No Such Object implies SE is polling a device for a specific OID and the device is returning NSO. If this is resulting in Contact Lost events I can only fathom that this is SNMPv3 and the switch SNMP settings are blocking visibility to standard sysUptime/sysName/sysLocation OIDs which are our means to 'health check' or 'SNMP Ping' a device.
Taking a tcpdump trace and capturing / decoding the interaction will expose more of what is going on here better than enabling any debugging and having to sift through all the logging it may produce.
