Merge branch 'master' into net-next

[mirror_iproute2.git] / doc / arpd.sgml

<!doctype linuxdoc system>

<article>

<title>ARPD Daemon
<author>Alexey Kuznetsov, <tt/kuznet@ms2.inr.ac.ru/
<date>some_negative_number, 20 Sep 2001
<abstract>
<tt/arpd/ is daemon collecting gratuitous ARP information, saving
it on local disk and feeding it to kernel on demand to avoid
redundant broadcasting due to limited size of kernel ARP cache. 
</abstract>


<p><bf/Description/

<p>The format of the command is:

<tscreen><verb>
       arpd OPTIONS [ INTERFACE [ INTERFACE ... ] ]
</verb></tscreen>

<p> <tt/OPTIONS/ are:

<itemize>

<item><tt/-l/ - dump <tt/arpd/ database to stdout and exit. Output consists
of three columns: interface index, IP address and MAC address.
Negative entries for dead hosts are also shown, in this case MAC address
is replaced by word <tt/FAILED/ followed by colon and time when the fact
that host is dead was proven the last time.

<item><tt/-f FILE/  - read and load <tt/arpd/ database from <tt/FILE/
in text format similar dumped by option <tt/-l/. Exit after load,
probably listing resulting database, if option <tt/-l/ is also given.
If <tt/FILE/ is <tt/-/, <tt/stdin/ is read to get ARP table.
 
<item><tt/-b DATABASE/  - location of database file. Default location is
<tt>/var/lib/arpd/arpd.db</tt>.

<item><tt/-a NUMBER/ - <tt/arpd/ not only passively listens ARP on wire, but
also send brodcast queries itself. <tt/NUMBER/ is number of such queries
to make before destination is considered as dead. When <tt/arpd/ is started
as kernel helper (i.e. with <tt/app_solicit/ enabled in <tt/sysctl/
or even with option <tt/-k/) without this option and still did not learn enough
information, you can observe 1 second gaps in service. Not fatal, but
not good.

<item><tt/-k/ - suppress sending broadcast queries by kernel. It takes
sense together with option <tt/-a/.

<item><tt/-n TIME/ - timeout of negative cache. When resolution fails <tt/arpd/
suppresses further attempts to resolve for this period. It makes sense
only together with option <tt/-k/. This timeout should not be too much
longer than boot time of a typical host not supporting gratuitous ARP.
Default value is 60 seconds.

<item><tt/-R RATE/ - maximal steady rate of broadcasts sent by <tt/arpd/
in packets per second. Default value is 1.

<item><tt/-B NUMBER/ - number of broadcasts sent by <tt/arpd/ back to back.
Default value is 3. Together with option <tt/-R/ this option allows
to police broadcasting not to exceed <tt/B+R*T/ over any interval
of time <tt/T/.

</itemize>

<p><tt/INTERFACE/ is name of networking inteface to watch.
If no interfaces given, <tt/arpd/ monitors all the interfaces.
In this case <tt/arpd/ does not adjust <tt/sysctl/ parameters,
it is supposed user does this himself after <tt/arpd/ is started.


<p> Signals

<p> <tt/arpd/ exits gracefully syncing database and restoring adjusted
<tt/sysctl/ parameters, when receives <tt/SIGINT/ or <tt/SIGTERM/.
<tt/SIGHUP/ syncs database to disk. <tt/SIGUSR1/ sends some statistics
to <tt/syslog/. Effect of another signals is undefined, they may corrupt
database and leave <tt/sysctl/ parameters in an unpredictable state.

<p> Note

<p> In order to <tt/arpd/ be able to serve as ARP resolver, kernel must be
compiled with the option <tt/CONFIG_ARPD/ and, in the case when interface list
is not given on command line, variable <tt/app_solicit/
on interfaces of interest should be set in <tt>/proc/sys/net/ipv4/neigh/*</tt>.
If this is not made <tt/arpd/ still collects gratuitous ARP information
in its database.

<p> Examples

<enum>
<item> Start <tt/arpd/ to collect gratuitous ARP, but not messing
with kernel functionality:

<tscreen><verb>
   arpd -b /var/tmp/arpd.db
</verb></tscreen>

<item> Look at result after some time:

<tscreen><verb>
   killall arpd
   arpd -l -b /var/tmp/arpd.db
</verb></tscreen>

<item> To enable kernel helper, leaving leading role to kernel:

<tscreen><verb>
   arpd -b /var/tmp/arpd.db -a 1 eth0 eth1
</verb></tscreen>

<item> Completely replace kernel resolution on interfaces <tt/eth0/
and <tt/eth1/. In this case kernel still does unicast probing to
validate entries, but all the broadcast activity is suppressed
and made under authority of <tt/arpd/: 

<tscreen><verb>
   arpd -b /var/tmp/arpd.db -a 3 -k eth0 eth1
</verb></tscreen>

This is mode which <tt/arpd/ is supposed to work normally.
It is not default just to prevent occasional enabling of too aggressive
mode occasionally.

</enum>

</article>