[Asterisk-Users] Asterisk and Cisco 7960 problems persist (for me, anyway)

[Rich Adamson]

Vic,

The problem you're having has been discussed multiple times on this list,
and can be easily seen using ethereal to inspect the timestamps contained
within the rtp packets sent "to" the 7960 phone. There are several issues
involved, including:
 1. the cisco phones drop any rtp packet that is not exactly 160 milliseconds
    between successive packets (thus causing choppy audio). That drop
    function seems to be the result of cisco changing DSPs in their v6.x
    code. I've not heard of anyone running v5.x sip code with the problem.
 2. iax2 had a bug in it that Mark fixed last month. The bug resulted in
    iax2/gsm timestamps that were erratic when they should have been
    exactly 20 milliseconds between successive packets.
 3. Code was added to rtp.c about a month or so ago that "ties" the
    iax2/gsm timestamps directly to the sip/rtp timestamps. When that code
    was added, it made the iax2 erratic timestamps "and" Cisco's dropping
    of packets extemely obvious to iax2 users. Other non-iax2 users are not
    impacted by this.

Cisco phones seem to be the only ones impacted by this. There are three
short term fixes available to you:
 a. upgrade (or insist your service provider) upgrade their iax2 code. (I
    don't believe the Stable branch has the fix in it as yet.) NuFone
    and some others have done that a few weeks ago.
 b. remove the two or three lines that were added in rtp.c (although
    Mark is discouraging this approach for other reasons), or, go back
    to source code cvs from about early March.
 c. Change the 7960's from v6.x code to v5.x code (and open a trouble
    ticket with Cisco). I've not heard anyone suggest that dropping rtp
    packets with uneven timestamps is necessary, a standard, or anything
    else. Therefore believe it's an anomaly that crept in with the DSP
    change in the sip v6.x code from Cisco.

Rich

------------------------

> On Fri, 7 May 2004, Brian Cuthie wrote:
> 
> > It seems that each time I get a new checkout of * from CVS my Cisco 7960 
> > works worse than before. I know this stuff's in flux, so I mention this 
> > in case it's news.  Anyone else having trouble?  What I'm seeing (er, 
> > hearing) is really choppy audio. The previous version I had installed 
> > had fairly frequent audio dropouts (not present when I make the same 
> > calls through the same * box using a TDM400P interface).
> 
> I had jittery audio with dropouts on a 7960 with SCCP, and started testing
> SIP hoping it would be better (based on the reports of the SIP-to-IAX2
> timestamping issue).  Here's my experience:
> 
> * As Brian mentions, when the other end of the call is from a non-VoIP
> path (e.g. Zaptel interface) the audio is fine.
> 
> * Calls over IAX start out okay, but within a few seconds the audio starts
> jittering.  It gets progressively worse until about a minute into the call
> (often less), by which time audio is unintelligible.  Calling the same
> number over the same IAX connection from an analogue phone attached to a
> SIP-image ATA-186 which in turn is plugged into the "PC" port of the same
> 7960 gives perfect audio.
> 
> * Calls over SIP are stable; I had an intermittent problem where audio
> into the 7960 would stop completely for up to three seconds, but that
> seems to be gone after doing a CVS update.  Side note: when I had this
> audio dropout problem, making the same call without * in the audio path
> (by using canreinvite=yes and removing t and T from Dial) resulted in
> perfect audio.
> 
> I'll try someone's suggestion to disable the jitterbuffer to fix the IAX2 
> problem, but I thought that the jitterbuffer was supposed to help this 
> kind of problem...  Besides, the same call over an ATA or using X-lite is 
> perfect.
> 
> Before anyone jumps in, yes, as soon as I can get there I will hit the bug 
> tracker.
> 
> Cheers,
> Vic Cross
> 
> 
> PS: I know that folks generally dislike 'me too' messages, but this time
> Too Bad -- I'm trying to provide more info to help anyone that might be
> working on problems.
> 
> <rant>
> I hope that Iain was exaggerating when he described his bug-reporting
> experience.  Many * users are unable to commit the time to poring over
> hundreds of lines of uncommented C code and ethereal traces with thousands
> of packets captured.  So, as our way of trying to help, we provide e-mails
> like this either in response to or as an attempt to gather more
> information about the problem.  To try and get people talking about a 
> problem.
> 
> How is does it help to jump on someone who is trying to get resolution to 
> a problem -- by driving them toward OpenPBX or VOCAL?  A few former 
> colleagues of mine may soon be about to learn (unfortunately) that you can 
> only piss off a customer so many times.
> 
> To the Asterisk developers, bug marshals, and coders: I am jealous of you!  
> You've created a wonderful thing.  I'd love to be able to spend the amount
> of time I'd like to on Asterisk.  I'd love to be able to do more to fix
> bugs and develop features.  But I can't.  Don't think less of me because
> of that. 
> </rant>
> 
> VC
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users

---------------End of Original Message-----------------