iftop/iftop.8

.TH IFTOP 8
.\"
.\" iftop.8:
.\" Manual page for iftop.
.\"
.\" $Id$
.\"

.SH NAME
iftop - display bandwidth usage on an interface by host

.SH SYNOPSIS
\fBiftop\fP \fB-h\fP |
[\fB-npbBP\fP] [\fB-i\fP \fIinterface\fP] [\fB-f\fP \fIfilter code\fP] [\fB-N\fP \fInet\fP/\fImask\fP]

.SH DESCRIPTION
\fBiftop\fP listens to network traffic on a named \fIinterface\fP, or on the
first interface it can find which looks like an external interface if none is
specified, and displays a table of current bandwidth usage by pairs of hosts.
\fBiftop\fP must be run with sufficient permissions to monitor all network
traffic on the \fIinterface\fP; see \fBpcap\fP(3) for more information, but on
most systems this means that it must be run as root.

By default, \fBiftop\fP will look up the hostnames associated with addresses it
finds in packets. This can cause substantial traffic of itself, and may result
in a confusing display. You may wish to suppress display of DNS traffic by
using filter code such as \fBnot port domain\fP, or switch it off entirely,
by using the \fB-n\fP option or by pressing \fBR\fP when the program is running.

By default, \fBiftop\fP counts all IP packets that pass through the filter, and
the direction of the packet is determined according to the direction the packet
is moving across the interface.  Using the \fB-N\fP option it is possible to
get \fBiftop\fP to show packets entering and leaving a given network.  For
example, \fBiftop -N 10.0.0.0/255.0.0.0\fP will analyse packets flowing in and
out of the 10.* network.

Some other filter ideas:
.TP
\fBnot ether host ff:ff:ff:ff:ff:ff\fP
Ignore ethernet broadcast packets.
.TP
\fBport http and not host \fP\fIwebcache.example.com\fP
Count web traffic only, unless it is being directed through a local web cache.
.TP
\fBicmp\fP
How much bandwith are users wasting trying to figure out why the network is
slow?

.SH OPTIONS

.TP
\fB-h\fP
Print a summary of usage.
.TP
\fB-n\fP
Don't do hostname lookups. This setting may be altered at run time.
.TP
\fB-p\fP
Run in promiscuous mode, so that traffic which does not pass directly through
the specified interface is also counted.
.TP
\fB-P\fP
Turn on port display.
.TP
\fB-b\fP
Don't display bar graphs of traffic. This setting may be altered at run time.
.TP
\fB-B\fP
Display bandwidth rates in bytes/sec rather than bits/sec.
.TP
\fB-i\fP \fIinterface\fP
Listen to packets on \fIinterface\fP.
.TP
\fB-f\fP \fIfilter code\fP
Use \fIfilter code\fP to select the packets to count. Only IP packets are ever
counted, so the specified code is evaluated as \fB(\fP\fIfilter code\fP\fB) and ip\fP.
.TP
\fB-N\fP \fInet\fP/\fImask\fP
Specifies a network for traffic analysis.  If specified, iftop will only
include packets flowing in to or out of the given network, and packet direction
is determined relative to the network boundary, rather than to the interface.
You may specify \fImask\fP as a dotted quad, such as /255.255.255.0, or as a
single number specifying the number of bits set in the netmask, such as /24.

.SH DISPLAY

When running, \fBiftop\fP uses the whole screen to display network usage. At
the top of the display is a logarithmic scale for the bar graph which gives a
visual indication of traffic.

The main part of the display lists, for each pair of hosts, the rate at which
data has been sent and received over the preceding 2, 10 and 40 second
intervals. The direction of data flow is indicated by arrows, <= and =>. For
instance,
.nf

foo.example.com  =>  bar.example.com      1Kb  500b   100b
                 <=                       2Mb    2Mb    2Mb

.Sp
.fi
shows, on the first line, traffic from \fBfoo.example.com\fP to
\fBbar.example.com\fP; in the preceding 2 seconds, this averaged 1Kbit/s,
around half that amount over the preceding 10s, and a fifth of that over the
whole of the last 40s. During each of those intervals, the data sent in the
other direction was about 2Mbit/s. On the actual display, part of each line
is inverted to give a visual indication of the 10s average of traffic.
You might expect to see something like this where host \fBfoo\fP is making
repeated HTTP requests to \fBbar\fP, which is sending data back which saturates
a 2Mbit/s link.

By default, the pairs of hosts responsible for the most traffic (10 second
average) are displayed at the top of the list.

At the bottom of the display, various totals are shown, including peak traffic
over the last 40s, total traffic transferred (after filtering), and total
transfer rates averaged over 2s, 10s and 40s.

.SH SOURCE / DEST AGGREGATION

By pressing \fBs\fP or \fBd\fP while \fBiftop\fP is running, all traffic
for each source or destination will be aggregated together.  This is most
useful when \fBiftop\fP is run in promiscuous mode, or is run on a gateway
machine.

.SH PORT DISPLAY

\fBS\fP or \fBD\fP toggle the display of source and destination ports
respectively. \fBp\fP will toggle port display on/off.

.SH DISPLAY ORDER

By default, the display is ordered according to the 10s average (2nd column).
By pressing \fB1\fP, \fB2\fP or \fB3\fP it is possible to sort by the 1st, 2nd
or 3rd column.   By pressing \fB<\fP or \fB>\fP the display will be sorted by
source or destination hostname respectively.

.SH DISPLAY FILTERING

\fBl\fP allows you to enter a POSIX extended regular expression that will be
used to filter hostnames shown in the display.  This is a good way to quickly
limit what is shown on the display.  Note that this happens at a much later
stage than filter code, and does not affect what is actually captured.

.SH PAUSE DISPLAY / FREEZE ORDER

\fBP\fP will pause the current display.

\fBo\fP will freeze the current screen order.  This has the side effect that
traffic between hosts not shown on the screen at the time will not be shown at
all, although it will be included in the totals at the bottom of the screen.

.SH FILTER CODE

\fBf\fP allows you to edit the filter code whilst iftop running.  This
can lead to some unexpected behaviour.

.SH QUIRKS (aka they're features, not bugs)

There are some circumstances in which iftop may not do what you expect.  In
most cases what it is doing is logical, and we believe it is correct behaviour,
although I'm happy to hear reasoned arguments for alternative behaviour.

\fBTotals don't add up\fP

There are several reasons why the totals may not appear to add up.  The
most obvious is having a screen filter in effect, or screen ordering
frozen.  In this case some captured information is not being shown to
you, but is included in the totals.

A more subtle explanation comes about when running in promiscuous mode
without specifying a \fB-N\fP option.  In this case there is no easy way
to assign the direction of traffic between two other hosts.  For the
purposes of the main display this is done in an arbitrary fashion (by
ordering of IP addresses), but for the sake of totals all traffic
between other hosts is accounted as incoming, because that's what it is
from the point of view of your interface.  

\fBPeak totals don't add up\fP

Again, this is a feature.  The peak sent and peak received didn't necessarily
happen at the same time.  The peak total is the maximum of sent plus received
in each captured time division.

\fBChanging the filter code doesn't seem to work\fP

Give it time.  Changing the filter code affects what is captured from
the time that you entered it, but most of what is on the display is
based on some fraction of the last 40s window of capturing.  After
changing the filter there may be entries on the display that are
disallowed by the current filter for up to 40s.  DISPLAY FILTERING has
immediate effect and does not affect what is captured.

.SH SEE ALSO
.BR tcpdump (8),
.BR pcap (3),
.BR driftnet (1).

.SH AUTHOR
Paul Warren <pdw@ex-parrot.com>

.SH VERSION
$Id$

.SH COPYING
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., 675 Mass Ave, Cambridge, MA 02139, USA.