/Hale

Botnet command & control monitor

Primary LanguagePython

About

Hale is a botnet command & control monitor/spy with a modular design to easily develop new modules that monitor new protocols used by C&C servers. Hale comes with IRC and HTTP monitors developed with Twisted to handle scalability of a large amount of connections. Theses modules have configurable protocol grammar and bot settings but can also be modified to fit your needs. All captured logs and files are saved to a database and in case of IRC, tracked IP numbers too.

To hide the location of the operator, connections can be made through SOCKSv5 proxies and this is configurable via the web interface where also all the logs are available to browse together with statistical charts and timelines. The interface was developed with Django and Google Visualization API. Some extras in the web ui are support for a RESTful API with OAuth support and a search engine. Screenshots of the interface are available here.

The main idea with Hale is to help botnet hunting and research to collaborate by creating a network of sensors (Hale monitors). To improve this idea, a XMPP bot is available to connect to a centralized XMPP server where currently two different grouprooms are used for coordinating between sensors and a room for sharing logs and files. The coordination room makes use of botnet hashes that are made out of the unique keys in the botnet settings, in this way, two botnets dont have to be monitored simultaneously that have the same hash (identity) and improves utilization. To help 3rd parties to make use of this network, a bot can join the coordination room and ask a sensor to start tracking a botnet if its untracked by sending the configurations for it. Additionally, in the share room, 3rd party bots can get their hands on logs and files captured by the sensors in realtime. To assist with log history the web API can be used that support GET requests.

Install

Hale has the following dependencies:

Python == 2.6
Django == 1.2.1
Twisted == 10.1.0
GeoIP-Python == 1.2.4 (and GeoIP C lib)
Whoosh == 0.3.18
django-haystack == 1.0.1-final
django-piston == 0.2.3rc1
pefile == 1.2.10-63
sleekxmpp == 0.9Rrc1
wsgiref == 0.1.2
zope.interface == 3.6.1
oauth2 == 1.2.0
httplib2 == 0.6.0

Additionally the monitor requires a database backend driver corresponding to the database used by django. When these libraries are installed download the source from here and extract it anywhere.

Setup

  1. First create a database that will be used by Hale, the database engine can be any of your choice. If you are using an existing database then skip this step.

  2. Next step is to install python database backend drivers corresponding to the one used by the server engine.

  3. Edit settings.py in hale/src/webdb/ and edit the following configurations: ENGINE, NAME, USER, PASSWORD, HOST and PORT where the engine setting is for example django.db.backends.mysql if your server engine is MySQL. The name setting is the name of your database used when creating it.

  4. If you dont want to start your own web ui then skip this step and go to 8). In settings.py, change PATH_TO_APP to point to the fullpath of the Django application, for example: '/home/..../Hale/src/webdb'. In the webdb directory run the following command: python manage.py syncdb. If you get any errors here its most likely that the database settings in settings.py are incorrect. Also, during the sync set the superuser that will be used when administrating the users.

  5. To support searching in the web interface, run python manage.py rebuild_index to let the search engine index first time. This will create a directory named whoosh.index/, to let the web server write here, issue a chmod o+w whoosh.index/. After this you run python manage.py update_index instead and should put this as a cron job to update indexes in a regular interval of your choice.

  6. Run python manage.py runserver and head to http://127.0.0.1:8000 to check if setup was correctly done.

  7. The runserver command deploys a development server that is not recommended for production use due to performance issues. Instead deploy the web ui by installing the mod_wsgi for Apache, assuming you're running Ubuntu, run apt-get install libapache2-mod-wsgi. In /etc/apache2/sites-available/default add the following:

    Alias /media /usr/local/lib/python2.6/dist-packages/django/contrib/admin/media/ <Location /media> Order allow,deny Allow from all

    WSGIDaemonProcess username processes=2 maximum-requests=500 threads=10 WSGIProcessGroup group

    WSGIScriptAlias / /home/..../webdb/django.wsgi

and edit the username and group for which the WSGI daemon should run as. WSGIScriptAlias should be set to the fullpath to the django.wsgi file located in the webdb directory. The Alias specified above is needed for the administration page to be rendered correctly. To be able to upload modules through the web interface, issue a chmod o+w webdb/modules.

  1. Upload modules that will be used from hale/src/modules/ or write your own (see Development section). Upload the desired module in the admin interface and edit for example the module name to irc and the filename to ircModule.py. Specifyc config rules for the module in the corresponding section located in hale/conf/modules.conf and put it in the textbox, also add the uniqueKeys sections for the module being uploaded.

  2. Before running the monitor edit hale.conf in hale/src/conf/ if you wish to use a XMPP server. To activate XMPP bot set use setting to True and either edit login info to an existing account and server or start your own XMPP server. An important step when starting up a XMPP server is to increase the max stanza size from the default value to something like 10Mb. Otherwise malware advertisement will not be possible. The channel settings in hale.conf are used for the share grouproom used by the bot and the coord setting is used for the grouproom where all coordination between sensors is taking place.

  3. Edit hale.conf and set client and server settings.

Usage

To start the monitor head to hale/src/ and execute python server.py. If it fires up with errors then the django settings.py file is not correctly set or some libraries are missing. With the client it's possible to issue commands to the server but it requires the user to authenticate himself against the server, access is granted for users that are set as staff members through the web interface. When the client is running type help or ? to get the available commands. Type help command to get more info about the specific command. Starting up a monitor bot is done by first editing the hale/src/conf/modules.conf file, for example using a irc configuration as follow:

[uniqueKeys]
irc = botnet, *grammar

[ircConf] 
module = irc 
botnet = irc.freenode.net 
port = 6667 
password = None 
nick = nickname
username = agent007 
realname = Spying 
channel = #channelname 
channel_pass = somepass 
pass_grammar = PASS 
nick_grammar = NICK 
user_grammar = USER 
join_grammar = JOIN 
version_grammar = VERSION 
time_grammar = TIME 
privmsg_grammar = PRIVMSG 
topic_grammar = TOPIC 
currenttopic_grammar = 332 
ping_grammar = PING 
pong_grammar = PONG

Edit or create a new config by specifying a new uniquely named section ([ircConf] part). At the top of the config file there is a section called uniqueKeys where all unique fields for a module are specified and used to generate the botnet hash, this should usually not be changed to preserve correct botnet tracking. When this is done run useconf section to load the configuration and then fire up the bot with exec modulename id where id is set by you to identify the botnet.

The web interface provides access to all captured data in the database which is accessible from the index page. There is also a search function which enables the user to search for botnet and file hashes, related IP numbers, botnet IDs, botnet modules used and botnet hosts. If the user got access to edit proxies or modules then this can be done in the admin section, url to this is http://.../admin. The administrator can set user modes and also add consumers for the web API.

Development

HOWTO add modules

  1. Implement module, for example:

    import moduleManager from utils import moduleInterface

    @moduleManager.register("irc") def module_setup(config, hash): """ Function to register modules, simply implement this to pass along the config and hash to the module object and return the it back. """

     return IRC(config, hash)
    

    must inherit from Module class

    class IRC(moduleInterface.Module):

    def __init__(self, config, hash):
        self.config = config
        self.hash = hash
    
    # must be implemented
    def stop(self):
        # stop execution
    
    # must be implemented
    def run(self):
        # start execution
    
    # must be implemented    
    def getConfig(self):
        return self.config
    

Add decorator for the register function (in this case module_setup) which will be called with the current configuration as argument and the config hash made of the unique keys. This function can be named anything. Pass along the configurations to the module object, the configHandler catches KeyErrors so if wrong configurations are sent to this function configHandler will notify you about it.

Also follow the naming convention nameModule.py and @moduleManager.register("name") and import the moduleManager, if not the moduleManager will notify you about any errors.

The rest of the module code is omitted but should create a twisted factory object and start this with the reactor in the run method, see the existing modules for an example. For tutorials on programming with Twisted, please see here. There are also some utils to make use of when developing modules, this is done as following:

Socksify:

# import all utils
from utils import *

# in the constructor create a new proxy object
self.prox = proxySelector.ProxySelector()

# in the run method add the following after having created the factory method.
proxyInfo = self.prox.getRandomProxy()
    if proxyInfo == None:
        self.connector = reactor.connectTCP(host, port, factory)
    else:
        proxyHost = proxyInfo['HOST']
        proxyPort = proxyInfo['PORT']
        proxyUser = proxyInfo['USER']
        proxyPass = proxyInfo['PASS']
        socksify = socks5.ProxyClientCreator(reactor, factory)
        if len(proxyUser) == 0:
            self.connector = socksify.connectSocks5Proxy(host, port, proxyHost, proxyPort, "HALE")
        else:
            self.connector = socksify.connectSocks5Proxy(host, port, proxyHost, proxyPort, "HALE", proxyUser, proxyPass)

Connection errors handling in the factory object:

def clientConnectionFailed(self, connector, reason):
    """
    Called on failed connection to server
    """

    moduleCoordinator.ModuleCoordinator().putError("Error connecting to " + self.config['botnet'], self.module)

def clientConnectionLost(self, connector, reason):
    """
    Called on lost connection to server
    """

    moduleCoordinator.ModuleCoordinator().putError("Connection lost to " + self.config['botnet'], self.module)

This will send the errors to the error bucket which is accessible by issuing a showlog in the CLI.

Logging:

# in the factory create the following method to handle logs (note that the hash and config must be sent to the factory)
# and call it in the protocol class with: self.factory.putLog(data)
def putLog(self, log):
    """
    Put log to the event handler
    """
        
    moduleCoordinator.ModuleCoordinator().addEvent(moduleCoordinator.LOG_EVENT, log, self.hash, self.config)

# apply reg expression to look for URLs containing possible malware
# and call it in the protocol class with: self.factory.checkForURL(data)
def checkForURL(self, data):
    """
    Check for URL in the event handler
    """
    
    moduleCoordinator.ModuleCoordinator().addEvent(moduleCoordinator.URL_EVENT, data, self.hash)

# if you module should detect IP numbers of other bots and herders implement the following method in the factory
def addRelIP(self, data):
    """
    Put possible ip related to the botnet being monitored
    in the event handler.
    """
        
    moduleCoordinator.ModuleCoordinator().addEvent(moduleCoordinator.RELIP_EVENT, data, self.hash)

handling related IPs is done by applying a regular expression to be used for the protocol that the module is going to support, in case of irc module, the code looks like this:

checkHost = data.split(':')[1].split(' ')[0].strip()
match = self.factory.expr.findall(checkHost)
if match:
    self.factory.addRelIP(data.split('@')[1].split(' ')[0].strip())

where the regular expression is as follow:

self.expr = re.compile('!~.*?@')
  1. Drag the file to the modules directory. The moduleManager will then automatically import it and check for errors.

  2. In modules.conf edit the configuration, in this case:

    specify unique configs for module

    a * means that all configs contaning the strings equal to the one

    after the * will be treated as unique.

    in this example all pass_grammar, mode_grammar etc should be

    treated as unique keys

    [uniqueKeys] irc = botnet, *grammar, ... etc

    name a section, can be anything as long as its unique

    and add module regname (in this case irc) as module option

    [myIrcConf] module = irc nick = SpyBot channel = #irc ... etc.

  3. Upload the module to the web ui by setting the module name to for example irc, filename ircModule.py and then add a config example for this module.

XMPP bot HOWTO

When sending a request for a botnet to track the request is made as follow to the groupchat coordination room

sensorLoadReq

where all sensors reply with their id and queue length (number of monitored botnets)

sensorLoadAck id=353f6650859547ed06597dbfa1dcfd88 queue=0

The feeder then choose one sensor based on this info like lowest queue length and if these are equal for several sensors, then the sensor id sorted alphabetically with lowest value is chosen.

When the feeder has chosen a sensor it sends a private chat message to the sensor

startTrackReq config

where config is a string representation of the configuration, for example

module=irc botnet=irc.freenode.net etc..

The sensor then replies with with an acknowledgement together with the config hash which can be used to distinguish the botnet logs from the other logs in the share channel. Example of acknowledgment:

startTrackAck hash

if no one else is monitoring this botnet, otherwise a startTrackNack is received if the botnet is already monitored or the sensor does not have the module installed for this botnet. Malware share is done by sensors sending a message like:

fileCaptured hash=353f6650... file content

where the content is Base64 encoded and comes directly after the file hash value.

RESTful Web API

To get access to the api you need a consumer key and secret key, this can be created by the admin and are used with OAuth to authenticate. The following URLs are available to fetch data in JSON format:

http://.../api/botnet will reply with all botnets monitored
http://.../api/botnet/botnethash will reply with the botnet with hash equal to botnethash
http://.../api/host/hostname will reply with all botnets monitored with host equal to hostname
http://.../api/type/module will reply with all botnets monitored with the module
http://.../api/botips/hash will reply with all ips captured by botnet with the value hash
http://.../api/bologs/hash will reply with all logs for botnet with value hash
http://.../api/bofiles/hash will reply with file hashes captured by botnet with value hash
http://.../api/file/hash returns botnet(s) info for those that have captured file with the hash specified
http://.../api/ip/addr will reply with botnet(s) info for those that have detected an IP with number addr

Note that currently only GET requests are possible.