merll / docker-fabric

Integration of Docker deployments into Fabric.
MIT License
79 stars 10 forks source link
docker docker-client docker-containers dockerfile fabric python

Docker-Fabric

Build Docker images, and run Docker containers in Fabric.

Project: https://github.com/merll/docker-fabric

Docs: https://docker-fabric.readthedocs.io/en/latest/

Overview

With a few preparations, Docker images can easily be generated and tested on development machines, and transferred on to a production environment. This package supports managing container configurations along with their dependencies within Fabric-based deployments. DockerFiles can also be easily implemented in Fabric tasks.

Local Docker clients can be controlled directly through docker-py. Remote Docker API services make use of Fabric's SSH connection.

API access

This project is based on Docker-Map, and adapts its container configuration methods.

As with Docker-Map, container configurations can be generated as objects, updated from Python dictionaries, or imported from YAML files in order to control remote clients via the API. Docker-Fabric includes the following enhancements:

Docker client

DockerFabricClient adds Fabric-like logging in the context of container instances on top of Fabric hosts, and enables automatic creation of tunnel connections for access to a remote Docker host using Fabric's SSH connection. By using the tool socat, the Docker client can access a remote service without re-configuration.

Client configuration

DockerClientConfiguration adds the capability of running containers to Fabric hosts with specific Docker settings for each, e.g. the version number.

Running container configurations

ContainerFabric is a simple wrapper that combines Docker-Map's DockerFabricClient, DockerClientConfiguration objects, and container maps.

Command-line based access

The following features are provided by running the appropriate commands on a remote Docker command line:

Tasks

All essential container actions (create, start, stop, remove) and some advanced (e.g. update) can be triggered from the command line as Fabric tasks and executed on the remote service, e.g. via SSH.

Additionally the following tasks are included in this package, that can be run by Fabric directly:

Contributions

Thanks to lfasnacht for publishing an implementation for a local tunnel to a Fabric client in the pull request 939 of Fabric.

Further contributions are maintained in CONTRIBUTIONS.md of the project.