WWWWWW

Current Directory: /usr/share/doc/mrtg/contrib/nt_n_cisco
Viewing File: /usr/share/doc/mrtg/contrib/nt_n_cisco/README
From: CmdrData <CmdrData@ex-pressnet.com>
Subject: mrtg for nt and cisco

I have been using MRTG running on Windows  NT now for a few months and find
it to be pretty cool.  Thanks so much for making a free tool.

The first time I installed it was on a Proliant 2500 w/ dual 200 Mhz CPUs ,
and the setup was sloppy b/c I was learning how to run the program and make
the config files, programatically.  The setup I had was less than ideal.

So I built a new Proliant 6500 w/ 4 x 200 Mhz CPUs and 3/4 GB RAM and
hardware RAID5 using 6 drives a couple of months later.  I also install
Microsoft SQL server and IIS on the box.  What I now have is an mrtg
database in sql wh/ has 2 tables 1 for NT devices and 1 for Cisco Devices.
The main web page I use has 2 combo boxes which load from these table
devices descriptors which are names and not ips.

The main reason I used sql was so the web page would get auto updated when a
devices was added and also so I could automate the configuration file
creation and the list of devices it choose to poll.  I made a DTS package in
sql which every 2 hours dumps parts of the two tables to text files (the sql
7.0 DTS wizard for this makes it easy to dump parts of a table to a file).
these 2 files are used to update all the configs every 12 hours.  To
accomplish this I had to write a bunch of batch files that would read text
file input and do some very repeatable tasks.  I also set a few environment
variables to make the writing and reading easier.

Since your site was very useful in getting me started I wanted to give a
little back.  I have attached a few nt batch files which would be useful for
anyone trying to set it up on NT.  I still have more automating to do like
making the default.asp page without needing hand editing, and I haven't
automated the ntconfig process.  But right now to add cisco devices  all I
need to do is add a record to a database!  I am using this setup to monitor
about 2700 + ports on mainly catalyst 6500's right now and it finishes its
polling cycle in about 4 minutes.  The machine seems to be cpu and disk
bound.  Quad 450's would be better and a raid with more drive heads would
also be better.  I used the START command with the /BELOWNORMAL switch in a
couple of places to allow the OS to jump in if it needs to b/c the CPU rides
at about 99-100% for most of the 4 minutes.  SQL is not needed to make use
of the attached scripts.  The files just contain 3 items, name (wh/equals
part of the workdir path), comm str and IP [or netbiosname].

I also had to hunt around for the nt info and put it in to one file which
you can do a find and replace on to customize what nt targets you poll.

Feel free to post these files if you like.

What I still have not figured out is how to get the detailed port
description to be used instead of the ifIndexs.  The post you have on this
doesn't work with the port for NT and I don't know Perl.  If you have any
suggestions on this I would appreciate it.  I tried to compare the mrtg file
from the nt port and the hack and the files were too different for me to
splice together without real knowlegde of perl, etc.

Anyhow love the app.  I work for a web hosting company.

thanks,

-Jon
F1l3 Manager
