[mrtg] mrtg performance

[Calvert, Neil]

Hi folks

Question for the floor. I'm monitoring one particular site that has a router
with about 80 frame relay links on it. The link to this site is bad so I
have MRTG doing a 5 retry and 10 second timeout for each SNMP query.

The problem I face is it seems like every time we lose contact to this
router for an extended period, the MRTG queries continue and build up,
queueing to run, querying and timeing out for the set period, creating more
and more processes and killing the box CPU in the process. The CRON jobs run
every 10 minutes so it seems the previous run isn't finished before the next
starts, and so on.

If I change the retry and timeout, then the quality of the link means a lot
of MRTG polls won't go through. (We average 25 percent packetloss and 400 ms
delay on a good day). 

Is anyone out there facing the same issue? Are there any nice elegant ways
to verify a box connectivity prior to firing off the report run? What I'd
like is a brief say 5 second query if the box doesn't respond (even on a bad
day it will usually respond within 5 seconds) then the report run is
cancelled until the next 10 minute interval.

Any suggestions welcomed...

Neil

--
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