The Tcpdump Group git mirrors - tcpdump/blob - README.md

   1 # TCPDUMP 4.x.y by [The Tcpdump Group](https://www.tcpdump.org/)
   2
   3 [![Build Status](https://api.travis-ci.com/the-tcpdump-group/tcpdump.svg?branch=master)](https://travis-ci.com/github/the-tcpdump-group/tcpdump)
   4 [![Build Status](https://ci.appveyor.com/api/projects/status/github/the-tcpdump-group/tcpdump?branch=master&svg=true)](https://ci.appveyor.com/project/guyharris/tcpdump)
   5
   6 **To report a security issue please send an e-mail to security@tcpdump.org.**
   7
   8 To report bugs and other problems, contribute patches, request a
   9 feature, provide generic feedback etc please see the
  10 [guidelines for contributing](CONTRIBUTING) in the tcpdump source tree root.
  11
  12 Anonymous Git is available via
  13
  14         https://github.com/the-tcpdump-group/tcpdump.git
  15
  16 This directory contains source code for tcpdump, a tool for network
  17 monitoring and data acquisition.
  18
  19 Over the past few years, tcpdump has been steadily improved by the
  20 excellent contributions from the Internet community (just browse
  21 through the [change log](CHANGES)).  We are grateful for all the input.
  22
  23 ### Dependency on libpcap
  24 Tcpdump uses libpcap, a system-independent interface for user-level
  25 packet capture.  Before building tcpdump, you must first retrieve and
  26 build libpcap.
  27
  28 Once libpcap is built (either install it or make sure it's in
  29 `../libpcap`), you can build tcpdump using the procedure in the
  30 [installation guide](INSTALL.txt).
  31
  32 ### Origins of tcpdump
  33 The program is loosely based on SMI's "etherfind" although none of the
  34 etherfind code remains.  It was originally written by Van Jacobson as
  35 part of an ongoing research project to investigate and improve TCP and
  36 Internet gateway performance.  The parts of the program originally
  37 taken from Sun's etherfind were later re-written by Steven McCanne of
  38 LBL.  To insure that there would be no vestige of proprietary code in
  39 tcpdump, Steve wrote these pieces from the specification given by the
  40 manual entry, with no access to the source of tcpdump or etherfind.
  41 ```text
  42 formerly from   Lawrence Berkeley National Laboratory
  43                 Network Research Group <tcpdump@ee.lbl.gov>
  44                 ftp://ftp.ee.lbl.gov/old/tcpdump.tar.Z (3.4)
  45 ```
  46
  47 ### See also
  48 Richard Stevens gives an excellent treatment of the Internet protocols
  49 in his book *"TCP/IP Illustrated, Volume 1"*. If you want to learn more
  50 about tcpdump and how to interpret its output, pick up this book.
  51
  52 Some tools for viewing and analyzing tcpdump trace files are available
  53 from the [Internet Traffic Archive](http://ita.ee.lbl.gov/).
  54
  55 Another tool that tcpdump users might find useful is
  56 [tcpslice](https://github.com/the-tcpdump-group/tcpslice).
  57 It is a program that can be used to extract portions of tcpdump binary
  58 trace files.
  59
  60 ### The original LBL README by Steve McCanne, Craig Leres and Van Jacobson
  61 ```
  62 This directory also contains some short awk programs intended as
  63 examples of ways to reduce tcpdump data when you're tracking
  64 particular network problems:
  65
  66 send-ack.awk
  67         Simplifies the tcpdump trace for an ftp (or other unidirectional
  68         tcp transfer).  Since we assume that one host only sends and
  69         the other only acks, all address information is left off and
  70         we just note if the packet is a "send" or an "ack".
  71
  72         There is one output line per line of the original trace.
  73         Field 1 is the packet time in decimal seconds, relative
  74         to the start of the conversation.  Field 2 is delta-time
  75         from last packet.  Field 3 is packet type/direction.
  76         "Send" means data going from sender to receiver, "ack"
  77         means an ack going from the receiver to the sender.  A
  78         preceding "*" indicates that the data is a retransmission.
  79         A preceding "-" indicates a hole in the sequence space
  80         (i.e., missing packet(s)), a "#" means an odd-size (not max
  81         seg size) packet.  Field 4 has the packet flags
  82         (same format as raw trace).  Field 5 is the sequence
  83         number (start seq. num for sender, next expected seq number
  84         for acks).  The number in parens following an ack is
  85         the delta-time from the first send of the packet to the
  86         ack.  A number in parens following a send is the
  87         delta-time from the first send of the packet to the
  88         current send (on duplicate packets only).  Duplicate
  89         sends or acks have a number in square brackets showing
  90         the number of duplicates so far.
  91
  92         Here is a short sample from near the start of an ftp:
  93                 3.00    0.20   send . 512
  94                 3.20    0.20    ack . 1024  (0.20)
  95                 3.20    0.00   send P 1024
  96                 3.40    0.20    ack . 1536  (0.20)
  97                 3.80    0.40 * send . 0  (3.80) [2]
  98                 3.82    0.02 *  ack . 1536  (0.62) [2]
  99         Three seconds into the conversation, bytes 512 through 1023
 100         were sent.  200ms later they were acked.  Shortly thereafter
 101         bytes 1024-1535 were sent and again acked after 200ms.
 102         Then, for no apparent reason, 0-511 is retransmitted, 3.8
 103         seconds after its initial send (the round trip time for this
 104         ftp was 1sec, +-500ms).  Since the receiver is expecting
 105         1536, 1536 is re-acked when 0 arrives.
 106
 107 packetdat.awk
 108         Computes chunk summary data for an ftp (or similar
 109         unidirectional tcp transfer). [A "chunk" refers to
 110         a chunk of the sequence space -- essentially the packet
 111         sequence number divided by the max segment size.]
 112
 113         A summary line is printed showing the number of chunks,
 114         the number of packets it took to send that many chunks
 115         (if there are no lost or duplicated packets, the number
 116         of packets should equal the number of chunks) and the
 117         number of acks.
 118
 119         Following the summary line is one line of information
 120         per chunk.  The line contains eight fields:
 121            1 - the chunk number
 122            2 - the start sequence number for this chunk
 123            3 - time of first send
 124            4 - time of last send
 125            5 - time of first ack
 126            6 - time of last ack
 127            7 - number of times chunk was sent
 128            8 - number of times chunk was acked
 129         (all times are in decimal seconds, relative to the start
 130         of the conversation.)
 131
 132         As an example, here is the first part of the output for
 133         an ftp trace:
 134
 135         # 134 chunks.  536 packets sent.  508 acks.
 136         1       1       0.00    5.80    0.20    0.20    4       1
 137         2       513     0.28    6.20    0.40    0.40    4       1
 138         3       1025    1.16    6.32    1.20    1.20    4       1
 139         4       1561    1.86    15.00   2.00    2.00    6       1
 140         5       2049    2.16    15.44   2.20    2.20    5       1
 141         6       2585    2.64    16.44   2.80    2.80    5       1
 142         7       3073    3.00    16.66   3.20    3.20    4       1
 143         8       3609    3.20    17.24   3.40    5.82    4       11
 144         9       4097    6.02    6.58    6.20    6.80    2       5
 145
 146         This says that 134 chunks were transferred (about 70K
 147         since the average packet size was 512 bytes).  It took
 148         536 packets to transfer the data (i.e., on the average
 149         each chunk was transmitted four times).  Looking at,
 150         say, chunk 4, we see it represents the 512 bytes of
 151         sequence space from 1561 to 2048.  It was first sent
 152         1.86 seconds into the conversation.  It was last
 153         sent 15 seconds into the conversation and was sent
 154         a total of 6 times (i.e., it was retransmitted every
 155         2 seconds on the average).  It was acked once, 140ms
 156         after it first arrived.
 157
 158 stime.awk
 159 atime.awk
 160         Output one line per send or ack, respectively, in the form
 161                 <time> <seq. number>
 162         where <time> is the time in seconds since the start of the
 163         transfer and <seq. number> is the sequence number being sent
 164         or acked.  I typically plot this data looking for suspicious
 165         patterns.
 166
 167
 168 The problem I was looking at was the bulk-data-transfer
 169 throughput of medium delay network paths (1-6 sec.  round trip
 170 time) under typical DARPA Internet conditions.  The trace of the
 171 ftp transfer of a large file was used as the raw data source.
 172 The method was:
 173
 174   - On a local host (but not the Sun running tcpdump), connect to
 175     the remote ftp.
 176
 177   - On the monitor Sun, start the trace going.  E.g.,
 178       tcpdump host local-host and remote-host and port ftp-data >tracefile
 179
 180   - On local, do either a get or put of a large file (~500KB),
 181     preferably to the null device (to minimize effects like
 182     closing the receive window while waiting for a disk write).
 183
 184   - When transfer is finished, stop tcpdump.  Use awk to make up
 185     two files of summary data (maxsize is the maximum packet size,
 186     tracedata is the file of tcpdump tracedata):
 187       awk -f send-ack.awk packetsize=avgsize tracedata >sa
 188       awk -f packetdat.awk packetsize=avgsize tracedata >pd
 189
 190   - While the summary data files are printing, take a look at
 191     how the transfer behaved:
 192       awk -f stime.awk tracedata | xgraph
 193     (90% of what you learn seems to happen in this step).
 194
 195   - Do all of the above steps several times, both directions,
 196     at different times of day, with different protocol
 197     implementations on the other end.
 198
 199   - Using one of the Unix data analysis packages (in my case,
 200     S and Gary Perlman's Unix|Stat), spend a few months staring
 201     at the data.
 202
 203   - Change something in the local protocol implementation and
 204     redo the steps above.
 205
 206   - Once a week, tell your funding agent that you're discovering
 207     wonderful things and you'll write up that research report
 208     "real soon now".
 209 ```