I see huge spikes on my graphs for Netapp filer. What is going on ?
Apparently, the Netapp stores counters internally as unsigned integers (Counter32 or just plain Counter in SNMP parlance), but sends them out as signed integers (INTEGER in SNMP parlance). Cricket collects its data using Perl, which (unlike C) will not fix the corruption caused by Netapp`s goof, so it will blithely operate on the large negative values it retrieves and (not surprisingly) come up with wrong values. Solution: add 2^31 to the collected value.