IPX: Network number collision 3 - eth0 802.3 & eth0 802.2 ?!?!?

This is a /var/log/messages kernel message on a Novel enterprise that includes several other UNIX machines, besides the Linux machine, on which this message is flooding the /var file system, as well as NT 4.0 and Windows 95 machines.  The answer is 200 points if the entire question is answered, in both parts:

a.)  What does it mean and how can it be fixed?

b.)  If IN FACT it cannot be fixed with a simple configuration adjustment, what is the priority level syntax for supressing this message in the syslog.conf and what will be implications of supressing this priority level message?
Who is Participating?
sauronConnect With a Mentor Commented:
What it means:-

Each IPX network is assigned an IPX network number. A machine that can function as an IPX router will have an 'Internal' IPX network, which is confined to that machine alone. All Novell servers can function as IPX routers. Each configured frame type counts as a separate logical IPX network. So, a Novell server running Ethernet_802.2 and Ethernet_802.3 will have 3 IPX network numbers allocated to it. 2 to the bit of wire running each frame type, and a third internal one. IPX network numbers must be unique, for routing reasons.

So, the message means that eth0 is running two frame types (i.e. two IPX networks) that have the same IPX network number. This is illegal.

When adding the ipx interface to a linux box, the sytax is:-

ipx_interface add [-p] device frame_type [network number]

Make sure the network numbers are unique. You should be aware that you can't arbitrarily assign network numbers - they must agree with what other machines think they should be. A logical segment must have an IPX network number assigned to each IPX frame type running on it, and all machines on that segment must agree as to that network number. It is possible that the misconfiguration is not on the linux box at all, but on a Novell server. Check the bind statements in the autoexec.ncf.

Suppressing the warnings is not a good idea - they are the result of a serious problem. It is likely that only one frame type is being used on this segment, as the second can't function in the presence of this error.
I really can't help much, but I know Novel uses MAC addresses to identify workstations... may be you have two network cards with the same address (difficult but not impossible). In this case you have to change one of them, or cahnge its MAC address. If you're lucky enough, any UNIX lets you change MAC address, under Linux:

#ifconfig eth0 hw ether 01:02:03:04:05:06

hope this helps...

-- Marcelo
otserAuthor Commented:
That only answered half the question.  Although the adminition in response to the second half of the question is appreciated and duely noted, the actual answer is still very important and greatly desired.

Thank You! Very Much, for the answer to the first half.  I'm sure it will be very helpful.
otserAuthor Commented:
Thank You, sauron:

Your answer, combined with other research, lead me to discover that, if edited the file /etc/sysconfig/ipx and changed IPX_AUTO_INTERFACE from 'on' to 'off' and IPX_CONFIGURED from 'no' to 'yes,' then rebooted, the system was able to then figure out how to properly assign the correct network number automatically.  Thereby ending the collision messages.

Thanks, again . . .
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.