RE: [Mon-devel] Re: Announcing ospf.monitor (beta-1 :-)

[Peter W. \(MO/EMW\) <pet...@er...>]

> -----Original Message-----
> From: Ed Ravin [mailto:er...@pa...]
> Sent: den 6 februari 2006 19:44
> To: Jim Trocki
> Cc: Ed Ravin; Peter Wirdemo (MO/EMW); mon...@li...
> Subject: Re: [Mon-devel] Re: Announcing ospf.monitor (beta-1 :-)

The SNMP modules requires the Net-SNMP package to be installed, which is =
not required for MON.
From the README in SNMP-4.2.0 (which i got using CPAN)

   SNMP module version 4.2.0 is being developed against NET-SNMP-4.2.0
   see http://sourceforge.net/projects/net-snmp for details.
  =20
   Compatibility with earlier or later versions of Net-SNMP or UCD-SNMP
   is not guaranteed due to the dynamic nature of open software
   development :).

Net-SNMP package is now 5.3.0.1 !

SNMP
The Perl5 'SNMP' Extension Module v3.1.0 for the UCD SNMPv3 Library
SNMP-4.2.0 - 12 Feb 2001 - Joe Marzot=20

Net::SNMP
Object oriented interface to SNMP
Net-SNMP-5.2.0 - 20 Oct 2005 - David M. Town=20

I dont know about stability, but I've nerver had problems using =
Net::SNMP

>=20
>=20
> On Mon, Feb 06, 2006 at 01:29:20PM -0500, Jim Trocki wrote:
> > The "heavy" SNMP includes the ability to parse MIBs, for one thing.
>=20
> Which I deliberately didn't use to keep the monitor a bit simpler,
> since I only had a handful of OIDs to fetch.

I think most monitors are like this, get a few OIDS, not the whole MIB

>=20
> > It also has  asynchronous mode operation, which is very useful
> > for when you need to gather a bunch of tables from a number of
> > different hosts.
>=20

Net::SNMP supports blocking and non-blocking mode!

> Now that's a good reason for using it - especially since one potential
> user of this script said he had 300 routers to poll.  In my network
> it doesn't matter since (a) I have nowhere near that many routers and
> (b) for various other reasons, I make each router a separate entry in
> Mon.
>=20
> > Long ago I embarked
> > on designing an OSPF monitor, and I realized that for it to=20
> be efficient I
> > needed to take advantage of this asynch. capability, and I=20
> wrote some
> > example code to rapidly fetch tables from dozens of snmp=20
> agents in parallel.
> > I'll post that code here so that maybe someone could=20
> integrate it into
> > Ed's OSPF monitor.
>=20
> It would be nice to have a sample monitor that had working async SNMP
> fetching, since many other monitors would benefit from it.  I've had
> trouble with some of my SNMP tests - they might work better if I gave
> them a longer timeout, but the serialization of the timeouts means the
> monitor might take longer than its 5 minute polling interval if enough
> hosts timed out.
>=20
> 	-- Ed
>=20