Chaosreader can trace TCP/UDP/... sessions and fetch application data from tcpdump or snoop logs. This is like an "any-snarf" program, it will fetch telnet sessions, FTP files, HTTP transfers (HTML, GIF, JPEG, ...), SMTP emails, etc ... from the captured data inside the network traffic logs. It creates a html index file that links to all the session details, including realtime replay programs for telnet, rlogin or IRC sessions; and reports such as image reports and HTTP GET/POST content reports. It also creates replay programs for telnet sessions, so that you can play them back in realtime (or even different speeds). Chaosreader can also run in standalone mode - where it invokes tcpdump or snoop (if they are available) to create the log files and then processes them. 01-May-2004, ver 0.94 (check for new versions, http://www.brendangregg.com) (or run a web search for "chaosreader") QUICK USAGE: tcpdump -s9000 -w out1; chaosreader out1; netscape index.html or, snoop -o out1; chaosreader out1; netscape index.html or, ethereal (save as "out1"); chaosreader out1; netscape index.html or, chaosreader -s 5; netscape index.html USAGE: chaosreader [-aehikqrvxAHIRTUXY] [-D dir] [-b port[,...]] [-B port[,...]] [-j IPaddr[,...]] [-J IPaddr[,...]] [-l port[,...]] [-L port[,...]] [-m bytes[k]] [-M bytes[k]] [-o "time"|"size"|"type"|"ip"] [-p port[,...]] [-P port[,...]] infile [infile2 ...] chaosreader -s [mins] | -S [mins[,count]] [-z] [-f 'filter'] chaosreader # Create application session files, indexes -a, --application # Create application session files (default) -e, --everything # Create HTML 2-way & hex files for everything -h # Print a brief help --help # Print verbose help (this) and version --help2 # Print massive help -i, --info # Create info file -q, --quiet # Quiet, no output to screen -r, --raw # Create raw files -v, --verbose # Verbose - Create ALL files .. (except -e) -x, --index # Create index files (default) -A, --noapplication # Exclude application session files -H, --hex # Include hex dumps (slow) -I, --noinfo # Exclude info files -R, --noraw # Exclude raw files -T, --notcp # Exclude TCP traffic -U, --noudp # Exclude UDP traffic -Y, --noicmp # Exclude ICMP traffic -X, --noindex # Exclude index files -k, --keydata # Create extra files for keystroke analysis -D dir --dir dir # Output all files to this directory -b 25,79 --playtcp 25,79 # replay these TCP ports as well (playback) -B 36,42 --playudp 36,42 # replay these UDP ports as well (playback) -l 7,79 --htmltcp 7,79 # Create HTML for these TCP ports as well -L 7,123 --htmludp 7,123 # Create HTML for these UDP ports as well -m 1k --min 1k # Min size of connection to save ("k" for Kb) -M 1024k --max 1k # Max size of connection to save ("k" for Kb) -o size --sort size # sort Order: time/size/type/ip (Default time) -p 21,23 --port 21,23 # Only examine these ports (TCP & UDP) -P 80,81 --noport 80,81 # Exclude these ports (TCP & UDP) -s 5 --runonce 5 # Standalone. Run tcpdump/snoop for 5 mins. -S 5,10 --runmany 5,10 # Standalone, many. 10 samples of 5 mins each. -S 5 --runmany 5 # Standalone, endless. 5 min samples forever. -z --runredo # Standalone, redo. Rereads last run's logs. -j 10.1.2.1 --ipaddr 10.1.2.1 # Only examine these IPs -J 10.1.2.1 --noipaddr 10.1.2.1 # Exclude these IPs -f 'port 7' --filter 'port 7' # With standalone, use this dump filter. eg1, tcpdump -s9000 -w output1 # create tcpdump capture file chaosreader output1 # extract recognised sessions, or, chaosreader -ve output1 # gimme everything, or, chaosreader -p 20,21,23 output1 # only ftp and telnet... eg2, snoop -o output1 # create snoop capture file instead chaosreader output1 # extract recognised sessions... eg3, chaosreader -S 2,5 # Standalone, sniff network 5 times for 2 mins # each. View index.html for progress (or .text) Output Files: Many will be created, run this in a clean directory. Short example, index.html Html index (full details) index.text Text index index.file File index for standalone redo mode image.html HTML report of images getpost.html HTML report of HTTP GET/POST requests session_0001.info Info file describing TCP session #1 session_0001.telnet.html HTML coloured 2-way capture (time sorted) session_0001.telnet.raw Raw data 2-way capture (time sorted) session_0001.telnet.raw1 Raw 1-way capture (assembeled) server->client session_0001.telnet.raw2 Raw 1-way capture (assembeled) client->server session_0002.web.html HTML coloured 2-way session_0002.part_01.html HTTP portion of the above, a HTML file session_0003.web.html HTML coloured 2-way session_0003.part_01.jpeg HTTP portion of the above, a JPEG file session_0004.web.html HTML coloured 2-way session_0004.part_01.gif HTTP portion of the above, a GIF file session_0005.part_01.ftp-data.gz An FTP transfer, a gz file. ... The convention is, session_* TCP Sessions stream_* UDP Streams icmp_* ICMP packets index.html HTML Index index.text Text Index index.file File Index for standalone redo mode only image.html HTML report of images getpost.html HTML report of HTTP GET/POST requests *.info Info file describing the Session/Stream *.raw Raw data 2-way capture (time sorted) *.raw1 Raw 1-way capture (assembeled) server->client *.raw2 Raw 1-way capture (assembeled) client->server *.replay Session replay program (perl) *.partial.* Partial capture (tcpdump/snoop were aware of drops) *.hex.html 2-way Hex dump, rendered in coloured HTML *.hex.text 2-way Hex dump in plain text *.X11.replay X11 replay script (talks X11) *.textX11.replay X11 communicated text replay script (text only) *.textX11.html 2-way text report, rendered in red/blue HTML *.keydata Keystroke delay data file. Used for SSH analysis. Modes: * Normal - eg "chaosreader infile", this is where a tcpdump/snoop file was created previously and chaosreader reads and processes it. * Standalone, once - eg "chaosreader -s 10", this is where chaosreader runs tcpdump/snoop and generates the log file, in this case for 10 i minutes, and then processes the result. Some OS's may not have tcpdump or snoop available so this will not work (instead you may be able to get Ethereal, run it, save to a file, then use normal mode). There is a master index.html and the report index.html in a sub dir, which is of the format out_YYYYMMDD-hhmm, eg "out_20031003-2221". * Standalone, many - eg "chaosreader -S 5,12", this is where chaosreader runs tcpdump/snoop and generates many log files, in this case it samples 12 times for 5 minutes each. While this is running, the master index.html can be viewed to watch progress, which links to minor index.html reports in each sub directory. * Standalone, redo - eg "chaosreader -ve -z", (the -z), this is where a standalone capture was previously performed - and now you would like to reprocess the logs - perhaps with different options (in this case, "-ve"). It reads index.file to determine which capture logs to read. * Standalone, endless - eg "chaosreader -S 5", like standalone many - but runs forever (if you ever had the need?). Watch your disk space! Note: this is a work in progress, some of the code is a little unpolished. Advice: * Run chaosreader in an empty directory. * Create small packet dumps. Chaosreader uses around 5x the dump size in memory. A 100Mb file could need 500Mb of RAM to process. * Your tcpdump may allow "-s0" (entire packet) instead of "-s9000". * Beware of using too much disk space, especially standalone mode. * If you capture too many small connections giving a huge index.html, try using the -m option to ignore small connections. eg "-m 1k". * snoop logs may actually work better. Snoop logs are based on RFC1761, however there are many varients of tcpdump/libpcap and this program cannot read them all. If you have Ethereal you can create snoop logs during the "save as" option. On Solaris use "snoop -o logfile". * tcpdump logs may not be portable between OSs that use different sized timestamps or endian. * Logs are best created in a memory filesystem for speed, usually /tmp. * For X11 or VNC playbacks, first practise by replaying a recent captured session of your own. The biggest problem is colour depth, your screen must match the capture. For X11 check authentication (xhost +), for VNC check the viewers options (-8bit, "Hextile", ...) * SSH analysis can be performed with the "sshkeydata" program as demonstrated on http://www.brendangregg.com/sshanalysis.html . chaosreader provides the input files (*.keydata) that sshkeydata analyses. Bugs: The following assumptions may cause problems (check for new vers); * A lower port number = the service type. Eg with ports 31247 and 23, the actual type of session is telnet (23). This may not work for some things (eg, VNC). * Time based order is more important for 2-way sessions (eg telnet), SEQ order is more import for 1-way transfers (eg ftp-data). * One particular TCP session isn't active for long enough that the SEQ number loops (or even wraps). WARNING: Please don't use this software for anything illegal. That definition differs for every country, please check the law first. This is a great network troubleshooting and development tool, not a "cracking" or "hacking" tool - a misidentification that could render owning this software illegal in some countries. SEE ALSO: ethereal (GUI packet viewer), dsniff (sniffing toolkit) COPYRIGHT: Copyright (c) 2003, 2004 Brendan Gregg. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. (http://www.gnu.org/copyleft/gpl.html) Author: Brendan Gregg [Sydney, Australia]