Difference between revisions of "Fingerprint Web Application (OTG-INFO-009)"

From OWASP
Jump to: navigation, search
(Common Application Identifiers)
 
(6 intermediate revisions by 3 users not shown)
Line 2: Line 2:
  
 
== Summary ==
 
== Summary ==
Web server fingerprinting is a critical task for the Penetration tester. Knowing the version and type of a running web server allows testers to determine known vulnerabilities and the appropriate exploits to use during testing.
 
  
There are several different vendors and versions of web servers on the market today. Knowing the type of web server that you are testing significantly helps in the testing process, and will also change the course of the test. This information can be derived by sending the web server specific commands and analyzing the output, as each version of web server software may respond differently to these commands. By knowing how each type of web server responds to specific commands and keeping this information in a web server fingerprint database, a penetration tester can send these commands to the web server, analyze the response, and compare it to the database of known signatures. Please note that it usually takes several different commands to accurately identify the web server, as different versions may react similarly to the same command. Rarely, however, different versions react the same to all HTTP commands. So, by sending several different commands, you increase the accuracy of your guess.
+
There is nothing new under the sun, and nearly every web application that one may think of developing has already been developed. With the vast number of free and open source software projects that are actively developed and deployed around the world, it is very likely that an application security test will face a target site that is entirely or partly dependent on these well known applications (e.g. Wordpress, phpBB, Mediawiki, etc). Knowing the web application components that are being tested significantly helps in the testing process and will also drastically reduce the effort required during the test. These well known web applications have known HTML headers, cookies, and directory structures that can be enumerated to identify the application.  
 +
 
  
 
== Test Objectives ==
 
== Test Objectives ==
  
== How to Test ==
+
Identify the web application and version to determine known vulnerabilities and the appropriate exploits to use during testing.
  
=== Black Box testing and example ===
 
The simplest and most basic form of identifying a Web server is to look at the Server field in the HTTP response header. For our experiments we use netcat.
 
Consider the following HTTP Request-Response:
 
<pre>
 
$ nc 202.41.76.251 80
 
HEAD / HTTP/1.0
 
  
HTTP/1.1 200 OK
+
== How to Test ==
Date: Mon, 16 Jun 2003 02:53:29 GMT
+
Server: Apache/1.3.3 (Unix)  (Red Hat/Linux)
+
Last-Modified: Wed, 07 Oct 1998 11:18:14 GMT
+
ETag: "1813-49b-361b4df6"
+
Accept-Ranges: bytes
+
Content-Length: 1179
+
Connection: close
+
Content-Type: text/html
+
</pre>
+
  
From the ''Server'' field, we understand that the server is likely Apache, version 1.3.3, running on Linux operating system.
+
=== Cookies ===
 +
A relatively reliable way to identify a web application is by the application-specific cookies.
  
Four examples of the HTTP response headers are shown below.
+
Consider the following HTTP-request:
  
From an '''Apache 1.3.23''' server:
 
 
<pre>
 
<pre>
HTTP/1.1 200 OK
+
GET / HTTP/1.1
Date: Sun, 15 Jun 2003 17:10: 49 GMT
+
User-Agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:31.0) Gecko/20100101 Firefox/31.0
Server: Apache/1.3.23
+
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Last-Modified: Thu, 27 Feb 2003 03:48: 19 GMT
+
Accept-Language: en-US,en;q=0.5
ETag: 32417-c4-3e5d8a83
+
'''Cookie: wp-settings-time-1=1406093286; wp-settings-time-2=1405988284'''
Accept-Ranges: bytes
+
DNT: 1
Content-Length: 196
+
Connection: keep-alive
Connection: close
+
Host: blog.owasp.org
Content-Type: text/HTML
+
 
</pre>
 
</pre>
  
From a '''Microsoft IIS 5.0''' server:
+
The cookie ''CAKEPHP'' has automatically been set, which gives information about the framework being used. List of common cookies names is presented in Cpmmon Application Identifiers section. However, it is possible to change the name of the cookie.
<pre>
+
HTTP/1.1 200 OK
+
Server: Microsoft-IIS/5.0
+
Expires: Yours, 17 Jun 2003 01:41: 33 GMT
+
Date: Mon, 16 Jun 2003 01:41: 33 GMT
+
Content-Type: text/HTML
+
Accept-Ranges: bytes
+
Last-Modified: Wed, 28 May 2003 15:32: 21 GMT
+
ETag: b0aac0542e25c31: 89d
+
Content-Length: 7369
+
</pre>
+
  
From a '''Netscape Enterprise 4.1''' server:
 
<pre>
 
HTTP/1.1 200 OK
 
Server: Netscape-Enterprise/4.1
 
Date: Mon, 16 Jun 2003 06:19: 04 GMT
 
Content-type: text/HTML
 
Last-modified: Wed, 31 Jul 2002 15:37: 56 GMT
 
Content-length: 57
 
Accept-ranges: bytes
 
Connection: close
 
</pre>
 
  
From a '''SunONE 6.1''' server:
+
=== HTML source code ===
<pre>
+
This technique is based on finding certain patterns in the HTML page source code. Often one can find a lot of information which helps a tester to recognize a specific web application. One of the common markers are HTML comments that directly lead to application disclosure. More often certain application-specific paths can be found, i.e. links to application-specific css and/or js folders. Finally, specific script variables might also point to a certain application.
HTTP/1.1 200 OK
+
 
Server: Sun-ONE-Web-Server/6.1
+
From the meta tag below, one can easily learn the application used by a website and its version. The comment, specific paths and script variables can all help an attacker to quickly determine an instance of an application.
Date: Tue, 16 Jan 2007 14:53:45 GMT
+
 
Content-length: 1186
+
Content-type: text/html
+
Date: Tue, 16 Jan 2007 14:50:31 GMT
+
Last-Modified: Wed, 10 Jan 2007 09:58:26 GMT
+
Accept-Ranges: bytes
+
Connection: close
+
</pre>
+
However, this testing methodology is not so good. There are several techniques that allow a web site to obfuscate or to modify the server banner string.
+
For example we could obtain the following answer:
+
 
<pre>
 
<pre>
403 HTTP/1.1 Forbidden
+
<meta name="generator" content="WordPress 3.9.2" />
Date: Mon, 16 Jun 2003 02:41: 27 GMT
+
Server: Unknown-Webserver/1.0
+
Connection: close
+
Content-Type: text/HTML; charset=iso-8859-1
+
 
</pre>
 
</pre>
  
In this case, the server field of that response is obfuscated: we cannot know what type of web server is running.
+
More frequently such information is placed between <head></head> tags, in <meta> tags or at the end of the page. Nevertheless, it is recommended to check the whole document since it can be useful for other purposes such as inspection of other useful comments and hidden fields.  
  
==== Protocol behaviour ====
+
=== Specific files and folders ===
More refined techniques take in consideration various characteristics of the several web servers available on the market. We will list some methodologies that allow us to deduce the type of web server in use.
+
Apart from information gathered from HTML sources, there is another approach which greatly helps an attacker to determine the application with high accuracy. Every application has its own specific file and folder structure on the server. It has been pointed out that one can see the specific path from the HTML page source but sometimes they are not explicitly presented there and still reside on the server.
  
'''HTTP header field ordering'''
+
In order to uncover them a technique known as dirbusting is used. Dirbusting is brute forcing a target with predictable folder and file names and monitoring HTTP-responses to emumerate server contents. This information can be used both for finding default files and attacking them, and for fingerprinting the web application. Dirbusting can be done in several ways, the example below shows a successful dirbusting attack against a WordPress-powered target with the help of defined list and intruder functionality of Burp Suite.
  
The first method consists of observing the ordering of the several headers in the response. Every web server has an inner ordering of the header. We consider the following answers as an example:
+
[[Image:Wordpress_dirbusting.png]]
  
Response from '''Apache 1.3.23'''
+
We can see that for some WordPress-specific folders (for instance, /wp-includes/, /wp-admin/ and /wp-content/) HTTP-reponses are 403 (Forbidden), 302 (Found, redirection to wp-login.php) and 200 (OK) respectively. This is a good indicator that the target is WordPress-powered. The same way it is possible to dirbust different application plugin folders and their versions. On the screenshot below one can see a typical CHANGELOG file of a Drupal plugin, which provides information on the application being used and discloses a vulnerable plugin version.
<pre>
+
$ nc apache.example.com 80
+
HEAD / HTTP/1.0
+
  
HTTP/1.1 200 OK
+
[[Image:Drupal_botcha_disclosure.png]]
Date: Sun, 15 Jun 2003 17:10: 49 GMT
+
Server: Apache/1.3.23
+
Last-Modified: Thu, 27 Feb 2003 03:48: 19 GMT
+
ETag: 32417-c4-3e5d8a83
+
Accept-Ranges: bytes
+
Content-Length: 196
+
Connection: close
+
Content-Type: text/HTML
+
</pre>
+
Response from '''IIS 5.0'''
+
<pre>
+
$ nc iis.example.com 80
+
HEAD / HTTP/1.0
+
  
HTTP/1.1 200 OK
+
Tip: before starting dirbusting, it is recommended to check the robots.txt file first. Sometimes application specific folders and other sensitive information can be found there as well. An example of such a robots.txt file is presented on a screenshot below.
Server: Microsoft-IIS/5.0
+
Content-Location: http://iis.example.com/Default.htm
+
Date: Fri, 01 Jan 1999 20:13: 52 GMT
+
Content-Type: text/HTML
+
Accept-Ranges: bytes
+
Last-Modified: Fri, 01 Jan 1999 20:13: 52 GMT
+
ETag: W/e0d362a4c335be1: ae1
+
Content-Length: 133
+
</pre>
+
Response from '''Netscape Enterprise 4.1'''
+
<pre>
+
$ nc netscape.example.com 80
+
HEAD / HTTP/1.0
+
  
HTTP/1.1 200 OK
+
[[Image:Robots_info_disclosure.png]]
Server: Netscape-Enterprise/4.1
+
Date: Mon, 16 Jun 2003 06:01: 40 GMT
+
Content-type: text/HTML
+
Last-modified: Wed, 31 Jul 2002 15:37: 56 GMT
+
Content-length: 57
+
Accept-ranges: bytes
+
Connection: close
+
</pre>
+
Response from a '''SunONE 6.1'''
+
<pre>
+
$ nc sunone.example.com 80
+
HEAD / HTTP/1.0
+
  
HTTP/1.1 200 OK
+
Specific files and folders are different for each specific application. It is recommended to install the corresponding application during penetration tests in order to have better understanding of what infrastructure is presented and what files might be left on the server. However, several good file lists already exist and one good example is FuzzDB wordlists of predictable files/folders (http://code.google.com/p/fuzzdb/).
Server: Sun-ONE-Web-Server/6.1
+
Date: Tue, 16 Jan 2007 15:23:37 GMT
+
Content-length: 0
+
Content-type: text/html
+
Date: Tue, 16 Jan 2007 15:20:26 GMT
+
Last-Modified: Wed, 10 Jan 2007 09:58:26 GMT
+
Connection: close
+
</pre>
+
We can notice that the ordering of the ''Date'' field and the ''Server'' field differs between Apache, Netscape Enterprise, and IIS.
+
  
'''Malformed requests test'''
+
== Common Application Identifiers ==
 +
=== Cookies ===
  
Another useful test to execute involves sending malformed requests or requests of nonexistent pages to the server.
+
{| class="wikitable"
Consider the following HTTP responses.  
+
|-
 +
| phpBB || phpbb3_
 +
|-
 +
| Wordpress || wp-settings
 +
|-
 +
| 1C-Bitrix || BITRIX_
 +
|-
 +
| AMPcms || AMP
 +
|-
 +
| Django CMS || django
 +
|-
 +
| DotNetNuke || DotNetNukeAnonymous
 +
|-
 +
| e107 || e107_tz
 +
|-
 +
| EPiServer || EPiTrace, EPiServer
 +
|-
 +
| Graffiti CMS || graffitibot
 +
|-
 +
| Hotaru CMS || hotaru_mobile
 +
|-
 +
| ImpressCMS || ICMSession
 +
|-
 +
| Indico || MAKACSESSION
 +
|-
 +
| InstantCMS || InstantCMS[logdate]
 +
|-
 +
| Kentico CMS || CMSPreferredCulture
 +
|-
 +
| MODx || SN4[12symb]
 +
|-
 +
| TYPO3 || fe_typo_user
 +
|-
 +
| Dynamicweb || Dynamicweb
 +
|-
 +
| LEPTON || lep[some_numeric_value]+sessionid
 +
|-
 +
| Wix || Domain=.wix.com
 +
|-
 +
| VIVVO || VivvoSessionId
 +
|}
  
Response from '''Apache 1.3.23'''
 
<pre>
 
$ nc apache.example.com 80
 
GET / HTTP/3.0
 
  
HTTP/1.1 400 Bad Request
+
=== HTML source code ===
Date: Sun, 15 Jun 2003 17:12: 37 GMT
+
Server: Apache/1.3.23
+
Connection: close
+
Transfer: chunked
+
Content-Type: text/HTML; charset=iso-8859-1
+
</pre>
+
Response from '''IIS 5.0'''
+
<pre>
+
$ nc iis.example.com 80
+
GET / HTTP/3.0
+
  
HTTP/1.1 200 OK
+
{| class="wikitable"
Server: Microsoft-IIS/5.0
+
|-
Content-Location: http://iis.example.com/Default.htm
+
! Application !! Keyword
Date: Fri, 01 Jan 1999 20:14: 02 GMT
+
|-
Content-Type: text/HTML
+
| Wordpress || <meta name="generator" content="WordPress 3.9.2" />
Accept-Ranges: bytes
+
|-
Last-Modified: Fri, 01 Jan 1999 20:14: 02 GMT
+
| phpBB || <body id="phpbb"
ETag: W/e0d362a4c335be1: ae1
+
|-
Content-Length: 133
+
| Mediawiki || <meta name="generator" content="MediaWiki 1.21.9" />
</pre>
+
|-
Response from '''Netscape Enterprise 4.1'''
+
| Joomla || <meta name="generator" content="Joomla! - Open Source Content Management" />
<pre>
+
|-
$ nc netscape.example.com 80
+
| Drupal || <meta name="Generator" content="Drupal 7 (http://drupal.org)" />
GET / HTTP/3.0
+
|-
 +
| DotNetNuke || DNN Platform - http://www.dnnsoftware.com
 +
|}
  
HTTP/1.1 505 HTTP Version Not Supported
+
More info https://www.owasp.org/index.php/Web-metadata
Server: Netscape-Enterprise/4.1
+
Date: Mon, 16 Jun 2003 06:04: 04 GMT
+
Content-length: 140
+
Content-type: text/HTML
+
Connection: close
+
</pre>
+
Response from a '''SunONE 6.1'''
+
<pre>
+
$ nc sunone.example.com 80
+
GET / HTTP/3.0
+
  
HTTP/1.1 400 Bad request
+
== Tools ==
Server: Sun-ONE-Web-Server/6.1
+
A list of general and well-known tools is presented below. There are also a lot of other utilities, as well as framework-based fingerprinting tools.
Date: Tue, 16 Jan 2007 15:25:00 GMT
+
Content-length: 0
+
Content-type: text/html
+
Connection: close
+
</pre>
+
We notice that every server answers in a different way. The answer also differs in the version of the server. Similar observations can be done we create requests with a non-existent protocol. Consider the following responses:
+
  
Response from '''Apache 1.3.23'''
+
=== WhatWeb ===
<pre>
+
Website:  http://www.morningstarsecurity.com/research/whatweb <br>
$ nc apache.example.com 80
+
Currently one of the best fingerprinting tools on the market. Included in a default [[Kali Linux]] build.
GET / JUNK/1.0
+
Language: Ruby
 +
Matches for fingerprinting are made with:
 +
* Text strings (case sensitive)
 +
* Regular expressions
 +
* Google Hack Database queries (limited set of keywords)
 +
* MD5 hashes
 +
* URL recognition
 +
* HTML tag patterns
 +
* Custom ruby code for passive and aggressive operations
  
HTTP/1.1 200 OK
 
Date: Sun, 15 Jun 2003 17:17: 47 GMT
 
Server: Apache/1.3.23
 
Last-Modified: Thu, 27 Feb 2003 03:48: 19 GMT
 
ETag: 32417-c4-3e5d8a83
 
Accept-Ranges: bytes
 
Content-Length: 196
 
Connection: close
 
Content-Type: text/HTML
 
</pre>
 
Response from '''IIS 5.0'''
 
<pre>
 
$ nc iis.example.com 80
 
GET / JUNK/1.0
 
  
HTTP/1.1 400 Bad Request
+
Sample output is presented on a screenshot below:
Server: Microsoft-IIS/5.0
+
 
Date: Fri, 01 Jan 1999 20:14: 34 GMT
+
[[Image:whatweb-sample.png]]
Content-Type: text/HTML
+
Content-Length: 87
+
</pre>
+
Response from '''Netscape Enterprise 4.1'''
+
<pre>
+
$ nc netscape.example.com 80
+
GET / JUNK/1.0
+
  
<HTML><HEAD><TITLE>Bad request</TITLE></HEAD>
+
 
<BODY><H1>Bad request</H1>  
+
 
Your browser sent to query this server could not understand.  
+
=== BlindElephant ===
</BODY></HTML>
+
Website: https://community.qualys.com/community/blindelephant <br>
</pre>
+
This great tool works on the principle of static file checksum based version difference thus providing a very high quality of fingerprinting.
Response from a '''SunONE 6.1'''
+
Language: Python
 +
 
 +
Sample output of a successful fingerprint:
 
<pre>
 
<pre>
$ nc sunone.example.com 80
+
pentester$ python BlindElephant.py http://my_target drupal
GET / JUNK/1.0
+
Loaded /Library/Python/2.7/site-packages/blindelephant/dbs/drupal.pkl with 145 versions, 478 differentiating paths, and 434 version groups.
 +
Starting BlindElephant fingerprint for version of drupal at http://my_target
  
<HTML><HEAD><TITLE>Bad request</TITLE></HEAD>
+
Hit http://my_target/CHANGELOG.txt
<BODY><H1>Bad request</H1>
+
File produced no match. Error: Retrieved file doesn't match known fingerprint. 527b085a3717bd691d47713dff74acf4
Your browser sent a query this server could not understand.
+
</BODY></HTML>
+
</pre>
+
  
== Tools ==
+
Hit http://my_target/INSTALL.txt
* httprint - http://net-square.com/httprint.html
+
File produced no match. Error: Retrieved file doesn't match known fingerprint. 14dfc133e4101be6f0ef5c64566da4a4
* httprecon - http://www.computec.ch/projekte/httprecon/
+
* Netcraft - http://www.netcraft.com
+
* Desenmascarame - http://desenmascara.me
+
* Shodan - http://www.shodanhq.com
+
* Nmap - http://nmap.org
+
  
=== Automated Testing ===
+
Hit http://my_target/misc/drupal.js
Rather than rely on manual bannering and analysis of the web server headers, a tester can use automated tools to achieve the same purpose. The tests to carry out in order to accurately fingerprint a web server can be many. Luckily, there are tools that automate these tests. "''httprint''" is one of such tools. httprint has a signature dictionary that allows one to recognize the type and the version of the web server in use.<br>
+
Possible versions based on result: 7.12, 7.13, 7.14
An example of running httprint is shown below:<br><br>
+
  
[[Image:httprint.jpg |800px|]]
+
Hit http://my_target/MAINTAINERS.txt
 +
File produced no match. Error: Retrieved file doesn't match known fingerprint. 36b740941a19912f3fdbfcca7caa08ca
  
 +
Hit http://my_target/themes/garland/style.css
 +
Possible versions based on result: 7.2, 7.3, 7.4, 7.5, 7.6, 7.7, 7.8, 7.9, 7.10, 7.11, 7.12, 7.13, 7.14
  
[http://www.nmap.org Nmap] version detection offers a lot of advanced features that can help in determining services that are running on a given host, it obtains all data by connecting to open ports and interrogating them by using probes that the specific services understand, the following example shows how Nmap connected to port 80 in order to fingerprint the service and its current version
+
...
  
<pre>
+
Fingerprinting resulted in:
localhost$ nmap -sV example.com
+
7.14
Starting Nmap 6.40 ( http://nmap.org ) at 2013-09-21 13:20 GST
+
Nmap scan report for example.com (127.0.0.1)
+
Host is up (0.028s latency).
+
Not shown: 997 filtered ports
+
PORT    STATE  SERVICE    VERSION
+
80/tcp  open  http      Microsoft IIS httpd 6.0
+
Service Info: OS: Windows; CPE: cpe:/o:microsoft:windows
+
</pre>
+
  
=== Online Testing ===
 
Online tools can be used if the tester wishes to test more stealthily and doesn't wish to directly connect to the target website. An example of online tool that often delivers a lot of information on target Web Server, are [http://www.netcraft.com Netcraft] and [http://www.shodanhq.com SHODAN]
 
  
With [http://www.netcraft.com Netcraft] we can retrieve information about operating system, web server used, Server Uptime, Netblock Owner, history of change related to Web server and O.S.<br> An example is shown below:
+
Best Guess: 7.14
<br><br>
+
</pre>
  
[[Image:netcraft2.png |800px|]]
 
  
 +
=== Wappalyzer ===
 +
Website: http://wappalyzer.com <br>
 +
Wapplyzer is a Firefox Chrome plug-in. It works only on regular expression matching and doesn't need anything other than the page to be loaded on browser. It works completely at the browser level and gives results in the form of icons. Although sometimes it has false positives, this is very handy to have notion of what technologies were used to construct a target website immediately after browsing a page.
  
[http://www.shodanhp.com SHODAN] combines an HTTP port scanner with a search engine index of the HTTP responses, making it trivial to find specific web servers. Shodan collects data mostly on web servers at the moment (HTTP port 80), but there is also some data from FTP (21), SSH (22) Telnet (23), SNMP (161) and SIP (5060) services. <br> An example is shown below:
 
<br><br>
 
  
[[File:Shodan.png |800px|]]
+
Sample output of a plug-in is presented on a screenshot below.
  
 +
[[Image:Owasp-wappalyzer.png]]
  
[[OWASP Unmaskme Project]] expect becomes another online tool to do fingerprinting in any website with an overall interpretation of all the [[Web-metadata]] extracted. The idea behind this project is that anyone in charge of a website could test the metadata their site is showing to the world and assess it from a security point of view.
 
While this project is being developed, you can test a [http://desenmascara.me/ Spanish Proof of Concept of this idea].
 
  
== Vulnerability References ==
+
== References ==
 
'''Whitepapers'''<br>
 
'''Whitepapers'''<br>
 
* Saumil Shah: "An Introduction to HTTP fingerprinting" - http://www.net-square.com/httprint_paper.html
 
* Saumil Shah: "An Introduction to HTTP fingerprinting" - http://www.net-square.com/httprint_paper.html
 
* Anant Shrivastava : "Web Application Finger Printing" - http://anantshri.info/articles/web_app_finger_printing.html
 
* Anant Shrivastava : "Web Application Finger Printing" - http://anantshri.info/articles/web_app_finger_printing.html
* Nmap "Service and Application Version Detection" - http://nmap.org/book/vscan.html
+
 
  
 
== Remediation ==
 
== Remediation ==
 +
The general advice is to use several of the tools described above and check logs to better understand what exactly helps an attacker to disclose the web framework. By performing multiple scans after changes have been made to hide framework tracks, it's possible to achieve a better level of security and to make sure of the framework can not be detected by automatic scans. Below are some specific recommendations by framework marker location and some additional interesting approaches.
 +
 +
 +
==== HTTP headers ====
 +
Check the configuration and disable or obfuscate all HTTP-headers that disclose information the technologies used. Here is an interesting article about HTTP-headers obfuscation using Netscaler:
 +
http://grahamhosking.blogspot.ru/2013/07/obfuscating-http-header-using-netscaler.html
 +
 +
 +
==== Cookies ====
 +
It is recommended to change cookie names by making changes in the corresponding configuration files.
 +
 +
 +
==== HTML source code ====
 +
Manually check the contents of the HTML code and remove everything that explicitly points to the framework.
 +
 +
General guidelines:
 +
*Make sure there are no visual markers disclosing the framework
 +
*Remove any unnecessary comments (copyrights, bug information, specific framework comments)
 +
*Remove META and generator tags
 +
*Use the companies own css or js files and do not store those in a framework-specific folders
 +
*Do not use default scripts on the page or obfuscate them if they must be used.
 +
 +
 +
==== Specific files and folders ====
 +
General guidelines:
 +
*Remove any unnecessary or unused files on the server. This implies text files disclosing information about versions and installation too.
 +
*Restrict access to other files in order to achieve 404-response when accessing them from outside. This can be done, for example, by modifying htaccess file and adding RewriteCond or RewriteRule there. An example of such restriction for two common WordPress folders is presented below.
 +
<pre>
 +
RewriteCond %{REQUEST_URI} /wp-login\.php$ [OR]
 +
RewriteCond %{REQUEST_URI} /wp-admin/$
 +
RewriteRule $ /http://your_website [R=404,L]
 +
</pre>
 +
 +
 +
However, these are not the only ways to restrict access. In order to automate this process, certain framework-specific plugins exist. One example for WordPress is StealthLogin (http://wordpress.org/plugins/stealth-login-page).
  
Protect the presentation layer web server behind a hardened reverse proxy.
 
  
Obfuscate the presentation layer web server headers.
+
==== Additional approaches ====
* Apache
+
General guidelines:
* IIS
+
*Checksum management
 +
*:The purpose of this approach is to beat checksum-based scanners and not let them disclose files by their hashes. Generally, there are two approaches in checksum management:
 +
*:*Change the location of where those files are placed (i.e. move them to another folder, or rename the existing folder)
 +
*:*Modify the contents - even slight modification results in a completely different hash sum, so adding a single byte in the end of the file should not be a big problem.
 +
*Controlled chaos
 +
*:A funny and effective method that involves adding bogus files and folders from other frameworks in order to fool scanners and confuse an attacker. But be careful not to overwrite existing files and folders and to break the current framework!

Latest revision as of 09:09, 22 September 2014

This article is part of the new OWASP Testing Guide v4.
Back to the OWASP Testing Guide v4 ToC: https://www.owasp.org/index.php/OWASP_Testing_Guide_v4_Table_of_Contents Back to the OWASP Testing Guide Project: https://www.owasp.org/index.php/OWASP_Testing_Project

Summary

There is nothing new under the sun, and nearly every web application that one may think of developing has already been developed. With the vast number of free and open source software projects that are actively developed and deployed around the world, it is very likely that an application security test will face a target site that is entirely or partly dependent on these well known applications (e.g. Wordpress, phpBB, Mediawiki, etc). Knowing the web application components that are being tested significantly helps in the testing process and will also drastically reduce the effort required during the test. These well known web applications have known HTML headers, cookies, and directory structures that can be enumerated to identify the application.


Test Objectives

Identify the web application and version to determine known vulnerabilities and the appropriate exploits to use during testing.


How to Test

Cookies

A relatively reliable way to identify a web application is by the application-specific cookies.

Consider the following HTTP-request:

GET / HTTP/1.1
User-Agent: Mozilla/5.0 (Windows NT 6.2; WOW64; rv:31.0) Gecko/20100101 Firefox/31.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: en-US,en;q=0.5
'''Cookie: wp-settings-time-1=1406093286; wp-settings-time-2=1405988284'''
DNT: 1
Connection: keep-alive
Host: blog.owasp.org

The cookie CAKEPHP has automatically been set, which gives information about the framework being used. List of common cookies names is presented in Cpmmon Application Identifiers section. However, it is possible to change the name of the cookie.


HTML source code

This technique is based on finding certain patterns in the HTML page source code. Often one can find a lot of information which helps a tester to recognize a specific web application. One of the common markers are HTML comments that directly lead to application disclosure. More often certain application-specific paths can be found, i.e. links to application-specific css and/or js folders. Finally, specific script variables might also point to a certain application.

From the meta tag below, one can easily learn the application used by a website and its version. The comment, specific paths and script variables can all help an attacker to quickly determine an instance of an application.

<meta name="generator" content="WordPress 3.9.2" />

More frequently such information is placed between <head></head> tags, in <meta> tags or at the end of the page. Nevertheless, it is recommended to check the whole document since it can be useful for other purposes such as inspection of other useful comments and hidden fields.

Specific files and folders

Apart from information gathered from HTML sources, there is another approach which greatly helps an attacker to determine the application with high accuracy. Every application has its own specific file and folder structure on the server. It has been pointed out that one can see the specific path from the HTML page source but sometimes they are not explicitly presented there and still reside on the server.

In order to uncover them a technique known as dirbusting is used. Dirbusting is brute forcing a target with predictable folder and file names and monitoring HTTP-responses to emumerate server contents. This information can be used both for finding default files and attacking them, and for fingerprinting the web application. Dirbusting can be done in several ways, the example below shows a successful dirbusting attack against a WordPress-powered target with the help of defined list and intruder functionality of Burp Suite.

Wordpress dirbusting.png

We can see that for some WordPress-specific folders (for instance, /wp-includes/, /wp-admin/ and /wp-content/) HTTP-reponses are 403 (Forbidden), 302 (Found, redirection to wp-login.php) and 200 (OK) respectively. This is a good indicator that the target is WordPress-powered. The same way it is possible to dirbust different application plugin folders and their versions. On the screenshot below one can see a typical CHANGELOG file of a Drupal plugin, which provides information on the application being used and discloses a vulnerable plugin version.

Drupal botcha disclosure.png

Tip: before starting dirbusting, it is recommended to check the robots.txt file first. Sometimes application specific folders and other sensitive information can be found there as well. An example of such a robots.txt file is presented on a screenshot below.

File:Robots info disclosure.png

Specific files and folders are different for each specific application. It is recommended to install the corresponding application during penetration tests in order to have better understanding of what infrastructure is presented and what files might be left on the server. However, several good file lists already exist and one good example is FuzzDB wordlists of predictable files/folders (http://code.google.com/p/fuzzdb/).

Common Application Identifiers

Cookies

phpBB phpbb3_
Wordpress wp-settings
1C-Bitrix BITRIX_
AMPcms AMP
Django CMS django
DotNetNuke DotNetNukeAnonymous
e107 e107_tz
EPiServer EPiTrace, EPiServer
Graffiti CMS graffitibot
Hotaru CMS hotaru_mobile
ImpressCMS ICMSession
Indico MAKACSESSION
InstantCMS InstantCMS[logdate]
Kentico CMS CMSPreferredCulture
MODx SN4[12symb]
TYPO3 fe_typo_user
Dynamicweb Dynamicweb
LEPTON lep[some_numeric_value]+sessionid
Wix Domain=.wix.com
VIVVO VivvoSessionId


HTML source code

Application Keyword
Wordpress <meta name="generator" content="WordPress 3.9.2" />
phpBB <body id="phpbb"
Mediawiki <meta name="generator" content="MediaWiki 1.21.9" />
Joomla <meta name="generator" content="Joomla! - Open Source Content Management" />
Drupal <meta name="Generator" content="Drupal 7 (http://drupal.org)" />
DotNetNuke DNN Platform - http://www.dnnsoftware.com

More info https://www.owasp.org/index.php/Web-metadata

Tools

A list of general and well-known tools is presented below. There are also a lot of other utilities, as well as framework-based fingerprinting tools.


WhatWeb

Website: http://www.morningstarsecurity.com/research/whatweb
Currently one of the best fingerprinting tools on the market. Included in a default Kali Linux build. Language: Ruby Matches for fingerprinting are made with:

  • Text strings (case sensitive)
  • Regular expressions
  • Google Hack Database queries (limited set of keywords)
  • MD5 hashes
  • URL recognition
  • HTML tag patterns
  • Custom ruby code for passive and aggressive operations


Sample output is presented on a screenshot below:

Whatweb-sample.png


BlindElephant

Website: https://community.qualys.com/community/blindelephant
This great tool works on the principle of static file checksum based version difference thus providing a very high quality of fingerprinting. Language: Python

Sample output of a successful fingerprint:

pentester$ python BlindElephant.py http://my_target drupal
Loaded /Library/Python/2.7/site-packages/blindelephant/dbs/drupal.pkl with 145 versions, 478 differentiating paths, and 434 version groups.
Starting BlindElephant fingerprint for version of drupal at http://my_target 

Hit http://my_target/CHANGELOG.txt
File produced no match. Error: Retrieved file doesn't match known fingerprint. 527b085a3717bd691d47713dff74acf4 

Hit http://my_target/INSTALL.txt
File produced no match. Error: Retrieved file doesn't match known fingerprint. 14dfc133e4101be6f0ef5c64566da4a4 

Hit http://my_target/misc/drupal.js
Possible versions based on result: 7.12, 7.13, 7.14

Hit http://my_target/MAINTAINERS.txt
File produced no match. Error: Retrieved file doesn't match known fingerprint. 36b740941a19912f3fdbfcca7caa08ca 

Hit http://my_target/themes/garland/style.css
Possible versions based on result: 7.2, 7.3, 7.4, 7.5, 7.6, 7.7, 7.8, 7.9, 7.10, 7.11, 7.12, 7.13, 7.14

...

Fingerprinting resulted in:
7.14


Best Guess: 7.14


Wappalyzer

Website: http://wappalyzer.com
Wapplyzer is a Firefox Chrome plug-in. It works only on regular expression matching and doesn't need anything other than the page to be loaded on browser. It works completely at the browser level and gives results in the form of icons. Although sometimes it has false positives, this is very handy to have notion of what technologies were used to construct a target website immediately after browsing a page.


Sample output of a plug-in is presented on a screenshot below.

Owasp-wappalyzer.png


References

Whitepapers


Remediation

The general advice is to use several of the tools described above and check logs to better understand what exactly helps an attacker to disclose the web framework. By performing multiple scans after changes have been made to hide framework tracks, it's possible to achieve a better level of security and to make sure of the framework can not be detected by automatic scans. Below are some specific recommendations by framework marker location and some additional interesting approaches.


HTTP headers

Check the configuration and disable or obfuscate all HTTP-headers that disclose information the technologies used. Here is an interesting article about HTTP-headers obfuscation using Netscaler: http://grahamhosking.blogspot.ru/2013/07/obfuscating-http-header-using-netscaler.html


Cookies

It is recommended to change cookie names by making changes in the corresponding configuration files.


HTML source code

Manually check the contents of the HTML code and remove everything that explicitly points to the framework.

General guidelines:

  • Make sure there are no visual markers disclosing the framework
  • Remove any unnecessary comments (copyrights, bug information, specific framework comments)
  • Remove META and generator tags
  • Use the companies own css or js files and do not store those in a framework-specific folders
  • Do not use default scripts on the page or obfuscate them if they must be used.


Specific files and folders

General guidelines:

  • Remove any unnecessary or unused files on the server. This implies text files disclosing information about versions and installation too.
  • Restrict access to other files in order to achieve 404-response when accessing them from outside. This can be done, for example, by modifying htaccess file and adding RewriteCond or RewriteRule there. An example of such restriction for two common WordPress folders is presented below.
RewriteCond %{REQUEST_URI} /wp-login\.php$ [OR]
RewriteCond %{REQUEST_URI} /wp-admin/$
RewriteRule $ /http://your_website [R=404,L]


However, these are not the only ways to restrict access. In order to automate this process, certain framework-specific plugins exist. One example for WordPress is StealthLogin (http://wordpress.org/plugins/stealth-login-page).


Additional approaches

General guidelines:

  • Checksum management
    The purpose of this approach is to beat checksum-based scanners and not let them disclose files by their hashes. Generally, there are two approaches in checksum management:
    • Change the location of where those files are placed (i.e. move them to another folder, or rename the existing folder)
    • Modify the contents - even slight modification results in a completely different hash sum, so adding a single byte in the end of the file should not be a big problem.
  • Controlled chaos
    A funny and effective method that involves adding bogus files and folders from other frameworks in order to fool scanners and confuse an attacker. But be careful not to overwrite existing files and folders and to break the current framework!