nv-l
[Top] [All Lists]

Re: FW: Some Questions about network device 's trap

To: nv-l@lists.tivoli.com
Subject: Re: FW: Some Questions about network device 's trap
From: Chris Cowan <chris.cowan@2ND-WAVE.COM>
Date: Thu, 28 Oct 1999 20:50:23 -0500
James Shanks wrote:

>
> 2. Attended MLM is something I know nothing about.  In other situations, as I
> explained, NetView does not forward traps to MLMs at all, but they forward to
> him.   I suspect that attended MLMs may be different, because in NT, trapd
> cannot forward events to other boxes.  The trapfrwd daemon does that.  So it
> would make sense that an attended MLM (which is netmon and midmand working
> together in the same NT box) would use MLM's trap forwarding capability.  But 
> I
> don't know how it works nor do I know if what you are seeing is normal or not.
>
> Perhaps someone else with experience with attended MLMs can answer your 
> question
> if you put it in those terms
>

Since I just did this, here goes.

First, there's an html doc in \usr\ov\doc that contains information
that's not in any of the Netview Manuals.   He needs to read it.

Stuff you need to know:

1. You have to configure trapfrwd (this means doing the ovobjadd, and
setting up the trapfrwd.conf file).  For the MLM, you send it to
localhost port 1662 (which is what the MLM uses when installed on top of
Netview).  BTW, trapfrwd could easily bypass the MLM and go to any other
system capable of receiving an SNMP trap.  It also has the ability to
filter traps based on Enterprise ID.

2. By default, the MLM has trap reception DISABLED.   You have to use
either the APM, smconfig, or do an   SNMP set and turn reception on in
the Trap Reception Table (there's separate switches for UDP and TCP
traps, also!).  You can use the snmptrap command to fire traps at the
MLM and Netview port to test that everything is truly enabled (there's a
port switch on the command).   You can also use the event command to
test the Netview piece (event has a switch for firing to another Netview
manager, it -h or -H I believe).

The traps he's seeing are the coming from the MLM status engine, not
Netmon, BTW.   It's also not clear but depending on how the environment
variables are set, you may have both the MLM and netmon doing redundant
status polls of nodes.   The MLM status/interface monitoring tables can
be totally decoupled from the underlying Netview Databases.  (Those
details and more are in the HTML doc).

Attachment: chris.cowan.vcf
Description: Card for Chris Cowan

<Prev in Thread] Current Thread [Next in Thread>

Archive operated by Skills 1st Ltd

See also: The NetView Web