Subject: [netatalk-admins] Multicast errors?
From: Fred Lindberg (lindberg@id.wustl.edu)
Date: Mon Aug 10 1998 - 11:36:44 EDT
Hi:
After recently upgrading to netatalk-1.4b2+asun2.1.0 (redhat 5.1), my
router installation has caused problems on the connected net (with my
router turned off 1 week, problems were gone, turning it on as a client
works fine, turning it on as a router caused the problem to recur
almost immediately).
I would be very grateful or any help in debugging this. At the moment,
I'm running as a client only with plans to go back to the previous
version of the code.
Network:
University network. "Internal Medicine" with cisco routers to the
backbone. A number of "FastPath"s on the "Internal Medicine" network.
My installation on the same network (eth0) and as a seed router on a
private network (eth1 - for IP 192.168. net with IP masquerading) with
netnumber 36035 (not used elsewhere). Starting a ID:AFPServer@ID Lab. 1
HP laserjet and 1 G3 Mac on the internal net. the fastpaths are
configured to use phase 1 on "the others side".
Problem:
A router (seemingly my netatalk installation) passes out incorrect
multicast addresses for the zone "Internal Medicine" (and possibly
others). As a consequence, clients that get the incorrect info cannot
see the nodes in the respective zone.
Here are the error as it is given from a Cisco 7000 running IOS
10.2(10) (one of the seed routers):
%AT-3-MCMISMATCH: Ethernet2/1: Appletalk Port Warning; zone
muilticast
000a.8c01.08e7 differs from our multicast 0900.0700.004c.
An explanation of this error is as follows: A computed multicast
address
disagrees with that provided by another Appletalk router. The other
Appletalk router may be misconfigured or faulty. The recommended
action is
to correct the problem at the other router.
We have also gotten the following as incorrect multicast addresses
(these are not even valid multicast addresses!)
000a.8c01.088a
000a.8c01.08eb
000a.8c01.08f4
360a.8232.0183
0020.8c01.08eb
0020.8c01.08f4
0020.8c01.08f7
0020.8c01.088a
0020.8c01.08e7
The configuration is:
eth0 -phase 2 -net 35841-35850 -addr 35850.28 -zone "Internal Medicine"
-zone "C
ardiology Jewish" -zone "Pulmonary Ethernet" -zone "Internal Medicine
IMBO" -zon
e "Cardiology Lab" -zone "BCC Net" -zone "ID Lab" -zone "Allergy" -zone
"Behavio
r Research" -zone "Pathology CSRB" -zone "Pathology West" -zone "Lab
Medicine"
eth1 -seed -phase 2 -net 36035 -addr 36035.232
The eth0 line was started without zones. They were added during
startup. The zones here match the list of local zones obtained with
getzones.
http://id.wustl.edu/~lindberg/atalkd.debug is a atalk.debug file that
shows the zones etc for our network. It was collected in client mode
(eth0 only) and the address was different. In client mode, getzones
from all routers agree (except that one is missing an unrelated zone).
Some zones have odd characters in their names (ESC, 8-bit set, etc) but
for the checksum calculation a byte is a byte[?].
The log says:
atalkd: zone_bcast Internal Medicine exists
multiple times. Occasionally, the zone mentioned is "Pulmonary" and
occasionally "Allergy".
I can't find the message in the netatalk source package and don't know
what it means.
Thanks!
-Sincerely, Fred
(Frederik Lindberg, Infectious Diseases, WashU, St. Louis, MO, USA)
This archive was generated by hypermail 2b28 : Sat Dec 18 1999 - 16:33:03 EST