Re: [M3UA] Action on Receipt of ERROR

"Sporton, Simon, VF-Group" <[email protected]> Thu, 17 Mar 2011 21:54:58 +0100
Newsgroups gmane.ietf.sigtran
Message-ID <957B5237851DB74283AFD29B017683D007384864@EITO-MBX01.internal.vodafone.com>
Hi David,

In our case the SGP included diagnostic information in it ERROR, so the
ASP could have determined that the problem was related to a DAUD it had
previously sent, but unfortunately the ASP doesn't look at diagnostic
info! 

To be honest, your pragmatic solution would probably be good enough...
it's the logic that each node needs to determine there is a miss-match
in the state machines and re-connect to get them in sync again.

Simon

-----Original Message-----
From: David Laight [mailto:[email protected]] 
Sent: 17 March 2011 10:21
To: Sporton, Simon, VF-Group; [email protected]
Subject: RE: [Sigtran] [M3UA] Action on Receipt of ERROR

 
> We have been looking at various vendors implementation of M3UA and 
> seen a significant divergence in the handling of M3UA ERROR 
> messages.....

The problem I had with the M3UA MGMT ERROR messages is that they are
completely 'out of band' with any requests.
We have code to detect 'Mangement blocking' as a reponse to an activate
request (an explicit negative ack to the activate would be much better),
but otherwise assume the MGMT errors are likely to be the result of M3UA
protocol mismatches. However we can't be certain of this so take no
immediate action.
If the remote system has rejected an important message then a guard
timer expires and we disconnect - since the far end clearly is very
confused - then retry later.

This may be a rather pragmatic solution....

	David