[mrtg-developers] Re: Deploying MRTG
Mohit.Tendolkar at thomson.net
Thu Dec 2 01:07:13 MET 2004
We are trying to deploy a monitoring solution using MRTG running on a
computer running Windows as follows:
* we have an MRTG config file for every device that we are
monitoring that contains the MRTG targets for the objects we are
monitoring on that specific device.
* We run each such config file with a separate instance of
perl/MRTG in daemon mode
We used this since it was the recommended approach to running/deploying
Since we monitor about 40+ devices, that implies that we get typically
about 40+ instances of the perl process (one can see it on the "Process"
tab of Windows task manager). Similarly when these instances of
perl/mrtg start the graphing process, they start another "cmd.exe" and
"rateup.exe" process each, resulting in those many more instances of
cmd.exe and rateup.exe.
This causes the system to run out on virtual memory and many times even
needs a reboot, since the instances of perl take up quite a bit of
virtual memory each.
We have also experimented with not running MRTG in daemon mode and
periodically running MRTG for each of the config files. What we observed
was that each time MRTG is run in this manner, since it parses the
config file each time, it takes up a bunch of CPU and the process is
quite resource intensive.
I see a lot of users/developers out there using MRTG, So, could other
MRTG developers out there let me know how they have deployed monitoring
solutions based on MRTG for a large number of systems from a Windows
Thanks & regards
Unsubscribe mailto:mrtg-developers-request at list.ee.ethz.ch?subject=unsubscribe
Help mailto:mrtg-developers-request at list.ee.ethz.ch?subject=help
More information about the mrtg-developers