Some devices use a 32bit integer for uptime which can cause counter wrapping (after roughly after 497 days).
As a result these devices (when they've been running for long enough) can provide a false reading to NMIS (indeed any SNMP poller).
We can rectify this by editing the model being used for the device and altering the attribute collected for updateuptime.
By default we retrieve sysUpTime from 1.3.6.1.2.1.1.3.
...