Eric, I would take exception to the statement that 1.1.1.* and 1.1.1.1-20
behave differently in a
seedfile. I can assure you that with Netview V4, a seedfile containing an
explicit list of nodes and
either of those ranges will discover only those explicit nodes, the management
station, subnets
for all interfaces on the requested devices (empty except for the requested
devices), and any
devices which actually have addresses in the range 1.1.1.* or 1.1.1.1-20.
Furthermore, the nodes
that DO match the ranges will have to make noise on their own, or be pinged to
be discovered.
Netview will not go out looking for them. It WILL go looking for the explicit
entries.
As James says, your best bet is to flush the databases. I like the option under
'Control...Restart
Automatic Map Generation' as it saves your Collection definitions, etc.
Changing the seedfile
will not get rid of things already in the database.
Since you already have a giant discovery, here's a tactic I have been using
lately: Ask the
giant database who you should keep:
/usr/OV/bin/ovtopodump -rl |
awk '/^HOSTNAME:/ {hostname=$2}
/^SNMP OBJECT ID:/ {oid=$4}
/^SUPPORTS SNMP:.*YES/ {print oid " : " hostname}' > /tmp/byoids.list
This gives you a list of all of the snmp-supported nodes you have discovered.
Sort it, and you
can go through it and identify routers, hubs, printers, servers, etc. Delete
the kinds of devices
you don't want. Move the routers to the top of the list, preferrably ordered
from near to far.
Awk it again to extract just the 3rd column and you have an explicit seedfile
of keepers.
cat /tmp/byoids.list |
awk '{ FS = " "}; {print $3}' > /tmp/seedfile.new
Add the limiting range and you have your complete, restrictive, explicit
seedfile.
Flush the database and rediscover.
Cordially,
Leslie Clark
IBM Global Services - Network & Systems Management - Detroit
______________________________________________________________________________________________________
James,
I use a dummy range to stop discovery but I thought the dummy range to use
was something like this :
1.1.1.10-20
I first tried to use the one you suggest and I had the followig result :
Netmon discovered nodes in this range and all the other nodes.
Later on, I was told at a IBM training session that :
1.1.1.* : netmon would discover at least all the nodes in the range
1.1.1.10-12 : netmon would discover at most all the nodes in the range
What's the point ? I'm running Netview 4.1 on AIX 4.1.4.
regards,
Eric.
-------------------------------------------------------------
Eric Filippi (efilippi@eurhone.e-mail.com)
Euhone
IBM Global Service Ouest
Lyon
France
-------------------------------------------------------------
James_Shanks@TIVOLI.COM on 15/06/98 22:04:04
Veuillez r?pondre ? NV-L@UCSBVM.ucsb.edu
Pour : NV-L@UCSBVM.ucsb.edu
cc : (ccc : Eric FILIPPI/VAISE/EURHONE/FR)
Objet : Re: I need to delete about 18,000 symbols
Content-type: text/plain; charset=us-ascii
Does your seed file also have a dummy range entry to halt discovery,
something like this?
1.1.1.*
I'll bet not .
The easiest thing to do is to define exactly the nodes you want in the seed
file, plus a dummy range like that to halt discovery, and then stop all the
daemon and delete your databases through SMIT. There is no easy way to
delete 17,000 nodes (there's no easy way to delete 100 nodes). When you
restart netmon he should discover just what's in your seed file an no more,
except for the connections and network symbols required to hold them.
You can read more about seed files in the dynatext books which come with
the product. The install manual has some especially good stuff.
James Shanks
Tivoli (NetView for UNIX) L3 Support
redder@YUMA.ACNS.COLOSTATE.EDU on 06/15/98 12:47:36 PM
Please respond to NV-L@UCSBVM.UCSB.EDU
To: NV-L@UCSBVM.UCSB.EDU
cc: (bcc: James Shanks)
Subject: I need to delete about 18,000 symbols
I'm running NetView 4.1 on AIX 4.2.1
It seemed neat at first that NetView could detect all 18,000 machines on
our campus, but our NetView machine doesn't really have the resources to
handle that large of a database. We have about 18,000 pieces of equipment
on this campus, and about 20 routers we manage. The seedfile has these
20 routers in it.
My dilemma is that I need to cut our object database down to the objects
I really care about - the 20 routers and their 120 interfaces plus
a scattering of "critical" servers - about 20. I'm not sure what the
best way is to trim out 17,000+ items from the database.
My questions are:
If I start over, NetView will discover all of the routers in the
seedfile and all of the machines on each of the interfaces and I'm
back to the problem I have now.
Maybe I have an empty seedfile and add routers manually? However,
if I add the routers I'm afraid that as soon as I manage the interface,
it will go out and discover all of the nodes on that network and populate
my database unneccessarily.
Ideally, I want the database to just contain the routers, their interfaces
and the network symbol associated with those interfaces, but none of the
machines in those networks. It's all of the machines on our network that
is making the database so large and slowing everything down.
What is the easiest way to get to that ideal situation from where
I'm at?
Thank you --Greg Redder
Network Analyst
Colorado State University
=============================================================================Greg
Redder Academic Computing & Networking
Services
Colorado State University, ACNS Phone:(970)491-7222 FAX: (970)
491-1958
601 S. Howes, Room 625 E-mail: redder@yuma.colostate.edu
Fort Collins, CO 80523 PGP
Fprint:68CEE78C86AC452881B27249785FEE91
=============================================================================
|