[3595] in linux-net channel archive
mrouted tunnel partial success
daemon@ATHENA.MIT.EDU (Malcolm Beattie)
Mon Jul 8 06:34:56 1996
From: Malcolm Beattie <malcolm.beattie@computing-services.oxford.ac.uk>
To: linux-net@vger.rutgers.edu
Date: Mon, 8 Jul 1996 10:46:36 +0100 (BST)
Cc: alan@cymru.net
With the kernel patches I mentioned in a previous message, I've
set up a partially successful mrouted tunnel. There are three hosts
involved: mcast and boris are on one subnet, plutonium is on another.
+-------+ +-------+ tunnel +-----------+
| boris |-----| mcast |- - - - - -| plutonium |
+-------+ eth +-------+ +-----------+
plutonium and boris are running Linux 2.0.0 + Alan's patchkit +
skb_reserve fix. mcast is running SunOS with an old non-pruning
mrouted (3.3 I think). I set up a single mrouted tunnel between
mcast and plutonium. Various multicasting programs started on
mcast and plutonium couldn't reach each other. However, starting
sdr (the mbone session application) on plutonium and boris leads
to interesting behaviour. A session created by boris appears on
sdr on plutonium but a session created on plutonium does not appear
on boris. It looks as though multicast datagrams that mcast's mrouted
sees on an ethernet vif get propagated down the tunnel and seen by
plutonium. However, multicast datagrams sent out by plutonium are
not propagated down its tunnel. On the other hand, IGMP packets
generated when a group is joined on plutonium are seen and logged by
plutonium's mrouted. It looks like there's some "loopback" (in a
general sense) oddity, perhaps with the forwarding of IP-in-IP?
Hope this helps to narrow down the problem a bit.
--Malcolm
--
Malcolm Beattie <mbeattie@sable.ox.ac.uk>
Unix Systems Programmer
Oxford University Computing Services
