Open GoogleCodeExporter opened 9 years ago
This problem is actually much more severe when the DNP3 stack is being used as
a slave instead of as a master. If unsolicited mode has not been requested,
when the cat5 cable is yanked the client will not be sending data and therefore
will never fill up the TCP send buffer. Thus, yanking the cat5 cable can
effectively cause a slave to hang forever, not accepting new requests.
Original comment by emeze...@gmail.com
on 9 Dec 2010 at 9:17
SunEdison has provided a mechanism to set the TCP keepalive on a per socket
basis and this change has been verified and will be merged pending the CLA.
It would be nice to have someone look into the equivalent windows functionality.
Original comment by jadamcrain
on 14 Dec 2010 at 3:14
I'm only vaguely familiar with DNP3, but why not just use the
"REQUEST_LINK_STATUS" (Function Code 9) per the recommendations of the DNP3
Spec? (Ref: DNP3 Spec, Volume 4 Data Link Layer, V2.01 Section 2.6.5)
Original comment by Master...@gmail.com
on 14 Jul 2011 at 3:40
Original issue reported on code.google.com by
jadamcrain
on 13 Jul 2010 at 8:02