingress controller's
Available as of v0.3.0
You have the ability to add taints for each node.