If you don't have a CoreOS machine running, check out the guides on running CoreOS on most cloud providers (EC2, Rackspace, GCE), virtualization platforms (Vagrant, VMware, OpenStack, QEMU/KVM) and bare metal servers (PXE, iPXE, ISO, Installer). With any of these guides you will have machines up and running in a few minutes.
It's highly recommended that you set up a cluster of at least 3 machines — it's not as much fun on a single machine. If you don't want to break the bank, Vagrant allows you to run an entire cluster on your laptop. For a cluster to be properly bootstrapped, you have to provide cloud-config via user-data, which is covered in each platform's guide.
CoreOS gives you three essential tools: service discovery, container management and process management. Let's try each of them out.
First, on the client start your user agent by typing:
Then, add your private key to the agent by typing:
Connect to a CoreOS machine via SSH as the user
core. For example, on Amazon, use:
$ ssh -A firstname.lastname@example.org CoreOS (beta)
-A forwards your ssh-agent to the machine, which is needed for the fleet section of this guide.
If you're using Vagrant, you'll need to connect a bit differently:
$ ssh-add ~/.vagrant.d/insecure_private_key Identity added: /Users/core/.vagrant.d/insecure_private_key (/Users/core/.vagrant.d/insecure_private_key) $ vagrant ssh core-01 -- -A CoreOS (beta)
The first building block of CoreOS is service discovery with etcd (docs). Data stored in etcd is distributed across all of your machines running CoreOS. For example, each of your app containers can announce itself to a proxy container, which would automatically know which machines should receive traffic. Building service discovery into your application allows you to add more machines and scale your services seamlessly.
If you used an example cloud-config from a guide linked in the first paragraph, etcd is automatically started on boot.
A good starting point would be something like:
#cloud-config hostname: coreos0 ssh_authorized_keys: - ssh-rsa AAAA... coreos: units: - name: etcd.service command: start - name: fleet.service command: start etcd: # generate a new token for each unique cluster from https://discovery.etcd.io/new?size=3 # specify the initial size of your cluster with ?size=X discovery: https://discovery.etcd.io/<token>
In order to get the discovery token, visit https://discovery.etcd.io/new and you will receive a URL including your token. Paste the whole thing into your cloud-config file.
The API is easy to use. From a CoreOS machine, you can simply use curl to set and retrieve a key from etcd:
Set a key
message with value
curl -L http://127.0.0.1:4001/v2/keys/message -XPUT -d value="Hello world"
Read the value of
curl -L http://127.0.0.1:4001/v2/keys/message
If you followed a guide to set up more than one CoreOS machine, you can SSH into another machine and can retrieve this same value.
The second building block, docker (docs), is where your applications and code run. It is installed on each CoreOS machine. You should make each of your services (web server, caching, database) into a container and connect them together by reading and writing to etcd. You can quickly try out a minimal busybox container in two different ways:
Run a command in the container and then stop it:
docker run busybox /bin/echo hello world
Open a shell prompt inside the container:
docker run -i -t busybox /bin/sh
The third building block of CoreOS is fleet, a distributed init system for your cluster. You should use fleet to manage the life cycle of your docker containers.
Fleet works by receiving systemd unit files and scheduling them onto machines in the cluster based on declared conflicts and other preferences encoded in the unit file. Using the
fleetctl tool, you can query the status of a unit, remotely access its logs and more.
First, let's construct a simple systemd unit that runs a docker container. Save this as
hello.service in the home directory:
[Unit] Description=My Service After=docker.service [Service] TimeoutStartSec=0 ExecStartPre=-/usr/bin/docker kill hello ExecStartPre=-/usr/bin/docker rm hello ExecStartPre=/usr/bin/docker pull busybox ExecStart=/usr/bin/docker run --name hello busybox /bin/sh -c "while true; do echo Hello World; sleep 1; done" ExecStop=/usr/bin/docker stop hello
The Getting Started with systemd guide explains the format of this file in more detail.
Then load and start the unit:
$ fleetctl load hello.service Unit hello.service loaded on 8145ebb7.../172.17.8.105 $ fleetctl start hello.service Unit hello.service launched on 8145ebb7.../172.17.8.105
Your container has been started somewhere on the cluster. To verify the status, run:
$ fleetctl status hello.service ● hello.service - My Service Loaded: loaded (/run/fleet/units/hello.service; linked-runtime) Active: active (running) since Wed 2014-06-04 19:04:13 UTC; 44s ago Main PID: 27503 (bash) CGroup: /system.slice/hello.service ├─27503 /bin/bash -c /usr/bin/docker start -a hello || /usr/bin/docker run --name hello busybox /bin/sh -c "while true; do echo Hello World; sleep 1; done" └─27509 /usr/bin/docker run --name hello busybox /bin/sh -c while true; do echo Hello World; sleep 1; done Jun 04 19:04:57 core-01 bash: Hello World ..snip... Jun 04 19:05:06 core-01 bash: Hello World
To stop the container, run:
fleetctl destroy hello.service
Fleet has many more features that you can explore in the guides below.