[asterisk-dev] chan_ooh323 issue

[Juan Carlos Castro y Castro]

Greetings. Although this message may not initially look like a -dev 
worthy post, keep reading because I believe it is.

I have the following problem: I have an Asterisk+GnuGK installation, in 
which the brand of H.323 ATAs we use (obscure OEM; I believe the maker 
is Aristel) are unable to receive calls. The phone rings, when I pick it 
up it's all mute and - that's the weird part - the originator keeps on 
hearing ring tones as if nobody had picked up. When the receiving (mute) 
end hangs up, the call is terminated. Calls in the other direction don't 
have this problem.

It's not an obvious ATA bug, neither a GnuGK bug -- two ATAs can call 
each other just fine. More: Asterisk can call an H.323 softphone 
(SJPhone) fine too. The traffic from the latter looks mighty different 
from that from a call to an ATA. Here's links to Ethereal analyses of 
three cases:

Asterisk to SJPhone (OK):
http://img78.imageshack.us/img78/9452/asterisktosoftphoneka1.png
Asterisk to ATA (Rings but doesn't complete):
http://img78.imageshack.us/img78/8003/asterisktoatasq5.png
ATA to ATA (OK):
http://img168.imageshack.us/img168/1830/atatoouratajd6.png

The big difference between Asterisk-to-ATA and ATA-to-ATA is, Asterisk 
(or, rather, chan_ooh323) never sends terminalCapabilitySet or 
masterSlaveDetermination packets. And, of course, there's no RTP coming 
from Asterisk.

All captures were made at the gatekeeper. It has two IP addresses (has 
to because of Linux-HA). Switching to it having only one IP didn't help. 
Yes, I have the .cap files that originated those.

I fear there will have to be some hacking in chan_ooh323 (hence me 
posting here) in order for it to speak properly with the ATAs like other 
ATAs do. If someone would like to test this, I can register my H.323 ATA 
with the gatekeeper you're using, and keep on making and receiving calls 
at your leisure. Plus chatting via MSN or the messenger of your choice, 
or IRC.

If this wasn't -devvy enough, I'll post to -users instead. Should it 
also go to bugs.digium.com?

Regards,
Juan