I use Nagios to monitor about 30 Cisco 2960 switches (check_snmp_load, check_snmp_int). I only check load and one or two backbone interfaces of every switch.
I get a lot of false positives caused by lack of an answer from devices. I suppose SNMP UDP packets get lost somewhere but I have not been able to troubleshoot it.
sh int summary on every switch and there is no information about dropped packets (IQD nor OQD).
The same Nagios instance monitors dozens of other devices (linux servers, iDRACs/iLOs, SAN switches, tape libraries etc.) and there is always everything alright with those.
The only difference in infrastructure connections between Nagios and 2960s / other devices is that 2960s are connected through quite old FC infrastructure. Is it possible it is a matter of these FC cables? How can I check it (without purchasing expensive network measurements)? I already used iperf but it did not show any problems.
What else can I do to troubleshoot this issue?
Update: My SNMP configuration on 2960s is quite simple:
#sh run | include snmp snmp-server community commstr RO