I went to the http://www.megacity.org/pong3 page and followed the links to mrtg and
its associated requirements.  I updgraded to the latest perl.

   On a "play" machine, I run the pong3server, and it monitors mail, http, ping, 
telnet,
disk percentage utilization, ssh, cpu load and other services on about 15 machines and
reports the results via a web page every 5 minutes.  The web page is up on a monitor in
the office where the technicians can glance at it.  Pong3 puts a Red background on the
page when an error occurs, and a Green background for situation normal.  this provides 
an
easy glance to know all is ok.  An e-mail is sent to people in a list whenever a 
service
changes from good to bad.  This gives a good log of when things go down.

   I currently monitor total number of users online for the entire network, and for 
each
machine using snmp via mrtg using a perl script to total various input numbers.  It was
easy to set up (however, compiling the packages was a necessary step), and mrtg comes
with some helper programs that allow you to build templates for routers without knowing
anything about snmp.

   I found some references on the net about enabling snmp for portmasters, to be able 
to
check the number of users online.  I have a virtual private network set up for the 
farmed
out dial in connections, and use scripts to monitor users online through that 
connection,
without using snmp.

  The mrtg also graphs in and out traffic densities on my main gateway to the network.
It gives 5 minute interval measurements over 24 hours, 30 minute averages over a week, 
2
hour averages over a month, and daily averages over a  year.

    This helped me find a network congestion problem due to an internal network based
user.  He had been connecting to a news server with no problems for a while.  They
upgraded their equipment, and suddenly became able to saturate my incoming bandwith.  
We
were running full out because he was subscribed to multiple newsgroups, and the 
download
of the headers and other information was lasting for several hours. He was also leaving
the machine online constantly checking for headers.  This is now stopped.

    This package checks every 5 minutes, and doing cpu load measurements indicates the
monitoring program does not affect the network performance or the performance of my 
main
server.

    This package will take about a day to set up, and is well work it.

bug

Doug Elznic wrote:

> On Tue, 7 Apr 1998, Bug Hunter wrote:
>
> >     I'll second that!  It works well, and when combined with mrtg, it allows you to
> > get a historic view of your network traffice density through routers.  You can also
> > get cpu load information and graph it.  If you have dial in users, you can (using
> > mrtg) graph the number of users online at any one time to check for overloads and
> > peaks, assuming your RAS boxes support snmp.
> I am interested in your setup. Can you give me some more info?
> Thanks...




-- 
  PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST ARCHIVES!
http://www.redhat.com/RedHat-FAQ /RedHat-Errata /RedHat-Tips /mailing-lists
         To unsubscribe: mail [EMAIL PROTECTED] with 
                       "unsubscribe" as the Subject.

Reply via email to