Nevermind those branch names, that’s just the way the deploy resource works and I guess it has to do with the shallow clone it uses.
In Git, branches are no more than “flags”, or “tags”, or “aliases” that point to a specific node (a commit, in this case) of the acyclic graph that’s the commit tree. Each commit has a unique revision ID that’s a hash of all of itself plus its parent(s), forming a graph.
The deploy_revision resource deploys the code in a releases/<commit_ID> directory, relative to the deploy root. Check that ID and see if it matches with the most current commit ID of your “development” branch. If it does, you’re good.
I find this 2010 OSDC talk very explanative and easy to follow if you’re interested in the way Git manages commits, branches, tags, etc: http://2010.osdc.com.au/proposal/196/git-ages-4-and — it certainly opened my eyes when I was beginning to learn it.
–
Cassiano Leal
http://cassianoleal.com
http://twitter.com/cassianoleal
On July 18, 2013 at 22:29:24, Liam Kirsher (liamk@numenet.com) wrote:
Well, actually it probably is working and I’m not doing something right.
I’m using deploy via the application cookbook, and my recipe looks like this…
application ‘bart’ do
path root_path
owner "nobody"
group “nogroup”
bart_data = data_bag_item(‘bti’, ‘bart’)
mongo_data = data_bag_item(‘bti’, ‘mongo’)
repository bart_data[‘repository’]
deploy_key bart_data[‘deploy_key’]
Chef::Log.warn(“BRANCH: #{node[‘bart’][:branch]}”)
revision node[‘bart’][‘branch’]
migrate false
The correct value for [:bart][:branch] is printed in the log (in this case it happens to be “development”)
However, in the installation dir, it doesn’t seem to have had any effect…?
git branch
- deploy
master
I’m expecting to see “development” there.
I’m assuming that the “deploy” branch is added by the deployment cookbook as a branch off of master.
In my case, I would want it to be branched off of development.
Can anyone explain how this works?
Liam
Liam Kirsher
PGP: http://liam.numenet.com/pgp/