1214478Srpaulo@(#) $Header: /tcpdump/master/tcpdump/README,v 1.68 2008-12-15 00:05:27 guy Exp $ (LBL) 217680Spst 3214478SrpauloTCPDUMP 4.x.y 475115SfennerNow maintained by "The Tcpdump Group" 575115SfennerSee www.tcpdump.org 617680Spst 7190207SrpauloPlease send inquiries/comments/reports to: 8190207Srpaulo tcpdump-workers@lists.tcpdump.org 975115Sfenner 10214478SrpauloAnonymous Git is available via: 11214478Srpaulo git clone git://bpf.tcpdump.org/tcpdump 1275115Sfenner 13214478SrpauloVersion 4.x.y of TCPDUMP can be retrieved with the CVS tag "tcpdump_4_xrely": 14214478Srpaulo cvs -d :pserver:cvs.tcpdump.org:/tcpdump/master checkout -r tcpdump_4_xrely tcpdump 1575115Sfenner 16251158SdelphijPlease submit patches by forking the branch on GitHub at 1775115Sfenner 18251158Sdelphij http://github.com/mcr/tcpdump/tree/master 19251158Sdelphij 20251158Sdelphijand issuing a pull request. 21251158Sdelphij 2275115Sfennerformerly from Lawrence Berkeley National Laboratory 2375115Sfenner Network Research Group <tcpdump@ee.lbl.gov> 2475115Sfenner ftp://ftp.ee.lbl.gov/tcpdump.tar.Z (3.4) 2575115Sfenner 2617680SpstThis directory contains source code for tcpdump, a tool for network 2775115Sfennermonitoring and data acquisition. This software was originally 2875115Sfennerdeveloped by the Network Research Group at the Lawrence Berkeley 2975115SfennerNational Laboratory. The original distribution is available via 3075115Sfenneranonymous ftp to ftp.ee.lbl.gov, in tcpdump.tar.Z. More recent 3175115Sfennerdevelopment is performed at tcpdump.org, http://www.tcpdump.org/ 3217680Spst 3375115SfennerTcpdump uses libpcap, a system-independent interface for user-level 3417680Spstpacket capture. Before building tcpdump, you must first retrieve and 3575115Sfennerbuild libpcap, also originally from LBL and now being maintained by 3675115Sfennertcpdump.org; see http://www.tcpdump.org/ . 3717680Spst 3817680SpstOnce libpcap is built (either install it or make sure it's in 3917680Spst../libpcap), you can build tcpdump using the procedure in the INSTALL 4017680Spstfile. 4117680Spst 4239297SfennerThe program is loosely based on SMI's "etherfind" although none of the 4339297Sfenneretherfind code remains. It was originally written by Van Jacobson as 4439297Sfennerpart of an ongoing research project to investigate and improve tcp and 4539297Sfennerinternet gateway performance. The parts of the program originally 4639297Sfennertaken from Sun's etherfind were later re-written by Steven McCanne of 4739297SfennerLBL. To insure that there would be no vestige of proprietary code in 4839297Sfennertcpdump, Steve wrote these pieces from the specification given by the 4939297Sfennermanual entry, with no access to the source of tcpdump or etherfind. 5017680Spst 5139297SfennerOver the past few years, tcpdump has been steadily improved by the 5239297Sfennerexcellent contributions from the Internet community (just browse 5339297Sfennerthrough the CHANGES file). We are grateful for all the input. 5417680Spst 5539297SfennerRichard Stevens gives an excellent treatment of the Internet protocols 5639297Sfennerin his book ``TCP/IP Illustrated, Volume 1''. If you want to learn more 5739297Sfennerabout tcpdump and how to interpret its output, pick up this book. 5817680Spst 5917680SpstSome tools for viewing and analyzing tcpdump trace files are available 6017680Spstfrom the Internet Traffic Archive: 6117680Spst 6239297Sfenner http://www.acm.org/sigcomm/ITA/ 6317680Spst 6439297SfennerAnother tool that tcpdump users might find useful is tcpslice: 6539297Sfenner 6639297Sfenner ftp://ftp.ee.lbl.gov/tcpslice.tar.Z 6739297Sfenner 6839297SfennerIt is a program that can be used to extract portions of tcpdump binary 6939297Sfennertrace files. See the above distribution for further details and 7039297Sfennerdocumentation. 7139297Sfenner 72111726SfennerProblems, bugs, questions, desirable enhancements, etc. should be sent 73190207Srpauloto the address "tcpdump-workers@lists.tcpdump.org". Bugs, support 74251158Sdelphijrequests, and feature requests may also be submitted on the GitHub issue 75251158Sdelphijtracker for tcpdump at 7617680Spst 77251158Sdelphij https://github.com/mcr/tcpdump/issues 7875115Sfenner 79111726SfennerSource code contributions, etc. should be sent to the email address 80251158Sdelphijabove or submitted by forking the branch on GitHub at 8175115Sfenner 82251158Sdelphij http://github.com/mcr/tcpdump/tree/master 83111726Sfenner 84251158Sdelphijand issuing a pull request. 85251158Sdelphij 86251158SdelphijCurrent versions can be found at www.tcpdump.org. 87251158Sdelphij 8875115Sfenner - The TCPdump team 8975115Sfenner 9075115Sfenneroriginal text by: Steve McCanne, Craig Leres, Van Jacobson 9175115Sfenner 9217680Spst------------------------------------- 9317680SpstThis directory also contains some short awk programs intended as 9417680Spstexamples of ways to reduce tcpdump data when you're tracking 9517680Spstparticular network problems: 9617680Spst 9717680Spstsend-ack.awk 9817680Spst Simplifies the tcpdump trace for an ftp (or other unidirectional 9917680Spst tcp transfer). Since we assume that one host only sends and 10017680Spst the other only acks, all address information is left off and 10117680Spst we just note if the packet is a "send" or an "ack". 10217680Spst 10317680Spst There is one output line per line of the original trace. 10417680Spst Field 1 is the packet time in decimal seconds, relative 10517680Spst to the start of the conversation. Field 2 is delta-time 10617680Spst from last packet. Field 3 is packet type/direction. 10717680Spst "Send" means data going from sender to receiver, "ack" 10817680Spst means an ack going from the receiver to the sender. A 10917680Spst preceding "*" indicates that the data is a retransmission. 11017680Spst A preceding "-" indicates a hole in the sequence space 11117680Spst (i.e., missing packet(s)), a "#" means an odd-size (not max 11217680Spst seg size) packet. Field 4 has the packet flags 11317680Spst (same format as raw trace). Field 5 is the sequence 11417680Spst number (start seq. num for sender, next expected seq number 11517680Spst for acks). The number in parens following an ack is 11617680Spst the delta-time from the first send of the packet to the 11717680Spst ack. A number in parens following a send is the 11817680Spst delta-time from the first send of the packet to the 11917680Spst current send (on duplicate packets only). Duplicate 12017680Spst sends or acks have a number in square brackets showing 12117680Spst the number of duplicates so far. 12217680Spst 12317680Spst Here is a short sample from near the start of an ftp: 12417680Spst 3.00 0.20 send . 512 12517680Spst 3.20 0.20 ack . 1024 (0.20) 12617680Spst 3.20 0.00 send P 1024 12717680Spst 3.40 0.20 ack . 1536 (0.20) 12817680Spst 3.80 0.40 * send . 0 (3.80) [2] 12917680Spst 3.82 0.02 * ack . 1536 (0.62) [2] 13017680Spst Three seconds into the conversation, bytes 512 through 1023 13117680Spst were sent. 200ms later they were acked. Shortly thereafter 13217680Spst bytes 1024-1535 were sent and again acked after 200ms. 13317680Spst Then, for no apparent reason, 0-511 is retransmitted, 3.8 13417680Spst seconds after its initial send (the round trip time for this 13517680Spst ftp was 1sec, +-500ms). Since the receiver is expecting 13617680Spst 1536, 1536 is re-acked when 0 arrives. 13717680Spst 13817680Spstpacketdat.awk 13917680Spst Computes chunk summary data for an ftp (or similar 14017680Spst unidirectional tcp transfer). [A "chunk" refers to 14117680Spst a chunk of the sequence space -- essentially the packet 14217680Spst sequence number divided by the max segment size.] 14317680Spst 14417680Spst A summary line is printed showing the number of chunks, 14517680Spst the number of packets it took to send that many chunks 14617680Spst (if there are no lost or duplicated packets, the number 14717680Spst of packets should equal the number of chunks) and the 14817680Spst number of acks. 14917680Spst 15017680Spst Following the summary line is one line of information 15117680Spst per chunk. The line contains eight fields: 15217680Spst 1 - the chunk number 15317680Spst 2 - the start sequence number for this chunk 15417680Spst 3 - time of first send 15517680Spst 4 - time of last send 15617680Spst 5 - time of first ack 15717680Spst 6 - time of last ack 15817680Spst 7 - number of times chunk was sent 15917680Spst 8 - number of times chunk was acked 16017680Spst (all times are in decimal seconds, relative to the start 16117680Spst of the conversation.) 16217680Spst 16317680Spst As an example, here is the first part of the output for 16417680Spst an ftp trace: 16517680Spst 16617680Spst # 134 chunks. 536 packets sent. 508 acks. 16717680Spst 1 1 0.00 5.80 0.20 0.20 4 1 16817680Spst 2 513 0.28 6.20 0.40 0.40 4 1 16917680Spst 3 1025 1.16 6.32 1.20 1.20 4 1 17017680Spst 4 1561 1.86 15.00 2.00 2.00 6 1 17117680Spst 5 2049 2.16 15.44 2.20 2.20 5 1 17217680Spst 6 2585 2.64 16.44 2.80 2.80 5 1 17317680Spst 7 3073 3.00 16.66 3.20 3.20 4 1 17417680Spst 8 3609 3.20 17.24 3.40 5.82 4 11 17517680Spst 9 4097 6.02 6.58 6.20 6.80 2 5 17617680Spst 17717680Spst This says that 134 chunks were transferred (about 70K 17817680Spst since the average packet size was 512 bytes). It took 17917680Spst 536 packets to transfer the data (i.e., on the average 18017680Spst each chunk was transmitted four times). Looking at, 18117680Spst say, chunk 4, we see it represents the 512 bytes of 18217680Spst sequence space from 1561 to 2048. It was first sent 18317680Spst 1.86 seconds into the conversation. It was last 18417680Spst sent 15 seconds into the conversation and was sent 18517680Spst a total of 6 times (i.e., it was retransmitted every 18617680Spst 2 seconds on the average). It was acked once, 140ms 18717680Spst after it first arrived. 18817680Spst 18917680Spststime.awk 19017680Spstatime.awk 19117680Spst Output one line per send or ack, respectively, in the form 19217680Spst <time> <seq. number> 19317680Spst where <time> is the time in seconds since the start of the 19417680Spst transfer and <seq. number> is the sequence number being sent 19517680Spst or acked. I typically plot this data looking for suspicious 19617680Spst patterns. 19717680Spst 19817680Spst 19917680SpstThe problem I was looking at was the bulk-data-transfer 20017680Spstthroughput of medium delay network paths (1-6 sec. round trip 20117680Spsttime) under typical DARPA Internet conditions. The trace of the 20217680Spstftp transfer of a large file was used as the raw data source. 20317680SpstThe method was: 20417680Spst 20517680Spst - On a local host (but not the Sun running tcpdump), connect to 20617680Spst the remote ftp. 20717680Spst 20817680Spst - On the monitor Sun, start the trace going. E.g., 20917680Spst tcpdump host local-host and remote-host and port ftp-data >tracefile 21017680Spst 21117680Spst - On local, do either a get or put of a large file (~500KB), 21217680Spst preferably to the null device (to minimize effects like 21317680Spst closing the receive window while waiting for a disk write). 21417680Spst 21517680Spst - When transfer is finished, stop tcpdump. Use awk to make up 21617680Spst two files of summary data (maxsize is the maximum packet size, 21717680Spst tracedata is the file of tcpdump tracedata): 21817680Spst awk -f send-ack.awk packetsize=avgsize tracedata >sa 21917680Spst awk -f packetdat.awk packetsize=avgsize tracedata >pd 22017680Spst 22117680Spst - While the summary data files are printing, take a look at 22217680Spst how the transfer behaved: 22317680Spst awk -f stime.awk tracedata | xgraph 22417680Spst (90% of what you learn seems to happen in this step). 22517680Spst 22617680Spst - Do all of the above steps several times, both directions, 22717680Spst at different times of day, with different protocol 22817680Spst implementations on the other end. 22917680Spst 23017680Spst - Using one of the Unix data analysis packages (in my case, 23117680Spst S and Gary Perlman's Unix|Stat), spend a few months staring 23217680Spst at the data. 23317680Spst 23417680Spst - Change something in the local protocol implementation and 23517680Spst redo the steps above. 23617680Spst 23717680Spst - Once a week, tell your funding agent that you're discovering 23817680Spst wonderful things and you'll write up that research report 23917680Spst "real soon now". 240