RKE2 (Rancher Kubernetes Engine 2) is SUSE Rancher's Kubernetes distribution, designed for security and compliance: it aims to pass the CIS Kubernetes benchmark with little operator intervention. You choose it when a cluster has to prove its compliance, or when your company already runs Rancher. If you are learning Kubernetes, or want a cluster that matches the upstream documentation, kubeadm remains the right starting point. This page helps you decide, then shows the installation from the official documentation in a few commands.
What you will learn
- Recognize what RKE2 already provides and what it requires of you.
- Compare RKE2 with kubeadm, Kubespray, Talos and k0s on concrete criteria.
- Choose RKE2 or another distribution for your needs.
- Spot what changes when you operate an RKE2 cluster.
What does RKE2 do that kubeadm does not?
kubeadm initializes the control plane and leaves the rest to you: you install containerd, the network plugin, DNS and the ingress controller yourself. RKE2 ships all of this in a single package, installed and configured on every node. The architecture documentation describes it this way: one binary per node, which then starts the components that fit the node's role.
At startup, RKE2 extracts the binaries it needs from its runtime image
(containerd, runc, kubelet, kubectl, crictl). The server runs
kube-apiserver, kube-controller-manager, kube-scheduler and etcd as
static pods, that is, pods the kubelet starts by itself from manifests
placed on disk. The add-ons (network, DNS, metrics-server, ingress controller)
arrive as Helm charts deployed by a built-in controller.
Why choose RKE2?
RKE2 answers one specific need, compliance. Its documentation announces default settings that let you pass the CIS Kubernetes benchmark (versions 1.12 or 1.11) with little operator intervention, FIPS 140-2 compliance, and regular scanning of its components with Trivy in its build pipeline.
Three situations point to RKE2. The first is having to demonstrate alignment with the CIS benchmark or with FIPS in front of an audit. The second is using Rancher Manager to drive several clusters: RKE2 is its native distribution, and the guide Rancher (in French) shows how that management works. The third is needing commercial support, which SUSE provides.
Installing it does not make a cluster compliant. RKE2 does not modify the
host operating system: kernel settings, a dedicated etcd user and the cis
profile remain your responsibility, and the
hardening guide lists them.
To audit the result, see
kube-bench and the CIS benchmark.
When should you choose RKE2, and when not?
The choice depends on what you must prove and on what you want to operate. The table compares RKE2 with kubeadm and the other distributions of the module. K3s (in French) serves as a reference point: it is the lightweight distribution of the same family.
| Need | Best fit | Reason |
|---|---|---|
| Learn Kubernetes, pass the CKA | kubeadm | You install every component yourself and see the cluster as the upstream documentation describes it |
| Prove CIS or FIPS compliance | RKE2 | The hardened settings are provided by default or enabled by profile: cis |
| Rancher Manager already drives your clusters | RKE2 | It is the native distribution of that ecosystem |
| Automate a fleet managed by Ansible | Kubespray | The playbooks install and upgrade the cluster on your existing machines |
| Stop managing the operating system | Talos | The system is dedicated to Kubernetes and is driven through an API |
| Very limited machine, edge | k0s or k3s | The minimum resources are lower |
Resource needs matter. For a control-plane node, the official documentation states a minimum of 4 GB of RAM and 2 CPUs for RKE2, 2 GB and 2 cores for k3s and 1 GB and 1 vCPU for k0s (checked on 2026-11-13). RKE2 therefore needs the most resources of the three.
Do not choose RKE2 in three cases. If nobody has to prove compliance, a simpler cluster is enough. If you want to understand every Kubernetes component, kubeadm shows them to you one by one. If the machine is very limited, k0s or k3s need fewer resources.
Which version of RKE2 and Kubernetes today?
The latest release found on the
releases page on 2026-11-13 is
v1.37.2+rke2r1, published on 2026-11-05. Its notes state that it updates
Kubernetes to v1.37.2. The RKE2 version number therefore follows the Kubernetes
one, followed by the +rke2r1 suffix.
The installer's channel service answered the same day: the stable channel
pointed to v1.36.6+rke2r1 and the latest channel to v1.37.2+rke2r1. To get
the same cluster back in six months, pin the version with
INSTALL_RKE2_VERSION, as in the demonstration below.
The packaged Ingress controller has changed
The community ingress-nginx project has been archived since March 2026, and
the RKE2 documentation states that Traefik is the default controller of new
clusters since version 1.36. On an older cluster, plan the switch. To expose
applications, see
Ingress with Traefik
and Gateway API.
What does the installation look like?
This demonstration follows RKE2's
official quick start, with the
version pinned. It is not a graded task: it shows the shape of the
installation. The task of installing a cluster is practiced with
kubeadm.
The official page requires root access (or sudo) and, if NetworkManager is
active, configuring it so that it ignores the interfaces managed by the network
plugin.
On the first server, three commands install RKE2, enable it at boot and start it. The script drops the binary and the systemd units.
curl -sfL https://get.rke2.io | sudo INSTALL_RKE2_VERSION="v1.37.2+rke2r1" sh -sudo systemctl enable rke2-server.servicesudo systemctl start rke2-server.serviceAfter this installation, the documentation states that the kubeconfig is
written to /etc/rancher/rke2/rke2.yaml and that an enrollment token is
created in /var/lib/rancher/rke2/server/node-token.
Getting kubectl access
RKE2 installs kubectl, but in /var/lib/rancher/rke2/bin/, which is not in
the default PATH. Two environment variables are enough on the server.
export PATH=$PATH:/var/lib/rancher/rke2/binexport KUBECONFIG=/etc/rancher/rke2/rke2.yamlkubectl get nodesThe rke2.yaml file grants full administrator rights: treat it as a
secret. It points to 127.0.0.1, so you must replace that address with the
server's address to use it from another workstation.
Joining agents
On every machine that must run workloads, you install RKE2 in agent mode, then give it the server's address and the token.
curl -sfL https://get.rke2.io | sudo INSTALL_RKE2_TYPE="agent" INSTALL_RKE2_VERSION="v1.37.2+rke2r1" sh -sudo systemctl enable rke2-agent.servicesudo mkdir -p /etc/rancher/rke2The agent's /etc/rancher/rke2/config.yaml file holds two lines, then the
service starts with sudo systemctl start rke2-agent.service.
server: https://<server-address>:9345token: <token read on the server>The server address carries port 9345, not 6443. Port 6443 is the one of the
Kubernetes API; 9345 is the one an agent uses to ask to register. With the
wrong port, the agent never shows up in kubectl get nodes. The documentation
also specifies that every machine must have a unique hostname, otherwise you
must set node-name in the same file.
What does RKE2 change for operations?
With kubeadm, you administer separate components. With RKE2, a single systemd
service per node (rke2-server or rke2-agent) drives everything, and logs
are read with journalctl -u rke2-server. There is no docker command on
these nodes: you inspect containers with crictl, shipped in the same
directory as kubectl.
Where to find the files
The node configuration lives in a single file that you write:
/etc/rancher/rke2/config.yaml. The rke2.yaml file, in the same place, is on
the contrary produced by RKE2: it is the administrator kubeconfig. The
cluster data, including the binaries and the etcd database, live in
/var/lib/rancher/rke2/.
The packaged components
The provided components (network, DNS, metrics-server, ingress controller) are
modules deployed from /var/lib/rancher/rke2/server/manifests. The
documentation specifies that RKE2 rewrites these files at every start: do
not edit them, customize them with a HelmChartConfig object. The network is
Canal by default (Flannel and Calico policies), and RKE2 also offers Cilium,
Calico and Flannel.
Not all of these components are mandatory. The disable option of the
config.yaml file (equivalent to --disable) removes a component, for example
disable: rke2-metrics-server, and the documentation specifies that a disabled
component is uninstalled from the cluster. To prefer your own ingress
controller, the setting is ingress-controller: none, and cni: none does the
same for the network plugin.
Choose the network plugin before the first start: the documentation states that switching plugins on a running cluster is not supported.
Upgrades and backups
Upgrades are done by hand (replace the RKE2 version on each node, servers
first) or with Rancher's system-upgrade-controller, which describes upgrades
in Plan objects. If Rancher manages the cluster, its interface drives the
upgrade. For the general approach, see
upgrading a Kubernetes cluster.
With embedded etcd, RKE2 takes a snapshot at midnight and at noon by default
and keeps five per server. The rke2 etcd-snapshot save command takes one on
demand, and rke2 etcd-snapshot ls lists them. Restoring requires the
original token: keep it. See also
backing up and restoring Kubernetes.
High availability
To tolerate the loss of a server, you need an odd number of servers, at least three, which keep the etcd quorum. Agents and additional servers register through a fixed address, load balancer or virtual IP, on port 9345. The principle is the one of the highly available control plane; the RKE2 documentation gives the exact configuration.
Key takeaways
- RKE2 is SUSE Rancher's Kubernetes distribution, oriented toward security and compliance.
- It ships containerd, the network, DNS and an ingress controller, where kubeadm leaves them to you.
- The default settings help pass the CIS benchmark, but the host system and the
cisprofile still have to be configured. - Choose kubeadm to learn, Kubespray for an Ansible fleet, Talos for an immutable system and k0s or k3s for a limited machine.
- On 2026-11-13, the latest release is v1.37.2+rke2r1, which ships Kubernetes 1.37.2; pin the version with
INSTALL_RKE2_VERSION. - An agent joins the server on port 9345 with the token, and a node's configuration lives in
/etc/rancher/rke2/config.yaml. - Traefik is the default ingress controller of new clusters since version 1.36.
- Choose the network plugin before the first start and keep the token for any restore.
Test your knowledge
The questions cover what sets RKE2 apart from the other distributions, the join port and the configuration files.
Knowledge check
Check what you have learned with this interactive quiz
How it works
- The timer starts when you click Start
- Multiple choice, true/false and short answer questions
- You can move between questions
- Detailed results are shown at the end
Starts the quiz and the timer
Review
(0/0)Score by domain
A domain groups several guides and gets several questions, so its score is more reliable than a per-topic score, which often rests on a single question.
Summary
Strengths
No topic passed at 80% on at least two questions yet.
Weak areas
No topic below the pass mark.
Recommended next
Hints before you try again?
Continue the course: k0s: when to choose this lightweight Kubernetes distribution
A new full quiz with random questions
Work again only on the questions you missed