i can do some snmp queries to the non-working hosts but not all (pool/vs do not work while cpu memory etc do). for example, here is an snmpwalk to the "working" machine i listed above:
snmpwalk -c public -v2c f5loadbalancer1.com .1.3.6.1.4.1.3375.2.2.5.2.3.1.1
SNMPv2-SMI::enterprises.3375.2.2.5.2.3.1.1.8.110.105.107.95.116.101.115.116 = STRING: "XXXXX"
SNMPv2-SMI::enterprises.3375.2.2.5.2.3.1.1.11.119.119.119.46.102.120.99.109.46.100.101 = STRING: "XXXXX"
SNMPv2-SMI::enterprises.3375.2.2.5.2.3.1.1.11.119.119.119.46.102.120.99.109.46.105.116 = STRING: "XXXXX"
and here is the "non-working" machine:
snmpwalk -c public -v2c f5loadbalancer2.com .1.3.6.1.4.1.3375.2.2.5.2.3.1.1
SNMPv2-SMI::enterprises.3375.2.2.5.2.3.1.1.8.110.105.107.95.116.101.115.116 = STRING: "XXXXX"
SNMPv2-SMI::enterprises.3375.2.2.5.2.3.1.1.11.119.119.119.46.102.120.99.109.46.100.101 = STRING: "XXXXX"
SNMPv2-SMI::enterprises.3375.2.2.5.2.3.1.1.11.119.119.119.46.102.120.99.109.46.105.116 = STRING: "XXXXX"
as you can see they're exactly the same. my cacti server can reach both of them just fine from the command line however through cacti i can only reach *some* of the oids. i'm going to do some packet captures later but i have double-checked my cacti settings about two-dozen times :]