README revision 75115
175115Sfenner@(#) $Header: /tcpdump/master/tcpdump/README,v 1.58 2000/12/08 06:59:11 mcr Exp $ (LBL) 217680Spst 375115SfennerTCPDUMP 3.6 475115SfennerNow maintained by "The Tcpdump Group" 575115SfennerSee www.tcpdump.org 617680Spst 775115SfennerPlease send inquiries/comments/reports to tcpdump-workers@tcpdump.org 875115Sfenner 975115SfennerAnonymous CVS is available via: 1075115Sfenner cvs -d cvs.tcpdump.org:/tcpdump/master login 1175115Sfenner (password "anoncvs") 1275115Sfenner cvs -d cvs.tcpdump.org:/tcpdump/master checkout tcpdump 1375115Sfenner 1475115SfennerVersion 3.6 of TCPDUMP can be retrived with the CVS tag "tcpdump_3_6": 1575115Sfenner cvs -d cvs.tcpdump.org:/tcpdump/master checkout -r tcpdump_3_6 tcpdump 1675115Sfenner 1775115SfennerPlease send patches against the master copy to patches@tcpdump.org. 1875115Sfenner 1975115Sfennerformerly from Lawrence Berkeley National Laboratory 2075115Sfenner Network Research Group <tcpdump@ee.lbl.gov> 2175115Sfenner ftp://ftp.ee.lbl.gov/tcpdump.tar.Z (3.4) 2275115Sfenner 2317680SpstThis directory contains source code for tcpdump, a tool for network 2475115Sfennermonitoring and data acquisition. This software was originally 2575115Sfennerdeveloped by the Network Research Group at the Lawrence Berkeley 2675115SfennerNational Laboratory. The original distribution is available via 2775115Sfenneranonymous ftp to ftp.ee.lbl.gov, in tcpdump.tar.Z. More recent 2875115Sfennerdevelopment is performed at tcpdump.org, http://www.tcpdump.org/ 2917680Spst 3075115SfennerTcpdump uses libpcap, a system-independent interface for user-level 3117680Spstpacket capture. Before building tcpdump, you must first retrieve and 3275115Sfennerbuild libpcap, also originally from LBL and now being maintained by 3375115Sfennertcpdump.org; see http://www.tcpdump.org/ . 3417680Spst 3517680SpstOnce libpcap is built (either install it or make sure it's in 3617680Spst../libpcap), you can build tcpdump using the procedure in the INSTALL 3717680Spstfile. 3817680Spst 3939297SfennerThe program is loosely based on SMI's "etherfind" although none of the 4039297Sfenneretherfind code remains. It was originally written by Van Jacobson as 4139297Sfennerpart of an ongoing research project to investigate and improve tcp and 4239297Sfennerinternet gateway performance. The parts of the program originally 4339297Sfennertaken from Sun's etherfind were later re-written by Steven McCanne of 4439297SfennerLBL. To insure that there would be no vestige of proprietary code in 4539297Sfennertcpdump, Steve wrote these pieces from the specification given by the 4639297Sfennermanual entry, with no access to the source of tcpdump or etherfind. 4717680Spst 4839297SfennerOver the past few years, tcpdump has been steadily improved by the 4939297Sfennerexcellent contributions from the Internet community (just browse 5039297Sfennerthrough the CHANGES file). We are grateful for all the input. 5117680Spst 5239297SfennerRichard Stevens gives an excellent treatment of the Internet protocols 5339297Sfennerin his book ``TCP/IP Illustrated, Volume 1''. If you want to learn more 5439297Sfennerabout tcpdump and how to interpret its output, pick up this book. 5517680Spst 5617680SpstSome tools for viewing and analyzing tcpdump trace files are available 5717680Spstfrom the Internet Traffic Archive: 5817680Spst 5939297Sfenner http://www.acm.org/sigcomm/ITA/ 6017680Spst 6139297SfennerAnother tool that tcpdump users might find useful is tcpslice: 6239297Sfenner 6339297Sfenner ftp://ftp.ee.lbl.gov/tcpslice.tar.Z 6439297Sfenner 6539297SfennerIt is a program that can be used to extract portions of tcpdump binary 6639297Sfennertrace files. See the above distribution for further details and 6739297Sfennerdocumentation. 6839297Sfenner 6975115SfennerProblems, bugs, questions, desirable enhancements, etc. 7075115Sfennershould be sent to the address "tcpdump-workers@tcpdump.org". 7117680Spst 7275115SfennerSource code contributions, etc. should be sent to the email address 7375115Sfenner"patches@tcpdump.org". 7475115Sfenner 7575115SfennerCurrent versions can be found at www.tcpdump.org 7675115Sfenner 7775115Sfenner - The TCPdump team 7875115Sfenner 7975115Sfenneroriginal text by: Steve McCanne, Craig Leres, Van Jacobson 8075115Sfenner 8117680Spst------------------------------------- 8217680SpstThis directory also contains some short awk programs intended as 8317680Spstexamples of ways to reduce tcpdump data when you're tracking 8417680Spstparticular network problems: 8517680Spst 8617680Spstsend-ack.awk 8717680Spst Simplifies the tcpdump trace for an ftp (or other unidirectional 8817680Spst tcp transfer). Since we assume that one host only sends and 8917680Spst the other only acks, all address information is left off and 9017680Spst we just note if the packet is a "send" or an "ack". 9117680Spst 9217680Spst There is one output line per line of the original trace. 9317680Spst Field 1 is the packet time in decimal seconds, relative 9417680Spst to the start of the conversation. Field 2 is delta-time 9517680Spst from last packet. Field 3 is packet type/direction. 9617680Spst "Send" means data going from sender to receiver, "ack" 9717680Spst means an ack going from the receiver to the sender. A 9817680Spst preceding "*" indicates that the data is a retransmission. 9917680Spst A preceding "-" indicates a hole in the sequence space 10017680Spst (i.e., missing packet(s)), a "#" means an odd-size (not max 10117680Spst seg size) packet. Field 4 has the packet flags 10217680Spst (same format as raw trace). Field 5 is the sequence 10317680Spst number (start seq. num for sender, next expected seq number 10417680Spst for acks). The number in parens following an ack is 10517680Spst the delta-time from the first send of the packet to the 10617680Spst ack. A number in parens following a send is the 10717680Spst delta-time from the first send of the packet to the 10817680Spst current send (on duplicate packets only). Duplicate 10917680Spst sends or acks have a number in square brackets showing 11017680Spst the number of duplicates so far. 11117680Spst 11217680Spst Here is a short sample from near the start of an ftp: 11317680Spst 3.00 0.20 send . 512 11417680Spst 3.20 0.20 ack . 1024 (0.20) 11517680Spst 3.20 0.00 send P 1024 11617680Spst 3.40 0.20 ack . 1536 (0.20) 11717680Spst 3.80 0.40 * send . 0 (3.80) [2] 11817680Spst 3.82 0.02 * ack . 1536 (0.62) [2] 11917680Spst Three seconds into the conversation, bytes 512 through 1023 12017680Spst were sent. 200ms later they were acked. Shortly thereafter 12117680Spst bytes 1024-1535 were sent and again acked after 200ms. 12217680Spst Then, for no apparent reason, 0-511 is retransmitted, 3.8 12317680Spst seconds after its initial send (the round trip time for this 12417680Spst ftp was 1sec, +-500ms). Since the receiver is expecting 12517680Spst 1536, 1536 is re-acked when 0 arrives. 12617680Spst 12717680Spstpacketdat.awk 12817680Spst Computes chunk summary data for an ftp (or similar 12917680Spst unidirectional tcp transfer). [A "chunk" refers to 13017680Spst a chunk of the sequence space -- essentially the packet 13117680Spst sequence number divided by the max segment size.] 13217680Spst 13317680Spst A summary line is printed showing the number of chunks, 13417680Spst the number of packets it took to send that many chunks 13517680Spst (if there are no lost or duplicated packets, the number 13617680Spst of packets should equal the number of chunks) and the 13717680Spst number of acks. 13817680Spst 13917680Spst Following the summary line is one line of information 14017680Spst per chunk. The line contains eight fields: 14117680Spst 1 - the chunk number 14217680Spst 2 - the start sequence number for this chunk 14317680Spst 3 - time of first send 14417680Spst 4 - time of last send 14517680Spst 5 - time of first ack 14617680Spst 6 - time of last ack 14717680Spst 7 - number of times chunk was sent 14817680Spst 8 - number of times chunk was acked 14917680Spst (all times are in decimal seconds, relative to the start 15017680Spst of the conversation.) 15117680Spst 15217680Spst As an example, here is the first part of the output for 15317680Spst an ftp trace: 15417680Spst 15517680Spst # 134 chunks. 536 packets sent. 508 acks. 15617680Spst 1 1 0.00 5.80 0.20 0.20 4 1 15717680Spst 2 513 0.28 6.20 0.40 0.40 4 1 15817680Spst 3 1025 1.16 6.32 1.20 1.20 4 1 15917680Spst 4 1561 1.86 15.00 2.00 2.00 6 1 16017680Spst 5 2049 2.16 15.44 2.20 2.20 5 1 16117680Spst 6 2585 2.64 16.44 2.80 2.80 5 1 16217680Spst 7 3073 3.00 16.66 3.20 3.20 4 1 16317680Spst 8 3609 3.20 17.24 3.40 5.82 4 11 16417680Spst 9 4097 6.02 6.58 6.20 6.80 2 5 16517680Spst 16617680Spst This says that 134 chunks were transferred (about 70K 16717680Spst since the average packet size was 512 bytes). It took 16817680Spst 536 packets to transfer the data (i.e., on the average 16917680Spst each chunk was transmitted four times). Looking at, 17017680Spst say, chunk 4, we see it represents the 512 bytes of 17117680Spst sequence space from 1561 to 2048. It was first sent 17217680Spst 1.86 seconds into the conversation. It was last 17317680Spst sent 15 seconds into the conversation and was sent 17417680Spst a total of 6 times (i.e., it was retransmitted every 17517680Spst 2 seconds on the average). It was acked once, 140ms 17617680Spst after it first arrived. 17717680Spst 17817680Spststime.awk 17917680Spstatime.awk 18017680Spst Output one line per send or ack, respectively, in the form 18117680Spst <time> <seq. number> 18217680Spst where <time> is the time in seconds since the start of the 18317680Spst transfer and <seq. number> is the sequence number being sent 18417680Spst or acked. I typically plot this data looking for suspicious 18517680Spst patterns. 18617680Spst 18717680Spst 18817680SpstThe problem I was looking at was the bulk-data-transfer 18917680Spstthroughput of medium delay network paths (1-6 sec. round trip 19017680Spsttime) under typical DARPA Internet conditions. The trace of the 19117680Spstftp transfer of a large file was used as the raw data source. 19217680SpstThe method was: 19317680Spst 19417680Spst - On a local host (but not the Sun running tcpdump), connect to 19517680Spst the remote ftp. 19617680Spst 19717680Spst - On the monitor Sun, start the trace going. E.g., 19817680Spst tcpdump host local-host and remote-host and port ftp-data >tracefile 19917680Spst 20017680Spst - On local, do either a get or put of a large file (~500KB), 20117680Spst preferably to the null device (to minimize effects like 20217680Spst closing the receive window while waiting for a disk write). 20317680Spst 20417680Spst - When transfer is finished, stop tcpdump. Use awk to make up 20517680Spst two files of summary data (maxsize is the maximum packet size, 20617680Spst tracedata is the file of tcpdump tracedata): 20717680Spst awk -f send-ack.awk packetsize=avgsize tracedata >sa 20817680Spst awk -f packetdat.awk packetsize=avgsize tracedata >pd 20917680Spst 21017680Spst - While the summary data files are printing, take a look at 21117680Spst how the transfer behaved: 21217680Spst awk -f stime.awk tracedata | xgraph 21317680Spst (90% of what you learn seems to happen in this step). 21417680Spst 21517680Spst - Do all of the above steps several times, both directions, 21617680Spst at different times of day, with different protocol 21717680Spst implementations on the other end. 21817680Spst 21917680Spst - Using one of the Unix data analysis packages (in my case, 22017680Spst S and Gary Perlman's Unix|Stat), spend a few months staring 22117680Spst at the data. 22217680Spst 22317680Spst - Change something in the local protocol implementation and 22417680Spst redo the steps above. 22517680Spst 22617680Spst - Once a week, tell your funding agent that you're discovering 22717680Spst wonderful things and you'll write up that research report 22817680Spst "real soon now". 229