(Heroku-ish) Slug Runner

A container image that runs Heroku-like slugs produced by slugbuilder.

What does it do exactly?

It takes a gzipped tarball of a "compiled" application via stdin or from a URL, letting you run a command in that application environment, or start a process defined in the application Procfile.

Using Slug Runner

First, you need Docker. Then you can either pull the image from the public index:

$ docker pull flynn/slugrunner

Or you can build from this source:

$ docker build -t flynn/slugrunner .

When you run the container, it always expects an app slug to be passed via stdin or by giving it a URL using the SLUG_URL environment variable. Lets run a Rake task that our app uses, attaching to stdout:

$ cat myslug.tgz | docker run -i -a stdin -a stdout flynn/slugrunner rake mytask

We can also load slugs using the SLUG_URL environment variable. This is currently the only way to run interactively, for example running Bash:

$ docker run -e SLUG_URL=http://example.com/slug.tgz -i -t flynn/slugrunner /bin/bash

Commands are run in the application environment, in the root directory of the application, with any default environment variables, and scripts sourced from .profile.d of the application.

Lastly there is a start command that will run any of the process types defined in the Procfile of the app, or of the default process types defined by the buildpack that built the app. For example, here we can start the web process:

$ cat myslug.tgz | docker run -i -a stdin -a stdout -a stderr flynn/slugrunner start web

Base Environment

The container image is based on cedarish, an image that emulates the Heroku Cedar stack environment. App slugs should include everything they need to run, but if something is missing it should be added to cedarish.