knife-wip

Chef should be the authoritative source of how your infrastructure looks like. Ideally what's in Chef is also what's running on the servers. However sometimes you have to go into a server, stop Chef and debug something by hand. This has all sorts of implications - depending on the length of the work - the server won't get updates, drifts out of sync with app configuration and worst of all nobody might know. This is why knife-wip provides a tag based way to track work in progress in your infrastructure. Knife-wip uses Chef node tags to track who is working on which servers. This way it supports all the great ways of searching for tags and testing for them in Chef recipe and doesn't rely on any external setup.

In order to provide an easy and standardized way to retrieve information and provide everyone with the maximum amount of information the tag for WIP looks like this:

wip:[USERNAME]:[provided description]

This should make it easy to immediately see who has added the WIP tag and what they are working on.

# mark a node as WIP
% knife node wip web01.example.com testing php build
Created WIP "wip:dschauenberg:testing php build" for node web01.example.com.

# show all nodes that are marked as WIP
% knife wip list
1 nodes found with work in progress

web01.example.com: dschauenberg:testing php build

# remove the WIP tag
% knife node unwip web01.example.com testing php build
Deleted WIP description "wip:dschauenberg:testing php build" for node web01.example.com.

knife-wip has a plugin system that makes it possible to perform custom actions when work is started or stopped.

knife-wip reads its configuration from different files in the following order:

• $COOKBOOKPATH/config/knife-wip-config.yml
• config/knife-wip-config.yml
• /etc/knife-wip-config.yml
• ~/.chef/knife-wip-config.yml

And the file should be YAML and look something like this:

plugins:
  irccat:
    server: irccat.example.com
    port: 12345
    channels:
      - "#chef"

The key of the plugin configuration is what knife-wip uses to try and load the corresponding ruby file under lib/knife-wip/plugins. So it needs to exist there first.

A simple plugin just needs to inherit from KnifeWip::Plugin and implement the two methods wip_start and wip_stop. Those methods get passed in the user, tag and node for the command that was performed. When the plugin gets instantiated it also gets its configuration from the config file passed in and it's available as the @config instance variable in there.

module KnifeWip
  module Plugins
    class LolFormatter < KnifeWip::Plugin

      def wip_start(user, tag, node)
      end

      def wip_stop(user, tag, node)
      end


    end
  end
end

gem install knife-wip

Probably. Please file a ticket here