cf-charmers team mailing list archive

[Whit Morriss <whit.morriss@xxxxxxxxxxxxx>]

This email is for the folks who will be working on integrating the package
creation and the ci, as well as for general awareness.

Currently the packages the cf charm and subcharm build off of live in S3:

  http://cf-packages.s3-website-us-east-1.amazonaws.com/

Hitting this url returns a json document containing a list of projects with
available packages along w/ a sublist of [version, architecture] tuple for
specific releases.  This information may be applied to traverse the url
structure of the serve s3 bucket.

The url/s3/file tree for a project follows this pattern:

     /{project}/{release}/jobs.tgz
     /{project}/{release}/{architecture}.json
     /{project}/packages/{package}-{sha1 checksum}.tgz

what the part do:

 - jobs.tgz is the exact copy of the jobs section of a bosh release at the
tag for the release number.
 - {architecture}.json maps each job to the list of architecture specific
packages on which it depends
 - the packages dir holds all the packages for all of a releases projects,
delimited by their sha1 checksums

Whenever someone deploys a cf-charm, the charm grabs the jobs.tgz and the
arch.json to determine what packages it needs and create the necessary
config.

I've collected some example output here to help illustrate the directory
structure and the contents of the arch.json:
https://gist.github.com/whitmo/d6fc13b27a86757a94ad

This gist also includes instructions on how to install the raindance
package which provides some utilities for converting a exported compiled
packages to the described repo structure or mirroring parts of the current
repository of packages.

-w

-- 
---------------
D. Whit Morriss
Developer, Juju Ecosystem
Canonical USA