Previous to a reboot (or when all processors are started at the same time), 1 processor will send SNMP v3 traps to another processors (sender to receiver node). After reboot, the senders traps are being dropped by snmptrapd for "usm: Message too old.". We believe this is happening because persistence is not being enabled on the sender side using the command snmptrap.

We believe this may tie back to a bug found in the snmp library.
please see the following links:

https://github.com/net-snmp/net-snmp/issues/827

https://github.com/net-snmp/net-snmp/pull/510