A proxy to add support for authentication and authorization to Kibana 4
Switch branches/tags
Nothing to show
Clone or download

README.md

Buckler: a proxy for adding authentication and authorization to Kibana

Buckler is an authentication frontend for ELK (Elasticsearch, Logstash, Kibana). It behaves like a proxy, allowing only access to those indexes the logged in user.

Buckler is implemented as a Django app. You will need a Django project to configure and run the application. A sample (buildout based) Django project is available in the Buckler-project repository on Kumina's GitHub.

How does it work?

Buckler provides a view that serves as the main entry point for accessing Kibana. All requests are forwarded to either Kibana or directly to Elasticsearch (in cases where Kibana will not properly handle the request, e.g. when access configuration indexes).

Buckler provides a very simple authentication mechanism. It does not define any models.

Buckler will inject some additional JavaScript code to add a logout button and (optionally) hide certain configuration options.

Lastly, Buckler will attempt to set up an initial index based on the configuration after first login.

Configuration

Buckler expects the following Django application settings:

  • KIBANA_UPSTREAM: The URL where Kibana can be found. E.g. http://localhost:5601 - no trailing slash!
  • ES_UPSTREAM: The URL where Elasticsearch can be found. E.g. http://localhost:9200 - no trailing slash!
  • ES_USERDATA_PREFIX: the Elasticsearch index prefix for storing user data. E.g. .kibana
  • CONFIG: A dictionary keyed by usernames that have access. Each value consists of a dictionary holding hashed password (using crypt.crypt()), the indexes the user has access to (as a tuple), and the index used to store the user's preferences. For example:
CONFIG = {
	'john': {
		'password': '$6$....',
		'indexes': ('logstash-john-\*', 'logstash-company-\*'),
		'userdata_index': 'john',
	},
	'demo': {
		'password': '$6$....',
		'indexes': ('logstash-demo-\*',)
		'userdata_index': 'demo',
	}
}

Buckler can (and will) set up initial indexes based on the 'indexes' property. However, if you use time stamp based indexes (e.g. [logstash-]YYYY.MM.DD), set these explicitly as 'autoindexes' if you want these to be created instead. For example:

{'john': {..., 'autoindexes': ('[logstash-john-]YYYY.MM.DD',) ..}

Currently, only 'daily' interval is supported!

Requirements

At this point, Buckler will only work with Kibana 4.1.x. It will not work with newer versions (e.g. 4.3) since the protocol has changed significantly.

Buckler has been tested with Django 1.4 and 1.8 on Python 2.7.

Running tests

You can test this package using 'tox'. Install it in a virtualenv if you don't already have it:

virtualenv .; bin/pip install tox
bin/tox

Credits

Buckler was initially developed by Ivo van der Wijk (m3r consultancy B.V.) under commission from Kumina b.v.

Kumina b.v. offers managed operations for companies that focus on creating and delivering online based applications, solutions and services. With highly scalable virtual private servers we provide fully customized robust IT environments for business critical applications. Our managed operations include advanced 24/7 monitoring, trend analysis, updates and patches for your entire infrastructure right up to your own applications, fully managed security measures and unlimited support hours. Find out more at kumina.nl.