On Tuesday 19 January 2016 23:51:22 Youngman Jung wrote:
> I think we need the method for handling the situation or the restrictions
> for it.

Sending packets every 10 ms is a bad idea. Especially if it's a unicast 
observation notification, because there may be other clients that need to be 
notified. This will seriously degrade network support.

The test should be done at a more reasonable interval of a second or more.

Regardless, timeouts should occur only after several seconds of non-response 
and the timer is reset the moment any response is obtained.
-- 
Thiago Macieira - thiago.macieira (AT) intel.com
  Software Architect - Intel Open Source Technology Center

Reply via email to