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

   1 # TCPDUMP 4.x.y by [The Tcpdump Group](https://www.tcpdump.org/)
   2
   3 **To report a security issue please send an e-mail to security@tcpdump.org.**
   4
   5 To report bugs and other problems, contribute patches, request a
   6 feature, provide generic feedback etc please see the
   7 [guidelines for contributing](CONTRIBUTING.md) in the tcpdump source tree root.
   8
   9 Anonymous Git is available via
  10
  11         https://github.com/the-tcpdump-group/tcpdump.git
  12
  13 This directory contains source code for tcpdump, a tool for network
  14 monitoring and data acquisition.
  15
  16 Over the past few years, tcpdump has been steadily improved by the
  17 excellent contributions from the Internet community (just browse
  18 through the [change log](CHANGES)).  We are grateful for all the input.
  19
  20 ### Supported platforms
  21 In many operating systems tcpdump is available as a native package or port,
  22 which simplifies installation of updates and long-term maintenance. However,
  23 the native packages are sometimes a few versions behind and to try a more
  24 recent snapshot it will take to compile tcpdump from the source code.
  25
  26 tcpdump compiles and works on at least the following platforms:
  27
  28 * [AIX](./doc/README.aix.md)
  29 * DragonFly BSD
  30 * FreeBSD
  31 * [Haiku](./doc/README.haiku.md)
  32 * HP-UX 11i
  33 * [illumos](./doc/README.solaris.md) (OmniOS, OpenIndiana)
  34 * GNU/Hurd
  35 * GNU/Linux
  36 * {Mac} OS X / macOS
  37 * [NetBSD](./doc/README.NetBSD.md)
  38 * OpenBSD
  39 * [Solaris](./doc/README.solaris.md)
  40 * [Windows](./doc/README.windows.md) (requires WinPcap or Npcap, and Visual
  41   Studio with CMake)
  42
  43 In the past tcpdump certainly or likely worked on the following platforms:
  44
  45 * 4.3BSD
  46 * BSD/386, later BSD/OS
  47 * DEC OSF/1, later Digital UNIX, later Tru64 UNIX
  48 * DOS
  49 * IRIX
  50 * LynxOS
  51 * QNX
  52 * SINIX
  53 * SunOS
  54 * Ultrix
  55 * UnixWare
  56
  57 ### Dependency on libpcap
  58 tcpdump uses libpcap, a system-independent interface for user-level
  59 packet capture.  If your operating system does not provide libpcap, or
  60 if it provides a libpcap that does not support the APIs from libpcap 1.0
  61 or later, you must first retrieve and build libpcap before building
  62 tcpdump,
  63
  64 Once libpcap is built (either install it or make sure it's in
  65 `../libpcap`), you can build tcpdump using the procedure in the
  66 [installation notes](INSTALL.md).
  67
  68 ### Origins of tcpdump
  69 The program is loosely based on SMI's "etherfind" although none of the
  70 etherfind code remains.  It was originally written by Van Jacobson as
  71 part of an ongoing research project to investigate and improve TCP and
  72 Internet gateway performance.  The parts of the program originally
  73 taken from Sun's etherfind were later re-written by Steven McCanne of
  74 LBL.  To insure that there would be no vestige of proprietary code in
  75 tcpdump, Steve wrote these pieces from the specification given by the
  76 manual entry, with no access to the source of tcpdump or etherfind.
  77 ```text
  78 formerly from   Lawrence Berkeley National Laboratory
  79                 Network Research Group <tcpdump@ee.lbl.gov>
  80                 ftp://ftp.ee.lbl.gov/old/tcpdump.tar.Z (3.4)
  81 ```
  82
  83 ### See also
  84 Richard Stevens gives an excellent treatment of the Internet protocols
  85 in his book *"TCP/IP Illustrated, Volume 1"*. If you want to learn more
  86 about tcpdump and how to interpret its output, pick up this book.
  87
  88 Another tool that tcpdump users might find useful is
  89 [tcpslice](https://github.com/the-tcpdump-group/tcpslice).
  90 It is a program that can be used to extract portions of tcpdump binary
  91 trace files.
  92
  93 ### The original LBL README by Steve McCanne, Craig Leres and Van Jacobson
  94 ```
  95 This directory also contains some short awk programs intended as
  96 examples of ways to reduce tcpdump data when you're tracking
  97 particular network problems:
  98
  99 send-ack.awk
 100         Simplifies the tcpdump trace for an ftp (or other unidirectional
 101         tcp transfer).  Since we assume that one host only sends and
 102         the other only acks, all address information is left off and
 103         we just note if the packet is a "send" or an "ack".
 104
 105         There is one output line per line of the original trace.
 106         Field 1 is the packet time in decimal seconds, relative
 107         to the start of the conversation.  Field 2 is delta-time
 108         from last packet.  Field 3 is packet type/direction.
 109         "Send" means data going from sender to receiver, "ack"
 110         means an ack going from the receiver to the sender.  A
 111         preceding "*" indicates that the data is a retransmission.
 112         A preceding "-" indicates a hole in the sequence space
 113         (i.e., missing packet(s)), a "#" means an odd-size (not max
 114         seg size) packet.  Field 4 has the packet flags
 115         (same format as raw trace).  Field 5 is the sequence
 116         number (start seq. num for sender, next expected seq number
 117         for acks).  The number in parens following an ack is
 118         the delta-time from the first send of the packet to the
 119         ack.  A number in parens following a send is the
 120         delta-time from the first send of the packet to the
 121         current send (on duplicate packets only).  Duplicate
 122         sends or acks have a number in square brackets showing
 123         the number of duplicates so far.
 124
 125         Here is a short sample from near the start of an ftp:
 126                 3.00    0.20   send . 512
 127                 3.20    0.20    ack . 1024  (0.20)
 128                 3.20    0.00   send P 1024
 129                 3.40    0.20    ack . 1536  (0.20)
 130                 3.80    0.40 * send . 0  (3.80) [2]
 131                 3.82    0.02 *  ack . 1536  (0.62) [2]
 132         Three seconds into the conversation, bytes 512 through 1023
 133         were sent.  200ms later they were acked.  Shortly thereafter
 134         bytes 1024-1535 were sent and again acked after 200ms.
 135         Then, for no apparent reason, 0-511 is retransmitted, 3.8
 136         seconds after its initial send (the round trip time for this
 137         ftp was 1sec, +-500ms).  Since the receiver is expecting
 138         1536, 1536 is re-acked when 0 arrives.
 139
 140 packetdat.awk
 141         Computes chunk summary data for an ftp (or similar
 142         unidirectional tcp transfer). [A "chunk" refers to
 143         a chunk of the sequence space -- essentially the packet
 144         sequence number divided by the max segment size.]
 145
 146         A summary line is printed showing the number of chunks,
 147         the number of packets it took to send that many chunks
 148         (if there are no lost or duplicated packets, the number
 149         of packets should equal the number of chunks) and the
 150         number of acks.
 151
 152         Following the summary line is one line of information
 153         per chunk.  The line contains eight fields:
 154            1 - the chunk number
 155            2 - the start sequence number for this chunk
 156            3 - time of first send
 157            4 - time of last send
 158            5 - time of first ack
 159            6 - time of last ack
 160            7 - number of times chunk was sent
 161            8 - number of times chunk was acked
 162         (all times are in decimal seconds, relative to the start
 163         of the conversation.)
 164
 165         As an example, here is the first part of the output for
 166         an ftp trace:
 167
 168         # 134 chunks.  536 packets sent.  508 acks.
 169         1       1       0.00    5.80    0.20    0.20    4       1
 170         2       513     0.28    6.20    0.40    0.40    4       1
 171         3       1025    1.16    6.32    1.20    1.20    4       1
 172         4       1561    1.86    15.00   2.00    2.00    6       1
 173         5       2049    2.16    15.44   2.20    2.20    5       1
 174         6       2585    2.64    16.44   2.80    2.80    5       1
 175         7       3073    3.00    16.66   3.20    3.20    4       1
 176         8       3609    3.20    17.24   3.40    5.82    4       11
 177         9       4097    6.02    6.58    6.20    6.80    2       5
 178
 179         This says that 134 chunks were transferred (about 70K
 180         since the average packet size was 512 bytes).  It took
 181         536 packets to transfer the data (i.e., on the average
 182         each chunk was transmitted four times).  Looking at,
 183         say, chunk 4, we see it represents the 512 bytes of
 184         sequence space from 1561 to 2048.  It was first sent
 185         1.86 seconds into the conversation.  It was last
 186         sent 15 seconds into the conversation and was sent
 187         a total of 6 times (i.e., it was retransmitted every
 188         2 seconds on the average).  It was acked once, 140ms
 189         after it first arrived.
 190
 191 stime.awk
 192 atime.awk
 193         Output one line per send or ack, respectively, in the form
 194                 <time> <seq. number>
 195         where <time> is the time in seconds since the start of the
 196         transfer and <seq. number> is the sequence number being sent
 197         or acked.  I typically plot this data looking for suspicious
 198         patterns.
 199
 200
 201 The problem I was looking at was the bulk-data-transfer
 202 throughput of medium delay network paths (1-6 sec.  round trip
 203 time) under typical DARPA Internet conditions.  The trace of the
 204 ftp transfer of a large file was used as the raw data source.
 205 The method was:
 206
 207   - On a local host (but not the Sun running tcpdump), connect to
 208     the remote ftp.
 209
 210   - On the monitor Sun, start the trace going.  E.g.,
 211       tcpdump host local-host and remote-host and port ftp-data >tracefile
 212
 213   - On local, do either a get or put of a large file (~500KB),
 214     preferably to the null device (to minimize effects like
 215     closing the receive window while waiting for a disk write).
 216
 217   - When transfer is finished, stop tcpdump.  Use awk to make up
 218     two files of summary data (maxsize is the maximum packet size,
 219     tracedata is the file of tcpdump tracedata):
 220       awk -f send-ack.awk packetsize=avgsize tracedata >sa
 221       awk -f packetdat.awk packetsize=avgsize tracedata >pd
 222
 223   - While the summary data files are printing, take a look at
 224     how the transfer behaved:
 225       awk -f stime.awk tracedata | xgraph
 226     (90% of what you learn seems to happen in this step).
 227
 228   - Do all of the above steps several times, both directions,
 229     at different times of day, with different protocol
 230     implementations on the other end.
 231
 232   - Using one of the Unix data analysis packages (in my case,
 233     S and Gary Perlman's Unix|Stat), spend a few months staring
 234     at the data.
 235
 236   - Change something in the local protocol implementation and
 237     redo the steps above.
 238
 239   - Once a week, tell your funding agent that you're discovering
 240     wonderful things and you'll write up that research report
 241     "real soon now".
 242 ```