Single Node Cluster

Running a development cluster on a single node without any high availability is also possible.

Configuration

When setting up a single-node cluster:

1. Set automatically_upgrade_os to false, especially with attached volumes the automatic reboots won't work properly.
2. Use only one control plane node and one agent node:

control_plane_nodepools = [
  {
    name        = "control-plane",
    server_type = "cx22",
    location    = "nbg1",
    count       = 1
  }
]

agent_nodepools = [
  {
    name        = "agent",
    server_type = "cx22",
    location    = "nbg1",
    count       = 1
  }
]

Ingress Controller

In this configuration, we don't deploy an external load-balancer but use the default k3s service load balancer on the host itself and open up port 80 & 443 in the firewall (done automatically).

Limitations

Single-node clusters:

• Do not provide high availability
• Should have automatic OS upgrades disabled
• Are not recommended for production workloads
• May have issues with persistent volumes

Use Cases

Single-node clusters are ideal for:

• Development and testing
• Learning Kubernetes
• Resource-constrained environments
• Edge computing scenarios