Hello,
while implementing a monitoring system we've noticed that Catalyst switches occasionally report incorrect ifSpeed and ifHighSpeed, defaulting to 10Mbit/s on 100Mbit/s, 1Gbit/s and 10Gbit/s interfaces.
SNMP query to get the interface speed is run once or twice a day, returned value is then cached and used in bandwidth utilization percentage calculation, which produces the result by an order of magnitude greater than expected ( > 100% ), triggering an alarm. The query has to be manually rerun to [temporarily] fix the alarm.
Subsequent manual queries return correct values. The culprit was identified by monitoring polling activities on the server doing the queries.
Switches and interfaces affected are seemingly random, as cpu and bandwidth utilization the moment the incorrect value is returned hasn't once been even remotely close to 100%.
Models and firmware versions which have been noticed to behave this way include (but are not limited to):
WS-C4948 - Version 12.2(46)SG
WS-C2960G-24TC-L - Version 12.2(44)SE5
WS-C3750G-12S - Version 12.2(55)SE5
Is this a known bug or expected behaviour, what could trigger it and how can it possibly be alleviated?
Thanks in advance.
