Website | Slack | Email | Twitter | Facebook

Docker to Kubernetes in one command

spread is a command line tool that builds and deploys a Docker project to a Kubernetes cluster in one command. The project's goals are to:

• Enable rapid iteration with Kubernetes
• Be the fastest, simplest way to deploy Docker to production
• Work well for a single developer or an entire team (no more broken bash scripts!)

See how we deployed Mattermost (and you can too!):

Spread is under open, active development. New features will be added regularly over the next few months - explore our roadmap to see what will be built next and send us pull requests for any features you’d like to see added.

See our philosophy for more on our mission and values.

Requirements

• Kubernetes cluster with kubectl installed

Installation

OS X

$ brew tap redspread/spread
$ brew install spread

Linux/Windows

Make sure Go and Git are installed.

go get rsprd.com/spread/cmd/spread

Quickstart

This assumes you have kubectl configured to a running Kubernetes cluster, whether local or remote.

1. Install Spread
2. Clone Mattermost, the open source Slack $ git clone http://github.com/redspread/kube-mattermost
3. Deploy Mattermost to Kubernetes: $ spread build . (local cluster) or $ spread deploy . (remote cluster)
4. Copy the IP and put it in your browser to see your self-hosted app!

For a more detailed walkthrough, see the full guide.

Localkube

Spread makes it easy to set up and iterate with localkube, a local Kubernetes cluster streamlined for rapid development.

Requirements:

• Make sure Docker is set up correctly, including starting docker-machine to bring up a VM. [1]

Getting started:

• Run spread cluster start to start localkube
• Sanity check: kubectl cluster-info

Suggested workflow:

• docker build the image that you want to work with [2]
• Create Kubernetes objects that use the image build above
• Run spread build . to deploy to cluster [3]
• Iterate on your application, updating image and objects running spread build . each time you want to deploy changes
• When finished, run spread cluster stop to stop localkube

[1] For now, we recommend everyone use a VM when working with localkube [2] spread will soon integrate building (#59)
[3] Since localkube shares a Docker daemon with your host, there is no need to push images :)

See more for our suggestions when developing code with Localkube.

What's been done so far

• spread deploy [-s] PATH [kubectl context]: Deploys a Docker project to a Kubernetes cluster. It completes the following order of operations:
  • Reads context of directory and builds Kubernetes deployment hierarchy.
  • Updates all Kubernetes objects on a Kubernetes cluster.
  • Returns a public IP address, if type Load Balancer is specified.
• Established an implicit hierarchy of Kubernetes objects
• Multi-container deployment
• Support for Linux and Windows
• localkube: easy-to-setup local Kubernetes cluster for rapid development

What's being worked on now

• Build functionality for spread deploy so it also builds any images indicated to be built and pushes those images to the indicated Docker registry.
• spread deploy -p: Pushes all images to registry, even those not built by spread deploy.
• Inner-app linking
• spread logs: Returns logs for any deployment, automatic trying until logs are accessible.
• spread build: Builds Docker context and pushes to a local Kubernetes cluster.
• spread rewind: Quickly rollback to a previous deployment.

See more of our roadmap here!

Future Goals

• Peer-to-peer syncing between local and remote Kubernetes clusters
• Develop workflow for application and deployment versioning
• Introduce paramaterization for container configuration

FAQ

How are clusters selected? Remote clusters are selected from the current kubectl context. Later, we will add functionality to explicitly state kubectl arguments.

How should I set up my directory? Spread requires a specific project directory structure, as it builds from a hierarchy of entities:

• Dockerfile
• *.ctr - optional container file, there can be any number
• pod.yaml - pod file, there can be only one per directory
• rc.yaml - replication controller file, there can be only one per directory
• /.k2e - holds arbitrary Kubernetes objects, such as services and secrets

What is the *.ctr file? The .ctr file is the container struct usually found in the pod.yaml or rc.yaml. Containers can still be placed in pods or replication controllers, but we're encouraging separate container files because it enables users to eventually reuse containers across an application.

Can I deploy a project with just a Dockerfile and *.ctr? Yes. Spread implicitly infers the rest of the app hierarchy.

Contributing

We'd love to see your contributions - please see the CONTRIBUTING file for guidelines on how to contribute.

Reporting bugs

If you haven't already, it's worth going through Elika Etemad's guide for good bug reporting. In one sentence, good bug reports should be both reproducible and specific.

Contact

Founders: founders@redspread.com
Slack: slackin.redspread.com
Planning: Roadmap
Bugs: Issues

License

Spread is under the Apache 2.0 license. See the LICENSE file for details.