Deis (pronounced DAY-iss) is an open source PaaS that makes it easy to deploy and manage applications on your own servers. Deis builds upon Docker and CoreOS to provide a lightweight PaaS with a Heroku-inspired workflow.

Deis is pre-release software. The current release is v0.13.1. Until there is a stable release, we recommend you check out the latest "master" branch code and refer to the latest documentation. Upgrading from a previous Deis release? See Upgrading Deis for additional information.

Deis is a set of Docker containers that can be deployed anywhere including public cloud, private cloud, bare metal or your workstation. Decide where you'd like to deploy Deis, then follow the deployment-specific documentation for Rackspace, EC2, DigitalOcean, Google Compute Engine or bare-metal provisioning. Want to see a particular platform supported? Please open an issue.

Trying out Deis? Continue following these instructions for a local installation using Vagrant.

• Due to its nature as a distributed system, we strongly recommend using Deis with a minimum of 3 nodes even for local development and testing
• The Deis "control plane" containers will consume approximately 2 GB of RAM across the cluster. Please be sure you have sufficient free memory before proceeding.
• Install Vagrant v1.6.5+ and VirtualBox

Note for Ubuntu users: the VirtualBox package in Ubuntu (as of the last known release for 14.04) has some issues when running in RAM-constrained environments. Please install the latest version of VirtualBox from Oracle's website.

Each time you spin up a new CoreOS cluster, you must provide a new discovery service URL in the CoreOS user-data file. This URL allows hosts to find each other and perform leader election.

Automatically generate a fresh discovery URL with:

$ make discovery-url

or manually edit contrib/coreos/user-data and add a unique discovery URL generated from https://discovery.etcd.io/new.

Start the CoreOS cluster on VirtualBox. From a command prompt, cd to the root of the Deis project code and type:

$ export DEIS_NUM_INSTANCES=3
$ vagrant up

This instructs Vagrant to spin up 3 VMs. To be able to connect to the VMs, you must add your Vagrant-generated SSH key to the ssh-agent (deisctl requires the agent to have this key):

$ ssh-add ~/.vagrant.d/insecure_private_key

Install the deisctl utility used to provision and operate Deis.

$ curl -sSL http://deis.io/deisctl/install.sh | sh

Export DEISCTL_TUNNEL so you can connect to one of the VMs using the deisctl client on your workstation.

$ export DEISCTL_TUNNEL=172.17.8.100

Use deisctl install platform to install all Deis components across the cluster, then deisctl start platform to start them.

$ deisctl install platform
$ deisctl start platform

This can take some time - the builder must download and install the beefy Heroku cedar stack. Grab some more coffee!

Your Deis platform should be accessible at deis.local3.deisapp.com. For clusters on other platforms see our guide to Configuring DNS.

Now that Deis is running there are a few administrative settings we need to provide.

Set the default domain used to anchor your applications. For a Vagrant environment, use local3.deisapp.com as it will resolve to your local routers:

$ deisctl config platform set domain=local3.deisapp.com

If you want to allow deis run for one-off admin commands, you must provide an SSH private key that allows Deis to gather container logs on CoreOS hosts:

$ deisctl config platform set sshPrivateKey=~/.vagrant.d/insecure_private_key

If you're using the latest Deis release, use pip install --upgrade deis to install the latest Deis Client or download pre-compiled binaries.

If you're working off master, precompiled binaries are likely out of date. You should either symlink the python file directly or build a local copy of the client:

$ sudo ln -fs $(pwd)/client/deis.py /usr/local/bin/deis

or

$ cd client && python setup.py install

Use the Deis Client to register a new user.

$ deis register http://deis.local3.deisapp.com
$ deis keys:add

Use deis keys:add to add your SSH public key for git push access -- normally $HOME/.ssh/id_rsa.pub.

Deis supports 3 deployment workflows:

• Heroku Buildpacks via git push -- Learn more about Using Buildpacks
• Dockerfiles via git push -- Learn more about Using Dockerfiles
• Docker Images via deis pull -- Learn more about Using Docker Images

As an example, we will walk through deploying a Ruby application using the Heroku Buildpack workflow.

Clone an example Ruby application:

$ git clone https://github.com/deis/example-ruby-sinatra.git
$ cd example-ruby-sinatra

From within the application directory, create an application on Deis:

$ cd example-ruby-sinatra
$ deis create

Use deis create --cluster=prod to place the app on a different cluster. Don't like our name-generator? Use deis create myappname.

When you created the application, a git remote for Deis was added automatically. Deploy with git push.

$ git push deis master

This will use the Deis builder to package your application as a Docker Image and automatically deploy it to the platform. Each build creates a new release, which can be rolled back.

Configure your application with environment variables. Each config change also creates a new release.

$ deis config:set DATABASE_URL=postgres://

Test the application by running your test suite inside an ephemeral Docker container.

$ deis run make test

Use the return code to integrate with a CI system.

Scale containers horizontally with ease.

$ deis scale web=8

Access to aggregated logs makes it easy to troubleshoot problems with your application.

$ deis logs

Use deis run to execute one-off commands and explore the deployed container.

Integration tests and corresponding documentation can be found under the tests/ folder.

Learn how to hack on Deis with a Docker-based development workflow.

Common issues that users have run into when provisioning Deis are detailed below.

Did you remember to add your SSH key to the ssh-agent? ssh-add -L should list the key you used to provision the servers. If it's not there, ssh-add -K /path/to/your/key.

The most common cause of this issue is that a new discovery URL wasn't generated and updated in contrib/coreos/user-data before the cluster was launched. Each Deis cluster must have a unique discovery URL, else there will be entries for old hosts that etcd will try and fail to connect to. Try destroying and relaunching the cluster with a fresh discovery URL.

This usually means the controller failed to submit jobs to the scheduler. deisctl journal controller will show detailed error information, but the most common cause of this is that the cluster was created with the wrong SSH key for the --auth parameter. The key supplied with the --auth parameter must be the same key that was used to provision the Deis servers. If you suspect this to be the issue, you'll need to clusters:destroy the cluster and recreate it, along with the app.

Use deisctl status <component> to view the status of the component. You can also use deisctl journal <component> to tail logs for a component, or deisctl list to list all components.

The most common cause of services failing to start are sporadic issues with Docker Hub. We are exploring workarounds and are working with the Docker team to improve Docker Hub reliability. In the meantime, try starting the service again with deisctl restart <component>.

Running into something not detailed here? Please open an issue or hop into #deis and we'll help!

Copyright 2014, OpDemand LLC

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.