Laurent,
This is certainly true. For these reasons, devices behind NATs should really do H.245 tunneling to avoid the problem. Alternatively, the NAT ALG handling the H.323 signaling could just discard the H.245 address received from, say, the calling party. In that case, only one side can initiate the H.245 connection. Discarding the called side's H.245 address is probably dangerous.
Anyway, these are just ideas... not necessarily the best solution. Certainly, this would be a good topic for discussion in Q5/16. (Copying the SG16 list.)
Paul
-----Original Message----- From: Laurent Barbero [mailto:barbero@col.bsf.alcatel.fr] Sent: Thursday, September 30, 2004 5:06 AM To: imtch323implementors@mail.imtc.org Subject: [imtch323implementors] conflict resolution in simultaneous H245 connections
Hi all, H323 recommandation says in case of separate H245 connections conflicts that the GW/Enpoint with the numerically smaller h245address shall close the TCP connection it opened to keep only one control channel.
If you have both ends A and B, each in a private network and H323 NAT in the middle, and if the address translation gives :
private H245address of A < translated H245address of B private H245address of B < translated H245address of A
Both H245connections are closed.
Is there a common implementation workaround except control policy of the H323 address translation in the NAT to avoid such an issue.
Thanks in advance. Best regards. L.B
Check out the IMTC Current News and Events! http://www.imtc.org/press/press_releases.asp