Here comes an opinion based on observing the behavior of netview discovery
over the span of several years and releases. Just my opinion.
It used to be that if you had different names configured in your DNS for
different interfaces, the worst that would happen might be that the name
displayed on the map would change. It did not get 'discovered' as a
new node. It appears to me that in recent releases (after 5.0?),increased
reliance is being placed on names, even in the absence of SNMP. For
instance, I can get Netview to create a 'router' on the top level of the map
by putting 4 PC addresses in the /etc/hosts file with the same hostname.
Netview will assume that they are addresses on the same device until it figures
out otherwise. Likewise, different names on different interfaces can trick it
into
thinking they are different devices, then it will complain about duplicate
IP addresses on these two devices when it gets around to checking further.
I think what you are seeing is a result of this increased reliance on name
resolution as an aid to discovery.
You can circumvent your problem by providing name resolution for
your routers this way: all of the addresses on the device resolve to the
same name, and the name always resolves to a single address (loopback
or something). Then you won't have this problem. If your corporate standard
for name resolution requires the arrangement you have now, and you are
using a DNS, then you can override that arrangement just for Netview by putting
the desired entries in /etc/hosts and setting /etc/netsvc.conf to
'hosts=local,bind'.
The desired entries in /etc/hosts would be a list of all of the addresses on a
device along with the exact same name on each record. The name will always
resolve to the first entry, and the addresses will always resolve to the
same name.
If you don't like the way names are being handled, you might want to
call Support and see if they really intended it to work that way. As far
as I know, there is nothing improper about providing different names for
different interfaces on a device, but Netview has grown to dislike it
more and more.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
---------------------------
How can you prevent the following behavior in NetView?
In our NetView configuration we have noticed that devices are sometimes
rediscovered (by autodiscovery) using the nodename of one of the serial
interface nodenames for that device. For example, a router in Los Angeles
(known by "labcn1") may be rediscovered and labeled as "labcn1dcbcn1". As a
result, the device already in the NetView database (which is associated with the
correct nodename, "labcn1") is removed and replaced with the "newly discovered"
device. This new device is placed in the new node area at the bottom of the IP
Internet submap. Basically, some of our routers are being renamed with their
serial interface nodenames by NetView.
More importantly how can this be fixed?
(We are using NetView ver 4.1 on AIX.)
Thanks in advance!
|