Argus at 1 Month

[Frithiof Jensen (AH/LMD)]

I used it as part of a setup to monitor a wierd network with a mixture
of IPv6 fixed networks, IPv6 ad-hoc networks and an IPv4 ad-hoc network.
About 30 nodes - some with SNMP parameters (GPS position) that I needed
to fetch also. 

The only complaint I had was that it was very hard work to maintain the
configuration manually when nodes and services are yanked around in the
network - as happens in an experimental network. 

One really need an inventory database and a tool to generate an argus
configuration based on that information. At least it is not too hard to
cludge something up with macros and make ... 

Graphing, I think, should be based *entirely* on RRDTool because, IMO,
RRDTool does such a nice job of time series plotting and storage that
there is no reason to go and reinvent the wheel! 

I would also like to see an Argus package for Ubuntu/Debian - It *is* a
bit fiddly to set up Argus as it is now. 

I did a simple Debian package for the above project - it was not
especially hard to do. Mostly a question replacing hard-coded path names
in the build/make scrips with Debian shell vars and getting the Perl
dependencies spelled out as Debian packages.    

  

> -----Original Message-----
> From: arguslist-bounces at tcp4me.com 
> [mailto:arguslist-bounces at tcp4me.com] On Behalf Of Patrick Giagnocavo
> Sent: 8. marts 2007 01:26
> To: arguslist at tcp4me.com
> Subject: Argus at 1 Month

> Areas for Improvement:
> -would like to understand graphing better -would like to see 
> more real-world examples -would like better explanation of 
> SNMP and how to determine the right variables to ask for

SNMP sucks!! 

Sadly, The Way to determine what to look for is to use a "MIB Browser"
on the device you want to monitor and copy the Object ID (OID) for the
data that one needs into the Argus config. For lots of pretty colours in
lots of plots, the simplest way is to add "Cacti" to the management
setup. 

Cacti comes with ready-made templates for SNMP variables assumed worth
collecting and code to work around interface-renumbering and other shite
that SNMP pushes upon the poor management server. 

In the end I used Argus for service monitoring and Cacti for plotting.
This setup worked well.