[mrtg] missed snmp poll problem?

[Rich Adamson]

Stock mrtg v2.9.18pre1, RedHat Linux

We've been seeing a problem with interface usage statistics when mrtg
misses an snmp poll/response.  

Example: I'm polling a remote Cisco 7206, 2621 and Catalyst 2924. We lost
power to our local router (mrtg running on Linux stayed up) for about
30 seconds plus the reboot time. Mrtg attempted to poll the remote boxes
but didn't get a response.  Approx 5 minutes later (next poll), mrtg did
in fact get a response, however the calculation of Octets In/Out was far
above the maximum speed of the interface (no where near realistic). The
remote devices were not rebooted and the snmp counters were not reset.

The chart reflected a one-time value of say 8,000,000 b/s when the
interface is a T1 (max 1,544,000 b/s). Same thing has happened on various
other interfaces including 10baseT and 100baseT's.

I'm not absolutely sure the issue is simply a missed poll, or maybe a
combination of a missed poll plus an snmp 32-bit counter rollover.

On the surface it would appear that code should be added to compare the
currently calculated bits/second to the MaxBytes value, and if exceeded,
discard the value. The current snmp counter values would still need to be
written to the log file, but not the five-number summary entry.

Anyone have any thoughts on this?

Rich


--
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
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi