[mrtg] Re: dead host detection

[Steve Shipway]

> My mrtg setup runs at the ragged edge - a "normal" cycle takes about 4
> minutes to complete, so any dead hosts that are supposed to be polled
> tend to kill the whole thing.  

We get around this in a different way.

Each device has a separate .cfg file, so I have a script which runs and
calls multiple instances of MRTG, spread out over a 5-min window.  This
helps decrease the load on the server, too, and spread out the various
writes.  The script also checks for the lock files existance, and also
for stale lock files.

As a result, a down host will only kill off the instance of MRTG for
that particular .cfg file.  Also, out server can cope with a lot more
targets - over 4000 at the last count on this 2Gb twin Xeon 2.5GHz box
running Linux RHEL3.  Our bottleneck is disk IO, so you might find your
server speeds up considerably if you install a really efficient fast
SCSI card and the fastest disks you can find.

Steve

--
Steve Shipway
ITSS, University of Auckland 
(09) 3737 599 x 86487
s.shipway at auckland.ac.nz

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://lists.ee.ethz.ch/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://lists.ee.ethz.ch/lsg2.cgi