Docs / Kubernetes
Kubernetes clusters
6 minute read
A managed Kubernetes cluster in Lagos: we run the control plane for you, and your worker nodes sit on your project's private network next to your databases and servers.
1. Create a cluster
Open Kubernetes → Create cluster. Pick the Kubernetes version (the newest is selected), the project whose private network the nodes join, and one or more node pools: a name, a node size and how many nodes. Then name the cluster and confirm. The summary shows the monthly estimate before you create anything.
The control plane starts first and the nodes join it; the cluster shows Running after about 5 minutes.
simplicloud kubernetes clusters create prod --pool pool-1:s2:3… creating prod (the control plane, then 3 nodes; about 5 minutes)✓ prod running · Kubernetes 1.37.2 · 3/3 ready
Run simplicloud kubernetes versions and simplicloud kubernetes sizes to see what you can pick.
2. Connect with kubectl
Each cluster has an admin kubeconfig: the cluster's address, its certificate authority and an admin token. Save it with the CLI (it is merged into ~/.kube/config and becomes the current context), or download it from the cluster's Kubeconfig tab.
simplicloud kubernetes kubeconfig save prod✓ Kubeconfig for prod merged to ~/.kube/config (context sc-los1-prod)kubectl get nodeskubectl create deployment hello --image=nginx --replicas=3
Helm, k9s, Lens and CI tools all work with the same file. Keep it like a password. If it leaks, use Regenerate token (or --regenerate): every kubeconfig with the old token stops working at once. Downloads and regenerations are in the audit log, and only developers, admins and owners can get a kubeconfig.
3. Node pools
A node pool is a group of identical worker nodes. Use one pool, or several when workloads need different sizes. From the cluster's Node pools tab you can:
- Add a pool: new nodes join in a few minutes.
- Resize a pool: more nodes are added; with fewer, the newest nodes are drained (their pods move elsewhere) and deleted.
- Delete a pool: its nodes are drained and deleted. A cluster keeps at least one pool.
- Recycle a node: the node is drained, deleted and replaced by a new one of the same size. Use it for a node that stays Not ready.
4. Storage and networking
The default storage class, local-path, keeps each volume on the disk of the node its pod runs on. It is fast and needs nothing set up, but the data stays on that node: if the node is recycled or deleted, the volume goes with it. Keep important data in a managed database or in Spaces.
Nodes have addresses on your project's private network, so pods reach your databases, servers and apps there directly, over the network you already use. Expose a service to the internet from a server or app in the same project for now (see the limits below).
5. What it costs
Worker nodes cost exactly the same as servers of the same size, billed by the hour and capped at the monthly price. The control plane is run by us for ₦7,300 a month (₦10 an hour, plus VAT) per cluster. Deleting a cluster stops billing for all its nodes in the same hour.
6. Current limits
| Today | |
|---|---|
| Region | Lagos (LOS1) |
| Control plane | One per cluster, managed by us (not highly available yet) |
| Node pools | Up to 5 per cluster, up to 10 nodes per pool and 20 per cluster, fixed size (no autoscaling yet) |
| Upgrades | Not in place yet: create a cluster on the newer version and move your workloads |
| Services of type LoadBalancer | Not offered yet (the service stays pending) |
| Storage | local-path on each node's disk; SimpliCloud volumes can't be attached to pods yet |
| Node access | No SSH to nodes; use kubectl exec and kubectl debug |