[asterisk-users] Nat Issue - is this Draytek || Asterisk?

[Brian]

I'm trying to debug a NAT issue and I can't make up my mind if the
problem is with my Vigor 2800 or Asterisk 1.6.2. I know the Draytek is
alleged to suffer from nat 'issues' but I did not have the issue with
1.6.1 - so I'm wondering if something has changed?

The Draytek offers 'NAT & Routed' on a single device - so my Asterisk
sits on a Public IP, and I have a number of SIP clients on a LAN being
natted. 

If I open a single client on the LAN, it all works as expected. However,
if another machine on the LAN opens a client no client will work.
Attempting to call anything like Voicemail fails and after a short while
Asterisk starts scrolling:

[Feb 10 11:10:31] WARNING[8852]: chan_sip.c:3779 retrans_pkt: Maximum
retries exceeded on transmission
1064dc5c-5101a8c0-13c4-3ba4-e88578-5c0 at 192.168.1.81 for seqno 1
(Critical Response) -- See doc/sip-retransmit.txt.
[Feb 10 11:10:31] WARNING[8852]: chan_sip.c:3779 retrans_pkt: Maximum
retries exceeded on transmission
1064dc5c-5101a8c0-13c4-3ba4-e88578-5c0 at 192.168.1.81 for seqno 1
(Critical Response) -- See doc/sip-retransmit.txt.
[Feb 10 11:10:32] WARNING[8852]: chan_sip.c:3779 retrans_pkt: Maximum
retries exceeded on transmission
1064dc5c-5101a8c0-13c4-3ba4-e88578-5c0 at 192.168.1.81 for seqno 1
(Critical Response) -- See doc/sip-retransmit.txt.

The only way to get service back is kill all other clients on the LAN
and restart the router. Naturally I'm questioning the router, but the
fly in the ointment is that it worked before I upgraded from 1.6.1 to
1.6.2 - which makes me think that it could be Asterisk itself. I'm
starting to wonder if there is an issue in the Asterisk NAT code as I'm
also seeing some 'stale nonce received' relating to the LAN IP of the
second client after I disconnect it.

I'm struggling to work out how can I debug this effectively and would
appreciate some guidance here.