[mrtg] Re: MRTG processor load

[Thomas Brian Granier]

There's gotta be something wrong then. Here's my setup:

Remote Lan I/O Graphs: 47
Remote Wan I/O Graphs: 52
Local Wan I/O Graphs: 46
Local Lan I/O Graphs: 50
Ping graph to remote Lan: 49
Total Bandwidth Graphs: 47

I do this on an NT workstation machine with a P III 450 and 128 MB of
RAM. I use the fiveminute.zip to run batch files every five minutes that
gather the MRTG graphs. I broke them up into 4 batch files to run 4 mrtg
sessions concurrently. With this setup my batch files finish in three
minutes under normal conditions. This machine is also running What's Up
Gold that sends out ping requests every minute. During the 3 minutes
that MRTG is gathering and graphing data, the CPU utilization on my
computer fluctuates between 95 percent and 100 percent utilization.

If your config files don't change very often I would recommend running
it in the Daemon mode. This eats up a lot less processor utilization.
Also, if it is possible to put all your configurations in one file that
will save MRTG from competing with itself for system resources as it
will only begin graphing the next item in the configuration file after
it finishes the current. This is different from running each config file
simultaneously which, I have discovered, eats up system resources way
too much. In my setup I have about 50 different config files broken up
into the 4 batch files. Each config file in the individual batch files
do not run into the previous one has finished.

Is the system running MRTG connected to your LAN via a hub or a switch?
It's possible that you're ending up creating a lot of collisions on your
Ethernet that are taking too long to resolve and therefore slowing down
your system unbearably. I have my system switched. Alone it creates no
more than 4 percent utilization of it's localized available bandwidth.

T. Brian Granier
MCSE MCP+I A+
Telecommunications Specialist
Zebec Data Systems

-----Original Message-----
From: mrtg-bounce at list.ee.ethz.ch [mailto:mrtg-bounce at list.ee.ethz.ch]On
Behalf Of Csaba Kovacs
Sent: Monday, May 15, 2000 9:06 AM
To: mrtg mailing list
Subject: [mrtg] MRTG processor load


Hi!

Coul anyone please tell me how much processing time is involved in
generating
the MRTG graphs? Do people in general find it to be fast enough? How
many
devices are you generating graphs for every 5 minutes?

I'm running the 5-minute snmp polling for around 100 devices, and the
machine
slows down enourmously. The polling doesn't even finish in time for the
next
batch of 5-minute graphs to be generated. At the moment there is a delay
in
the graphs of up to 2 hrs. The 'last updated' may say 14:00, but the
graph
only shows values until 12:00. 

I'm running it in Solaris, on a Sun Ultra 80, with 1GB of RAM. 

Any ideas how to make it more efficient?

Regards,
Csaba

____________________________________________________________________
Get free email and a permanent address at http://www.netaddress.com/?N=1

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=subscribe
Archive     http://www.ee.ethz.ch/~slist/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org


--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://www.ee.ethz.ch/~slist/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org