This is a (very simple) heads-up-display for your nodes. You can see an example in action here.
It solves a few problems:
- Helps you keep tabs on what nodes are failing and why they're failing.
- Keeps you in the loop on how many nodes you have running, how many haven't reported in at a variety of increments.
- Reports resource groupings -- this can assist with identifying non-idempotent recipes and providers, e.g., a large batch of nodes is executing the same thing every time.
You can do a couple of things here:
- here is a cookbook that will do most of the work for you. Consider it beta at the time of this writing. :)
- Alternatively you can do is fork the project and do what's below:
Creating Capistrano configuration to deploy this application should be cake and
should follow a normal deployment pattern. This app should be cooperative with
any rack-capable server with bundle exec support, internal or external,
although it already has support to run under unicorn. Since configuration of
capistrano is out of scope for this project and tends to vary by environment, I
will leave it as an exercise for the reader.
You may need to run the bin/create_database script manually, which should
dump a dashboard.db in your project's root directory. Future support for
other databases is planned, but sqlite3 has served my purpose with excellent
performance on initial testing with around 10k reports and 100 nodes, which
should be more than enough for small to midsize environments.
You will want to extract the chef dashboard handler
to make full effectiveness of the HUD (or write your own). This installs as
both an exception and reporting handler in your /etc/chef/client.rb. You can
use schisamo's wonderful chef handler provider
to drive this installation.
Sending data to the /report handler is pretty simple. The payload is just
JSON data and should be relatively easy to populate from anything that uses
Chef as a framework.
If you look in int/test.json you will see what a payload looks like:
- node
fqdn,name, andipaddressas string. successboolean value.resourcesarray of string.
These are sent in a PUT query as application/json. The appropriate response
is a 200 OK and empty json object (e.g., {}).
More plans will be expressed in a TODO file in the future. I want this project to improve on its basic features, and gradually become a complete reporting solution for Chef-enabled networks. No plans to make it cloudkick or sensu. If you need those, use them. KISS.
- Fork the project
- Make your edits
- Do not modify any LICENSE or CREDITS files, or similar data. If you have concerns about being credited for your work, say so in the pull request.
- Send a pull request.
Erik Hollensbe erik+chef@hollensbe.org