On this page
Managed Kubernetes clusters
Patrii Cloud runs the Kubernetes control plane for you and provisions worker nodes on demand. Create a cluster, download its kubeconfig, and deploy your workloads with the standard Kubernetes tools you already use.
A managed cluster removes the toil of building and operating Kubernetes yourself. Patrii provisions the control-plane and worker instances, wires up networking and the cloud integrations, and gives you a cluster you drive with kubectl, Helm and your CI pipelines.
The control plane can run as a single node or in a highly-available configuration with several control-plane nodes so one failure does not take the cluster's API offline. Worker nodes run your pods and are grouped so you can scale their count as your workloads grow.
Key concepts
Cluster — A Kubernetes control plane plus its worker nodes, managed as one resource.
Control plane — The nodes that run the Kubernetes API, scheduler and controllers; run several for high availability.
Worker node — An instance that runs your pods; the cluster has one or more of them.
kubeconfig — The credentials-and-endpoint file kubectl uses to talk to your cluster's API.
Node count — How many worker nodes the cluster runs; you can change it later.
Before you start
- Enough compute quota for the control-plane and worker sizes you choose.
kubectlinstalled on your machine (and Helm if you use it).- A project network the cluster's nodes can use.
Step by step
- Open the create wizard
Go to Kubernetes and choose Create cluster.
Screenshot — The Create cluster button on the Kubernetes pageThe Create cluster button on the Kubernetes page - Name it and pick a version
Give the cluster a name and select a Kubernetes version from the supported list.
Screenshot — The cluster name and Kubernetes version selectionThe cluster name and Kubernetes version selection - Choose the control plane
- Select the size for the control-plane nodes.
- Choose a single control-plane node, or several for high availability so the API survives a node failure.
Screenshot — The control-plane sizing stepThe control-plane sizing step - Choose the worker nodes
- Select the worker node size.
- Set how many worker nodes to start with; you can change this later.
Screenshot — The worker size and node-count stepThe worker size and node-count step - Create and wait for ready
Review the summary and create. The cluster provisions its nodes and moves to a ready state; this takes a few minutes.
Screenshot — A cluster transitioning to the ready stateA cluster transitioning to the ready state - Download the kubeconfig and connect
Open the cluster and download its kubeconfig, then point kubectl at it and confirm the nodes are Ready.
Screenshot — The cluster detail page with the Download kubeconfig actionThe cluster detail page with the Download kubeconfig action
Examples
export KUBECONFIG=~/Downloads/kubeconfig-my-cluster kubectl get nodes kubectl get pods -A
kubectl create deployment web --image=nginx kubectl scale deployment web --replicas=3 kubectl expose deployment web --port=80 --type=LoadBalancer
helm repo add bitnami https://charts.bitnami.com/bitnami helm install my-release bitnami/nginx
Limits & specifications
What the console gives you per cluster:
| Capability | What it does |
|---|---|
Create / delete | Provision or tear down a whole cluster. |
Scale workers | Change the worker node count as load changes. |
kubeconfig | Download admin or user-scoped credentials for kubectl and CI. |
Versions | Pick from the supported Kubernetes versions at create time. |
Metrics & nodes | See cluster metrics and the state of each node. |
Access (RBAC) | Grant users and groups scoped access to the cluster. |
Best practices
- Run a highly-available control plane for anything you cannot afford to have offline during a node failure.
- Keep the admin kubeconfig secret; hand out user-scoped, RBAC-limited access for day-to-day work.
- Pin your manifests to a Kubernetes version and test before upgrading.
- Expose services through a load balancer or ingress rather than individual node IPs.
- Right-size workers and adjust the node count from metrics instead of over-provisioning up front.
FAQ
kubectl cannot reach the cluster.
How do I add more capacity?
Can I give a teammate limited access?
How do I run a LoadBalancer service?
Learn more
Everything you can do here in the console is also available through the Patrii Cloud API for automation and scripting.