Table of Contents
-
Copy config-example.json to config.json and modify with your desired layout. This is the core config file for the web console, agent, and server. Config.json should be copied to the picluster directory on each node.
-
You can run the server and agent on the same node since they are listening on different ports.
-
Copy the example Dockerfile layout depending on architecture (Optional) Arch Linux ARM
cd /opt/picluster
cp -r example/arm/archlinux/* docker/
x86_64 Ubuntu x86_64
cd /opt/picluster
cp -r example/x86_64/ubuntu/* docker/
An example of the Docker folder layout: Based on the config snippet below, I have two container images that will be called "mysql" and "nginx" that will run on host 192.168.0.100.
"layout": [
{"node":"192.168.0.100", "mysql":"-p 3306:3306","nginx":"-p 80:80"}
The Docker folder will need to be setup like this to match the container names:
/root/docker/mysql/Dockerfile /root/docker/nginx/Dockerfile When it is time to build the containers, PiCluster will use the "docker" variable from config.json plus the container name to locate and build the images.
Server Installation The server is the brain of PiCluster and the agents and web console connect to it.
cd server
npm install
node server.js
Agent Installation
The server will send commands to be executed on the agents nodes. The agent should be installed on each host in the cluster.
cd agent
npm install
node agent.js
Web Console Installation
The web console will send commands to the server that will run commands or gather information from the agent nodes.
cd web
npm install
node webconsole.js