Summary:ASTERISK-01718: iax2/gsm timestamps not consistent at 20ms
Reporter:radamson (radamson)Labels:
Date Opened:2004-05-28 12:33:09Date Closed:2011-06-07 14:10:46
Versions:Frequency of
Environment:Attachments:( 0) iax2-May28.txt
Description:iax2/gsm timestamps are not consistent causing choppy audio with Cisco 7960 phones, using Head cvs from May 25 & 26th. When audio is choppy, transfer call to snom 200, and audio is okay. Ethereal trace indicates 20ms, 21ms, 19ms, 20ms on iax2; rtp indicates 160ms, 168ms, 152ms, 160ms.


This has been an ongoing problem. Ethereal trace is attached. Txt file is frames 662 -> 677 showing the observation as an example. iax2-May28.cap is full ethereal trace with 3,716 packets and several occurances of the timestamp problem and is available if needed. Forcing iLBC codec seems to improve the audio quality.
Comments:By: radamson (radamson) 2004-05-28 16:11:10

The cisco dropped packet problem (irregular rtp timestamps) is still in their sip v7.1 code for the 7940/7960. I understand from others the supera also drops packets due to the timestamp problem, but I don't have one to prove it.

Would it be practical to add an rtp.conf parameter (or somewhere else) that relaxes the iax -> rtp timestamp requirements?

By: Brian West (bkw918) 2004-05-28 22:40:57

what provider are you using?  They will have to update to the lastest as of I think monday on stable.  Also I need just "tethereal -n not port ssh" what you have given me isn't very parsable.

By: Mark Spencer (markster) 2004-05-29 22:44:36

Agreed, this bug screams of bogosity.  We've seen this before and in *every* case it has been that part of the picture isn't updated to the appropriate CVS.

By: Brian West (bkw918) 2004-05-29 22:55:46

Ya that is what I suspect.

By: zoa (zoa) 2004-05-30 18:36:52

Reminder sent to radamson

any updates ?

if you have no idea what version the other end is running or the other end says its using latest cvs try contacting me (zoa) or one of the other bug marshalls and we'll let you connect to a test server to see if you have this with latest cvs.

By: zoa (zoa) 2004-05-30 19:17:58

i spoke to radamson on irc, looks like he has no problems except for this specific termination company.

He's awaiting a bugnote from them, lets be wait for that before we do any further testing.

By: Mark Spencer (markster) 2004-05-31 11:30:48

I'm relabling this a "MINOR" bug since it's tied to just a single service provider and is almost certainly a code versioning issue.

By: Mark Spencer (markster) 2004-06-03 00:41:19

Any more feedback here?

By: radamson (radamson) 2004-06-09 07:12:40

Service provider is not talking, however test calls yesterday indicate they've fixed the iax2 issue with 20ms interpacket timestamps. (They have unrelated issues that appear as though Silence Suppression is in use (or something like that), as 20 to 60 iax2 pkts will frequently be missing, yet the timestamps before-after the missing pkts are very reasonable.) Since they have not contributed to this bug record, we'll have to assume this was a self created problem that was being blamed on * code. I think this bug record can be closed or deleted now. Rich