Rancher RKE2

What is RKE2?

RKE2

Rancher Kubernetes Engine 2
A CNCF-certified Kubernetes distribution developed by SUSE Rancher.
Designed as the next generation of RKE, focusing on:
- Security first (FIPS 140-2 compliance, SELinux, CIS hardening profiles).
- Simplicity of deployment (single binary installer).
- Production-ready for on-premise, edge, and cloud environments.

CNCF

Cloud Native Computing Foundation, part of the Linux Foundation
To support and advance cloud-native technologies, especially Kubernetes.
Other projects:
- Prometheus: open-source monitoring and alerting system designed for reliability and scalabiliaty.
- Envoy: a high-performance service proxy and communication bus designed for microservices
- Helm: a package manager for Kubernetes, aims to simplify the deployment and management of applications using reusable configuration templates (charts)
- Containerd: lightweight container runtime.

flowchart TD
subgraph Kubernetes["Kubernetes Cluster"]
    direction TB

    subgraph ControlPlane["Control Plane"]
        Helm["Helm\n(Package Manager)"]
        Prometheus["Prometheus\n(Monitoring & Alerting)"]
    end

    subgraph DataPlane["Data Plane"]
        Containerd["Containerd\n(Container Runtime)"]
        Envoy["Envoy\n(Service Proxy / Networking)"]
    end
end

Helm -->|Deploys apps| Containerd
Prometheus -->|Monitors| ControlPlane
Envoy -->|Routes traffic| Containerd

Key Features

Security Enhancements

Built-in support for CIS benchmarks (kube-bench ready).
- CIS: Center for Internet Security
SELinux policies enforced by default.
Containerd (not Docker) as runtime for security and performance.

Operational Simplicity

Single binary distribution (rke2).
Automated etcd management.
Rancher integration for centralized multi-cluster management.

Flexibility

Works in data centers, cloud, and at the edge.
Supports air-gapped environments.

Architecture Overview

Server Nodes: Run the Kubernetes control plane + etcd.
Agent Nodes: Run workloads (similar to workers).
Runtime: Containerd, not Docker.
Networking: Uses CNI plugins (default: Canal, but others supported).
Ingress: NGINX ingress controller by default.

Why RKE2 Matters

A production-grade Kubernetes distro hardened out-of-the-box.
Ideal for regulated industries (finance, healthcare, government).
Designed for multi-cluster management with Rancher.
Future-facing: aligns with CNCF standards and cloud-native practices.

Hands-on: Typical Installation Workflow

Fabric

Run a git pull (run git stash if necessary) on your fabric-examples repository. Navigate to 478-examples and open up the notebook. Run the notebook!

flowchart TB
  subgraph Laptop["Your laptop"]
    Browser["Browser / curl<br/>http://127.0.0.1:5555"]
    HostDocker["Docker Desktop<br/>publishes host:5555 → container:5555"]
  end

  subgraph Jupyter["fabric-jupyter container"]
    NB["Notebook + Ansible + FABlib"]
    Tunnel["SSH local forward<br/>-L 0.0.0.0:5555:127.0.0.1:8080"]
  end

  subgraph FabricEdge["FABRIC access path"]
    Bastion["bastion.fabric-testbed.net"]
  end

  subgraph Slice["FABRIC slice: RancherK8s"]
    subgraph Mgmt["Management network (often IPv6)"]
      N1mgmt["node1 mgmt IP"]
      N2mgmt["node2 mgmt IP"]
    end

    subgraph DP["L2 dataplane 192.168.1.0/24"]
      N1dp["node1<br/>192.168.1.1"]
      N2dp["node2<br/>192.168.1.2"]
    end

    subgraph RKE2["RKE2 cluster"]
      Server["rke2-server<br/>control-plane + etcd"]
      Agent["rke2-agent<br/>worker"]
      API["Kubernetes API :6443"]
      Canal["Canal / Flannel CNI<br/>iface-can-reach 192.168.1.1"]
      Svc["Service nginx-demo<br/>ClusterIP + NodePort 30080"]
      Pods["Pods nginx-demo<br/>replicas on cluster network"]
      PF["kubectl port-forward<br/>127.0.0.1:8080 → svc/nginx-demo:80"]
    end
  end

  Browser -->|"http://127.0.0.1:5555"| HostDocker
  HostDocker --> Tunnel
  NB -->|"provision / ansible"| Bastion
  Tunnel -->|"SSH via bastion"| Bastion
  Bastion --> N1mgmt
  Bastion --> N2mgmt

  N1mgmt --- Server
  N2mgmt --- Agent
  N1dp --- Server
  N2dp --- Agent
  Server --- API
  Server --- Canal
  Agent --- Canal
  Canal --- Pods
  API --> Svc
  Svc --> Pods
  Server --> PF
  Tunnel -->|"after SSH lands on node1<br/>connect to loopback :8080"| PF
  PF -->|"cluster networking"| Svc

  classDef warn fill:#fff3cd,stroke:#856404
  classDef ok fill:#d4edda,stroke:#155724
  classDef path fill:#cce5ff,stroke:#004085
  class Tunnel,PF ok
  class Canal path
  class Svc warn