Your First Palette Cluster¶
This walks you through standing up a working Kubernetes cluster in Palette on the bare-metal node that was registered for you last week. No prior Kubernetes experience assumed.
By the end you will have built a cluster profile — a reusable recipe — and used it to deploy a live single-node cluster.
Roughly 20 minutes of clicking, then 15–25 minutes of waiting while the cluster builds itself.
What you are actually doing¶
Palette separates what a cluster should be from the machine it runs on.
graph LR
A[Your node<br/>edge-0639…be53d] --> C[Cluster]
B[Cluster profile<br/>OS + Kubernetes + Networking] --> C
C --> D[Working Kubernetes]
You will build an infrastructure profile first — the type that holds the OS, Kubernetes and networking layers — then point it at the node. The profile can be reused for the next ten clusters — that is the whole idea.
The three layers you will add¶
A profile is a stack of layers. Each layer is filled by a pack — Palette's word for a pre-packaged, versioned component. You need exactly three:
| Layer | Pack you will pick | What it does |
|---|---|---|
| Operating System | BYOS Edge OS 2.1.0 |
Tells Palette the OS is already on the machine |
| Kubernetes | Palette eXtended Kubernetes – Edge 1.36.2 |
Kubernetes itself |
| Network | Cilium 1.20.0 |
Lets containers talk to each other |
There is no storage layer here on purpose. This is the smallest cluster that actually works.
What you need before you start¶
| Item | Value |
|---|---|
| Palette | console.spectrocloud.com |
| Project | SA-Craig-Smith — pick this from the project dropdown, top-left |
| Your node | Already registered. You will confirm this in step 1. |
If you get stuck
Every step has a What good looks like box. If your screen does not match it, stop there rather than pressing on — the next step will not fix it. The troubleshooting page covers what usually goes wrong.
Steps¶
- Check your node — confirm the machine is registered and free
- Build the cluster profile — the three layers
- Turn the CIDRs into variables — the interesting part
- Deploy the cluster — put the profile on the node
- Confirm it worked