Archon is a Kubernetes controller managing computing resources based on the operator pattern. Sysadmins could use it to automate their daily work eg. bootstrapping, updating and scaling servers. It is primarily designed for Kubernetes clusters but can easily be extended to other distributed clusters due to its declarative nature.
Archon is designed following the principles of Kubernetes and it works as
an extension of Kubernetes using ThirdPartyResource. Users create abstract cluster
definitions with yaml files. Archon will render the definition resources and create
computing resources accordingly in the configured cloud provider. Then users could
update and scale the cluster using
kubectl in the same way as managing containers.
WARNING: Archon is currently in beta status.
See it in action
Archon itself is a general purpose execution engine. It provides following fundamental capabilities:
- Certificates management. Creating CA, signing new certificates with
- Network management. Create VPC from definition resource and manage its lifecyle.
- User management. Create user definitions used as default users for instances.
- Instance group management. Manage a group of instances with similar configurations.
Scale the group by changing the
- Instance management. Create instance with generated
userdatafor cloudinit from definition resource. Manage the instance lifecyle by watching its status.
We have put together some examples to showcase various ways to bootstrap and manage a Kubernetes cluster with Archon with following features:
- HA master to prevent single point of failure
- Rolling update by creating a new instance group and delete the old one
- Support various operating systems using bootkube or kubeadm
- Scale the cluster with one command
- Manage baremetal servers using matchbox and PXE
- Tools exist today are mainly imperative. Declarative tools will help users share their experience by publish their own cluster definitions. Also users could easily make small customization based on existing template using declarative tools.
- Self-hosted Kubernetes sounds like a promising idea and we need cluster management tools which could be easily integrated into current Kubernetes architecture.
- There're many different environment to launch a Kubernetes cluster in, an unified experience and easy to extend codebase will help developers to collaborate by adding more cloud provider and os support.
How it works
Archon use Kubernetes as its base. Just like the way you use Kubernetes,
define resources in your cluster by creating yaml files using our customized
resource types. Then manage its lifecycle using
kubectl. Until now, we
support these resources:
archon-controller which you should launch beforehand in the cluster will
create and manage its status based on your definition.
You might find the idea that bootstrapping a new Kubernetes cluster with an existing
one too complex. But after your cluster got initialized, you can move all the
definitions and controllers into the new cluster and let it manage itself. It will
be very convenient that you can manage both applications running in your cluster
and the cluster itself with just the
kubectl cli tool.
Supported cloud providers:
Supported baremetal provisioners:
Supported operating system:
Supported bootstrapping tool:
At the moment, we only support limited cloud providers and oses. More cloud providers and operating systems support will be added when the core is stable.
You could launch Archon locally or install it into your cluster.
Download the latest release from Github.
wget https://github.com/kubeup/archon/releases/download/v0.3.0/archon-controller-v0.3.0-linux-amd64.gz gunzip archon-controller-v0.3.0-linux-amd64.gz chmod +x archon-controller-v0.3.0-linux-amd64 mv archon-controller-v0.3.0-linux-amd64 /usr/local/bin/archon-controller
On OSX, just change
wget https://github.com/kubeup/archon/releases/download/v0.3.0/archon-controller-v0.3.0-darwin-amd64.gz gunzip archon-controller-v0.3.0-darwin-amd64.gz chmod +x archon-controller-v0.3.0-darwin-amd64 mv archon-controller-v0.3.0-darwin-amd64 /usr/local/bin/archon-controller
Then config AWS credentials and start running the embeded Kubernetes server along with
export AWS_ACCESS_KEY_ID=YOUR_AWS_KEY_ID export AWS_SECRET_ACCESS_KEY=YOUR_AWS_SECRET export AWS_ZONE=YOUR_CLUSTER_ZONE archon-controller --local --cloud-provider aws
You can also use Aliyun:
export ALIYUN_ACCESS_KEY=YOUR_ALIYUN_KEY_ID export ALIYUN_ACCESS_KEY_SECRET=YOUR_ALIYUN_SECRET archon-controller --local --cloud-provider aliyun
The server will listen on
localhost:8080 and server data will be saved to
Create cluster resource with kubectl
kubectl will talk with the server on
localhost:8080, you can create an example cluster with:
kubectl create -f examples/k8s-simple
After a while, you could get the ip address for the server:
kubectl get instance -o yaml
And ssh into the server with the default password
In cluster deployment
Please follow these instructions if you plan to deploy Archon into your cluster:
We believe every cluster is different. So we provides some ways to bootstrap a Kubernetes cluster as examples. You could choose one as the starting point and customize it to match your needs.
- Simple one machine Kubernetes cluster
- One master multiple nodes Kubernetes cluster
- Self-hosted Kubernetes cluster with bootkube
- Kubernetes cluster with Ubuntu and kubeadm
- Kubernetes cluster with CentOS and kubeadm
- Kubernetes cluster with RedHat and kubeadm
- Aliyun Kubernetes cluster
- Three nodes etcd cluster
- Baremetal Kubernetes cluster with PXE and matchbox