[lug] ifupdown mapping

[Chris Riddoch]

"J. Wayde Allen" <wallen at its.bldrdoc.gov> writes:
> What I can't seem to get my head around is the mapping script.

As others have said[1], this isn't exactly documented. In this thread,
Anthony Towns says, "It's not documented yet because I've been busy on
other things -- actually it might be documented in either the BTS or
in the NMUed ifupdown in unstable."

Looking at the package bug reports, I found bug #86895: [PATCH]
ifupdown: mapping protocol not really documented.

After this bug was open for 2 years 241 days, it finally got fixed by
a non-maintainer update last month, but it being in unstable means
that you probably have an outdated manual page.  (Bringing to the list
my rant in last month's BLUG meeting about why I've switched from
Debian Testing to SuSE 8.2.)

I found another email[2] which describes what's going on with mapping,
I'll just paste it here; hopefully this will make Google cluster
things better when someone searches for

> It starts to become slightly more clear when the
> /usr/share/doc/ifupdown/examples/* are taken into consideration. Given the
> man page example:
> 
>       mapping eth0
>             script /usr/local/sbin/map-scheme
>             map HOME eth0-home
>             map WORK eth0-work
> 
>         iface eth0-home inet static
>             address 192.168.1.1
>             netmask 255.255.255.0
> 
>         iface eth0-work inet dhcp
> 
> The script /usr/local/sbin/map-scheme (which doesn't really exist, it's a
> place holder for a script provided by the user) is called whenever eth0
> needs to be brought up (or something like this). The string it returns
> ("HOME" or "WORK" in this case) determines which iface stanza should be
> used. I'm guessing. I don't have an immediate opportunity to play with it.
> Some data arrives on the command line to the script, some on stdin
> (according to the examples).
> 
> Turns out there's more documentation in the source including diagrams. But
> I don't have a box available with X on it (requires dia to build
> diagrams). The included PDF diagrams in the Debian source package aren't
> scaled for direct viewing (image bleeds off page). The documentation is
> tangled in the source (a la nowebm). So, I have no idea if there's
> anything helpful there.
> 
> I think the theory is that mapping allows an interface to work out its own
> personality issues without user intervention.

This answer was provided by Andrew B. Sweger.
Hope it helps.  I also found this[3] which might help.


1] http://lists.debian.org/debian-devel/2002/debian-devel-200207/msg01697.html
2] http://www.ssc.com/pipermail/linux-list/2002-June/052614.html
3] http://blog.opencg.org/blogs/main/archives/000179.html

-- 
Christopher.Riddoch at colorado.edu
 - epistemological humility -