Chook

Documentation is a work in progress. Please get in touch for assistance. <3

Version 1.1.5 is probably the last release of v.1x for chook. Version 2 will be a major reworking of the code. While the general principles will remain the same, a lot will be simplified, some will be jettisoned (e.g. the whole TestEvent aspect) and hopefully lots will be optimized to better handle more and faster incoming webhooks. We'll get some test code up to Github asap.

• Introduction
• Installing Chook
• The Server
  • Server Configuration
  • SSL
  • Logging
  • Admin Interface
  • Event Handlers
  • Internal Handlers - Ruby
    • External Handlers - Any Language
    • Logging from handlers
  • Pointing Jamf Pro at your Chook server
• The Framework
  • Events and Subjects
  • Putting It Together
• TODOs

Chook is a Ruby module that implements a framework for working with webhook events sent by a Jamf Pro Server, a management tool for Apple devices.

Chook also provides a simple sinatra-based HTTP server for receiving and processing those Events, and classes for sending simulated TestEvents to any Jamf webhook handling server.

You do not need to be a Ruby developer to use Chook!

The Chook webhook handling server can use "Event Handlers" written in any language. See below for more information.

Although Chook integrates well with ruby-jss, especially for processing webhook events, it's a separate tool. However, ruby-jss is required when using Jamf-based admin page authentication, or using sampling features to generate TestEvents.

For more detail about the Jamf Pro Webhooks API and the JSON data it passes, please see JAMF's developer reference.

Note: When enabling webhooks from your Jamf Pro server to be handled by the framework, you must specify JSON in the "Content Type" section. This framework does not support XML and will only generate Test Events in JSON format.

As with most Ruby gems: gem install chook -n /usr/local/bin

It will automatically install "sinatra" and "thin", and their dependencies.

If you'll be using a Jamf Pro server to authenticate access to Chook's admin web page, or if you'll be generating Chook TestEvents using data sampled from a Jamf Pro server you'll also need to gem install ruby-jss

Chook comes with a simple HTTP(S) server that uses the Chook framework to handle incoming webhook POST requests from a Jamf Pro server via a single URL https://my.chookserver.org/handle_webhook_event.

After Installing chook, just run /usr/local/bin/chook-server and then point your Jamf Pro webhooks at: http://my_hostname/handle_webhook_event

It will then process incoming webhook POST requests using whatever handlers you have installed.

To automate it on a dedicated Mac, just make a LaunchDaemon plist to run that command and keep it running.

The Chook server looks for a config file at /etc/chook.conf. If not found, default values are used. Full descriptions of the config values are provided in the sample config file at: /path/to/your/gem/folder/chook-/data/chook.conf.example

Each config setting is on a single line like: key: value. Blank lines and those starting with # are ignored.

Here's a summary of possible configuration keys:

• port: The server port
  • default = 443 (SSL), or 80 (no SSL)
• concurrency: Should events be processed simultaneously? (otherwise, one at a time)
  • default = true
• handler_dir: The directory holding the event handler files to load.
  • default = /Library/Application Support/Chook
• use_ssl: Should the server use SSL (https)
  • default = false
• ssl_cert_path: If SSL is used, the path to the server certificate
  • no default
• ssl_private_key_path: If SSL is used, the path to the certificate key
  • no default
• log_file: The path to the server log file
  • default = /var/log/chook-server.log
• log_level: The severity level for log entries
  • default = info
• logs_to_keep: How many old log files to keep when rotating
  • default = 10
• log_max_megs: How big can a log file get before it's rotated.
  • default = 10
• webhooks_user: The username for Basic Authentication
  • no default, leave unset for no authentication
• webhooks_user_pw: The file path, or command, to get the password for the webhooks_user.
  • no default
• admin_user: the username for access to the Chook admin web page, or 'use_jamf'
  • no default, leave unset for no authentication
• admin_pw: if the admin user is NOT 'use_jamf', The file path, or command, to get the password for the admin_user.
  • no default
• admin_session_expires: How many seconds is an admin login valid?
  • default: 86400 (24 hours)
• jamf_server: if admin_user is 'use_jamf', the Jamf Pro server to use for admin authentication
  • default: none, but /etc/ruby-jss.conf is honored.
• jamf_port: if admin_user is 'use_jamf', the Jamf Pro server port to use for admin authentication
  • default: none, but /etc/ruby-jss.conf is honored.
• jamf_use_ssl: if admin_user is 'use_jamf', use SSL to talk to the Jamf Pro server? true/false
  • default: none, but /etc/ruby-jss.conf is honored.
• jamf_verify_cert: if admin_user is 'use_jamf', verify the SSL certificate from the Jamf Pro server? true/false
  • default: none, but /etc/ruby-jss.conf is honored.

See the sample config file for details about all of these settings.

It is recommended to use SSL (https) if possible for security, although its beyond the scope of this document to go into a lot of detail about SSL and certificates. That said, here are some pointers:

• The certificate and key files should be in .pem format

• Make sure you use a certificate that can be verified by the JSS.

  • This might involved adding a CA to the JSS's Java Keystore.

• If running on macOS, the 'thin' webserver and it's underlying 'eventmachine' gem may not like the OS's openssl replacement 'libressl'.

  • One solution is to use homebrew to install openssl and then install eventmachine using that openssl, something like this:

    brew install openssl ; brew link openssl --force ; gem install eventmachine -- --with-ssl-dir=/usr/local/

The Chook server logs activity into the file defined in the log_file config setting, /var/log/chook-server.log by default.

It uses a standard ruby Logger instance, which provides 5 severity levels: fatal (lowest), error, warn, info, and debug (highest).

The log_level config setting defines the level when the server starts up, and log messages of that level or lower will be written to the log.

The log can automatically rotate when it reaches a certain size, as specified by the log_max_megs configuration setting, and the logs_to_keep settings tells chook how many it should keep in total - older log files will be deleted automatically.

If you want to manage the log rotation on your own, set logs_to_keep to zero, or leave it unset, and the log will never automatically rotate.

See below for how to write to the Chook log from within a handler

If you point your web browser at your Chook server http(s)://your.chookserver.org/ , you'll see a simple admin interface.

If an admin_user is set in the configuration, you'll need to provide that name and password, or if admin_user is use_jamf you'll need to provide the username and password of any Jamf Pro user on the server indicated in the config.

The first section provides a live-stream of the server log file, and provides a way to change the server's log level on the fly. Note that this change affects the server itself not just the view in your browser. If you'd like to stop the stream temporarily (e.g. to scroll back, or select some text), just pause and unpause with the checkbox.

The second section lets you see which handlers are currently loaded, and if they are internal or external. The (view) button shows the contents of the handler file.

There's also a button to reload the handlers from the handler directory without restarting the server - useful when you add, delete, or modify them.

The final section just shows your current /etc/chook.conf file, or if there is none, the sample config file is shown, since it shows the default values.

The admin page cannot be used to edit or upload handlers or change the config. For security reasons, you must do that on the server machine itself though normal administrative methods.

A handler is a file containing code to run when a webhook event is received. These files are located in a specified directory, /Library/Application Support/Chook/ by default, and are loaded when the server starts, or the (reload) button is clicked on the admin web page.

Handler files must begin with the name of the event they handle, e.g. ComputerAdded, followed by: nothing, a dot, a dash, or an underscore. Handler filenames are case-insensitive.

All of these file names are valid handlers for ComputerAdded events:

• ComputerAdded
• computeradded.sh
• COMPUTERAdded_notify_team
• Computeradded-update-ldap

Each kind of event can have as many handlers as desired, all will be executed when webhook event is recieved. If all four of the above files existed in the handler directory, every ComputerAdded event would run all four of them.

There are two kinds of handlers, distinguished by their file-executability.

These handlers are non-executable files containing Ruby code. The code is loaded at runtime and executed as a thread in the Chook server process when a matching event is received.

Internal handlers must be defined as a ruby code block passed to the Chook.event_handler method. The block must take one parameter, the Chook::Event subclass instance being handled. Here's a simple example of a handler for a ComputerAdded webhook event.

Chook.event_handler do |event|
  cname = event.subject.deviceName
  uname = event.subject.realName
  event.logger.info "Computer '#{cname}' was just added to the JSS for user #{uname}."
end

The code block, between do and end, takes one parameter which will be a Chook::HandledEvents::ComputerAddedEvent object. In this example the object is stored in the variable "event" and used inside the block.

This handler then extracts the "deviceName" and "realName" values from the subject contained in the event, and uses them to log a message in the chook log.

The subject of an event is the thing that the event affected. In the case of ComputerAdded events, the subject is a Computer. In Chook, it's an object of the class Chook::HandledSubjects::ComputerAdded.

NameSpacing

Be careful when writing internal handlers - they all run in the same Ruby process!

Not only do they have to be thread-safe, but be wary of cluttering the default namespace with constants or methods that might overwrite each other.

A good, very ruby-like, practice is to put the guts of your code into a Module or a Class and use that from inside the handler definition. Here's an example using a Class:

require 'slack-em' # ficticious Slack-chat gem, for demonstation purposes

class ComputerAdder

  SLACK_CHANNEL = '#mac-notifications'

  def initialize(event)
    @event = event
    @comp_name = @event.subject.deviceName
    @user_name = @event.subject.realName
  end

  def run
    @event.logger.info "Adder Starting for computer #{@comp_name}"
    notify_admins
    @event.logger.info "Adder Finished for computer #{@comp_name}"
  end

  def notify_admins
    msg = "Computer '#{@comp_name}' was just enrolled for user #{@user_name}."
    SlackEm.send message: msg, channel: SLACK_CHANNEL
    @event.logger.debug "Admins notified about computer #{@comp_name}"
  end

end

Chook.event_handler do |event|
  ComputerAddeder.new(event).run
end

Here, the handler file defines a 'ComputerAdder' class that does all the work. The handler block merely creates an instance of ComputerAdder, passing it the event, and tells the ComputerAdder instance to run. The instance's run method can then perform any steps desired.

In this example, the SLACK_CHANNEL constant is defined inside the ComputerAdder class. Access to it from inside the class is done using just the constant itself. If you need to access that particular value from outside of the class, you can use ComputerAdder::SLACK_CHANNEL.

This way, similar handlers can have their own SLACK_CHANNEL constants and there won't be any interference.

For more details about event and subject classes, see The Framework

NOTE: Internal handlers must not be executable files. Executability is how the framework determines if a handler is internal or external.

External handlers are executable files that are executed when called by an event. They can be written in any language, but they must accept raw JSON on their standard input. It's up to them to parse that JSON and react to it as desired. In this case the Chook server is merely a conduit for passing the Posted JSON from the Jamf Pro server to the executable program.

Here's a simple example using bash and jq to do something similar to the first ruby example above:

#!/bin/bash
JQ="/usr/local/bin/jq"
while read line ; do JSON="$JSON $line" ; done
cname=`echo $JSON | "$JQ" -r '.event.deviceName'`
uname=`echo $JSON | "$JQ" -r '.event.realName'`
echo "Computer '${cname}' was just added to the JSS for user ${uname}."

External handlers must be executable files. Executability is how the framework determines if a handler is internal or external.

See data/sample_handlers/RestAPIOperation-executable for a more detailed bash example that handles RestAPIOperation events.

See the Jamf Developer Site for details about the JSON contents of webhook events.

Internal handlers

To write to the Chook log file from within an internal handler, use the #logger method of the event object inside the handler block, like so:

Chook.event_handler do |event|
  event.logger.debug "This line appears in the log if the level is debug"
  event.logger.info "This line appears in the log if the level is info or debug"
  event.logger.error "This line appears in the log if the level is error, warn, info, or debug"
end

If you want to log a Ruby exception with its backtrace, you can pass the entire exception to the event logger's 'log_exception' method like this:

  begin
    # something horribly wrong happens here
  rescue => execption
    event.logger.log_exception exception
  end

Log entries written through event objects are preceded with Event xxxxxxxxx: where xxxxxxxxx is an internal ID number for the specific even that wrote the entry.

External handlers

External Handlers can use a URL to make log entries by POSTing to https://my.chookserver/log

The request body must be a JSON object wth 2 keys 'level' and 'message' where both values are strings.

The 'level' must be one of the known log levels: fatal, error, warn, info, or debug. The message is a single line of text to be added to the log.

If your chook server is using Basic Authentication for webhook events, it must be provided for logging also.

Here's an example with curl, split to multi-line for clarity:

curl -u 'auth-user:auth-pw' \
  -H 'Content-Type: application/json' \
  -X POST \
  --data '{"level":"debug", "message":"It Worked"}' \
  https://chookserver.myorg.org/log

Messages logged via this url show up in the log preceded by ExternalEntry:

Any info needed to indentify a log entry with a specific event must be included in your log message.

Once your server is up and running, and you have a handler or two in place, you can create webhooks in your Jamf Pro interface:

1. Navigate to Settings => Global Management => Webhooks
2. Click "New"
3. Give your webhook a display name
4. Enter the URL for your Chook server, ending with 'handle_webhook_event', e.g:

• http://my.chookserver.edu/handle_webhook_event
• https://my.chookserver.edu:8443/handle_webhook_event

5. If you use Basic Authentication, enter the name and password
6. The default timeouts should be OK, but raise them a bit if you're experiencing errors.
7. Set the content type to JSON
8. Select the event that triggers the webook
9. Click "Enabled" at the top.
10. Click "Save"

Watch the Chook log, with the level at info or debug, to see events come in.

While most folks will get along fine using the chook server and writing handlers, the server is built upon a framework implemented in the Chook ruby module, available after doingrequire 'chook'. For those with very specific needs, this framework can be used to implement your own webhook handling service, or to simulate a JamfPro server sending webhook events to some handling service.

The Chook framework abstracts webhook Events and their components as Ruby classes, grouped in two namespaces: HandledEvents, and TestEvents.

When the JSON payload of a JSS webhook POST request is passed into the Chook::Event.parse_event method, an instance of the appropriate subclass of Chook::Event is returned, for example, given the JSON for a ComputerInventoryCompleted webhook event, a Chook::Event::ComputerInventoryCompletedEvent instance is returned by Chook::Event.parse_event.

Each such event instance contains these important attributes:

• webhook_id: The webhook ID stored in the JSS which caused the POST request. This attribute matches the "webhook[:id]" value of the POSTed JSON.

• webhook_name: A read-only instance of the webhook name stored in the JSS which caused the POST request. This attribute matches the "webhook[:name]" value of the POSTed JSON.

• subject: A read-only instance of a Chook::Subject::<Class> representing the "subject" that accompanies the event that triggered the webhook. It comes from the "event" object of the POSTed JSON, and different events come with different subjects attached. For example, the ComputerInventoryCompleted event comes with a "computer" subject containing data about the JSS computer that completed inventory.

  This is not a ruby-jss JSS::Computer object from the REST API, but rather a group of named attributes about that computer.

• event_json: The JSON content from the POST request, parsed into a Ruby hash with symbolized keys (meaning the JSON key "deviceName" becomes the symbol :deviceName).

• raw_json: A String containing the raw JSON from the POST request.

Here are the Event classes supported by the framework and the Subject classes they contain.

For details about the attributes of each Subject, see the Jamf Developer documentation.

A special note about Subjects

In Jamf's documentation, what Chook refers to as a 'Subject' is called an 'event object' because it is a JSON 'object' (a.k.a. dictionary, hash, associative array) labeled 'event'. We've chosen the word 'subject' to make talking about this thing a bit more clear in the context of object-oriented programming.

Each Event class is a subclass of Chook::Event, where all of their functionality is defined.

The Subject classes aren't subclasses, but are dynamically-defined members of the Chook::Subjects module.

Here is a commented sample of ruby code that uses the framework to process a ComputerAdded Event:

# load the framework
require 'chook'

# The framework comes with sample JSON files for each Event type.
# In reality, a webserver would extract this from the data POSTed from the JSS
posted_json = Chook.sample_jsons[:ComputerAdded]

# Create Chook::HandledEvents::ComputerAddedEvent instance for the event
event = Chook::HandledEvent.parse_event posted_json

# Call the events #handle method, which will execute any ComputerAdded
# handlers that were in the Handler directory when the framework was loaded.
event.handle

Of course, you can use the framework without using the built-in #handle method, and if you don't have any handlers in the directory, it won't do anything anyway. Instead you are welcome to use the objects as desired in your own Ruby code.

• Better YARD docs
• more documentation beyond this README