dc83f36 Sep 8, 2018
3 contributors

Users who have contributed to this file

@bacongobbler @michelleN @bjornmagnusson
73 lines (53 sloc) 3.56 KB

+++ dep = 3 title = "Draft Pack Format" authors = [ "Matt Fisher" ] created = 2017-10-18 +++


Packs consist of a Dockerfile and a Helm Chart that demonstrates best practices for deploying applications of a given language. For example:

$ draft create
--> Draft detected the primary language as Python with 97.267760% certainty.
--> Ready to sail


Draft Packs allows draft create to bootstrap an application from a defined pack in their local filesystem. Without language packs available to bootstrap applications, users will need to create applications from scratch. This eases the development workflow down to draft create for preparing an application, ready to deploy to Kubernetes.

The Pack Structure

A pack is organized inside a directory in $(draft home)/packs. Inside the pack's directory, there will be a template chart/ and a Dockerfile that will be injected into the application when the pack is requested.

Inside this directory, Draft will expect a structure like this:

python/               # the name of the directory is the name of the pack
    Chart.yaml        # A YAML file containing information about the chart
    LICENSE           # OPTIONAL: A plain text file containing the license for the chart         # OPTIONAL: A human-readable README file
    values.yaml       # The default configuration values for this chart
    charts/           # OPTIONAL: A directory containing any charts upon which this chart depends.
    templates/        # OPTIONAL: A directory of templates that, when combined with values,
                      # will generate valid Kubernetes manifest files.
  Dockerfile          # A Dockerfile for building the application

We could then run draft create with this pack like so:

$ draft create --pack=python
--> Draft detected the primary language as Python with 97.267760% certainty.
--> Ready to sail

The easiest way to create and work with a pack is with the following commands:

$ cd $(draft home)/packs
$ mkdir python
$ cd python
$ helm create chart
Creating chart
$ echo "FROM python" > Dockerfile

Plans and ideas to improve the pack creation workflow can be found in issue 287.

You can optionally define a set of tasks to run at different points while using draft to build and deploy your application in a tasks.toml file inside of a draft pack. The tasks.toml file will get copied to .draft-tasks.toml inside of your application's root directory. If no tasks.toml is provided in the pack, draft create will generate an empty .draft-tasks.toml.

Pack Detection

When draft create is executed on an application, Draft performs a deep search on the current directory to determine the language. It displays language percentages based on the files present in the current directory and subdirectories. The percentages are calculated based on the bytes of code for each language as reported by a Naive Bayesian Classifier, which is trained on files provided by github/linguist. Draft then starts iterating through the packs available in $(draft home)/packs. If it finds a pack that matches the language description, it will then use that pack to bootstrap the application.

Draft's smart pack detection can be overridden with the --pack flag. The detection logic will not be run and Draft will bootstrap the app with the specified pack, no questions asked.