We also use a seedfile listing our routers, switches and other critical
devices, and face the same issues of rapid growth, many
moves/adds/changes/deletes, and delayed or missing notification of such
changes. I too am interested in learning how others are dealing with this.
I've been contemplating a way to automate the seedfile updates, but have not
had a chance to test it yet. My plan is to set up a rule to detect the
Node Add and Node Delete traps issued by netmon (IBM_NVADD and IBM_NVDEL),
check if they are the type of device I would include in the seedfile by
querying a database field or collection, and call a script to edit the
seedfile accordingly. For routers, the scripts could do further processing
to determine if a loopback address is configured, if other interfaces have a
name associated with the device, etc., and add the appropriate name or
address to the seedfile. It would be REALLY nice to be able to be able to
delete the device via CLI and clean up the map too.... I need to do some
testing to determine what info is available in the database and how
up-to-date the collections are at the time of trap receipt.
Thoughts / comments? Tricks to do the delete from the map that I haven't
heard about?
Re: the nearest interface on the router being used to issue the trap. This
gets particularly messy if there are multiple nearest interfaces (i.e. for
redundancy). On Cisco routers, and perhaps others, the interface used as
the trap source can be configured on the router. The syntax for Cisco is:
snmp-server trap-source <interface-name>
Since we don't have loopback addresses configured on most of our routers
:-( , I'm careful to specify the interface name that has the IP address that
is set up in DNS for reverse lookups.
Regards,
Karin
-----Original Message-----
From: lclark@us.ibm.com <lclark@us.ibm.com>
To: NV-L@tkg.com <NV-L@tkg.com>
Date: Wednesday, March 29, 2000 7:51 PM
Subject: Re: [NV-L] Netview Seed File with oid's
>
>
>I think this is a fairly common issue, the trade-off between control
>and the adminstrative burden it entails.
>
>If I were letting routers be discovered by whatever interface is
>found, I would just add matching names to the hosts file for those
>interfaces,
>without bothering with the delete/rediscover. I would override the DNS
>with /etc/hosts entries for those interfaces (on AIX this is done in
>/etc/netsvc.conf with hosts=local,bind, Solaris does it differently).
>I might set up a collection (smartset) that contains things without name
>resolution, and, if needed, things that have goofy interface-oriented
>names that need changing. Treat that as an after-discovery to-do list.
>
>I have found it to be very useful to have consistent name resolution
>not only on the loopback, but also on the nearest interface, since
>that is usually the one that the traps come in from. It allows me to
>correlate events with objects on the map, among other things.
>
>Anybody else?
>
>Cordially,
>
>Leslie A. Clark
>IBM Global Services - Systems Mgmt & Networking
>Detroit
>
>---------------------- Forwarded by Leslie Clark/Southfield/IBM on
>03/29/2000 08:30 PM ---------------------------
>
>"Jewan, P. (Pritesh)" <PriteshJ@nedcor.com>@tkg.com on 03/29/2000 06:59:50
>PM
>
>Please respond to IBM NetView Discussion <nv-l@tkg.com>
>
>Sent by: owner-nv-l@tkg.com
>
>
>To: "'nv-l@tkg.com'" <nv-l@tkg.com>
>cc:
>Subject: [NV-L] Netview Seed File with oid's
>
>
>
>
>
>Hello People,
>
>We currently have a network with about 550 Cisco routers and 250 Cisco
>switches as well a large number of other servers(NT,AIX,SUN etc). The way
>we have been doing the discovery in the past was to put the IP addresses of
>the device into the seed file and let netmon discover the nodes. This
>method was important to us because we wanted to discover the routers using
>their loopback addresses as the loopback addresses are tied to names in the
>DNS server. However our environment is growing at a rapid pace and as a
>result new routers go out in the network and their addresses are sometimes
>not reported and as a result are never put into the seedfile.
>
>We are currently investigating other methods of creating a seedfile to
>alleviate this problem. The only suitable method I could think of was to
>use Cisco OID's for the Cisco devices but then this leaves us with the
>problem of the routers been discovered with the first address that netmon
>finds. The only thing this helps us with is that we now know that there is
>a router out there and can then put it's loopback address into the
>seedfile, delete the object and let netmon rediscover it bye the loopback
>address.
>
>Has anyone encountered a similar situation? Any ideas/suggestion would be
>greatly appreciated.
>
>Regards
>Pritesh Jewan
>ESM - Technology & Operations Division
>Nedcor Bank Limited (South Africa)
>
>Tel : +27 - 011 - 320 5417
>Cell: +27 - 82 570 5046
>Fax : +27 - 011 - 8814743
>e-mail : priteshj@nedcor.com
>
>
>
>
>_________________________________________________________________________
>
>NV-L List information (unsubscribing, policies, posting, digest version,
>searchable archives): http://www.tkg.com/nv-l
>
|