agent++3.3 SnmpRequest::trap

[Jochen Katz]

Guy.HINDELL____opentrade.co.uk wrote:
> 
> I can see a couple of curious pieces of behaviour with traps sent using
> Agent++. I am using the UCD 'snmptrapd' program as a receiver, and it shows
> an 'Uptime' as part of the trap (both version1 and version2c) which is way
> higher than the actual system uptime (a couple of months), or the uptime of
> the agent (a few seconds). I see this when sending traps from both my
> Solaris and Linux agents.

SNMP++ does the following to set the timestamp:
    gettimeofday(&tp, NULL);
    tp.tv_sec -= 642816000;   // knock off 20years worth of seconds
    timestamp = (tp.tv_sec * 100) + (tp.tv_usec / 10000);
    pdu.set_notify_timestamp( timestamp);

So this value isn't used as "uptime", but it is used as a timestamp
that increases continuously and wraps to zero after 497 days.
I will look into the standards and change this if it's necessary.

Best Regards,
  Jochen