Re: [Asterisk-Users] Re: Audio flutter on OH323 output?

2005-05-23 Thread Michael Manousos 

Tony Mountifield wrote:

In article [EMAIL PROTECTED], I wrote:


In article [EMAIL PROTECTED],
Michael Manousos [EMAIL PROTECTED] wrote:


Can you get an ethereal trace on a call with that problem?
Run an RTP analysis on the captured stream (Tools Menu) and save
the contents of the RTP packets in audio files. Then check if
the playback of these files is normal or not.


Aha, sounds like Ethereal has even more clever features - I didn't
know it could do that.

Fortunately, I've been running a continuous tcpdump capture on the
Asterisk box of the traffic between it and the switch. I'll see
what Ethereal thinks of it.





Hi Tony,



Hi Michael,

I've now done some RTP analysis of affected streams, and have found that
the times when we get the audio flutter correspond with parts of the
analysis showing repeated packets in the RTP stream. This only happens
on the outgoing stream; the incoming stream from the switch is perfect.

When a packet is repeated, it is identical to the previous and occurs
about 7 to 14 microsec after, as if something in the software has decided
to send the packet twice in immediate succession.



This seems as a bug in OpenH323, but also as a bug in the switch's RTP
implementation. Normally, the remote RTP stack should reject the
duplicate packet. Of course this does not cancel the need to fix the
bug in OpenH323. I'll prepare a version which compiles with the latest
code of OpenH323/Pwlib. You can try it and see if the problem is still
there.



I've pasted below an extract from the CVS file saved from the RTP
stream analysis. I also saved the payload (pity it converts it to uLaw
instead of saving it in the stream's native format) and listened to it
to confirm the periods of distortion.

I would assume the problem is somewhere in the depths of openh323, but
any pointers in the right direction would be appreciated!

Cheers
Tony



Regards,
Michael.




Here is the CSV extract:

Packet,Sequence,Delay (s),Jitter (s),Marker,Status,Date,Length
144209,17353,0.02,0.03,,[ Ok ],05/17/2005 17:05:48.420,214
144225,17354,0.020006,0.03,,[ Ok ],05/17/2005 17:05:48.440,214
144226,17354,0.07,0.03,,Wrong sequence nr.,05/17/2005 17:05:48.440,214
144244,17355,0.019987,0.04,,[ Ok ],05/17/2005 17:05:48.460,214
144260,17356,0.019998,0.04,,[ Ok ],05/17/2005 17:05:48.480,214
144261,17356,0.08,0.04,,Wrong sequence nr.,05/17/2005 17:05:48.480,214
144276,17357,0.019991,0.04,,[ Ok ],05/17/2005 17:05:48.500,214
144296,17358,0.019992,0.04,,[ Ok ],05/17/2005 17:05:48.520,214
144298,17358,0.09,0.05,,Wrong sequence nr.,05/17/2005 17:05:48.520,214
144313,17359,0.019990,0.05,,[ Ok ],05/17/2005 17:05:48.540,214
144329,17360,0.02,0.05,,[ Ok ],05/17/2005 17:05:48.560,214
144349,17361,0.019992,0.05,,[ Ok ],05/17/2005 17:05:48.580,214
144368,17362,0.020003,0.05,,[ Ok ],05/17/2005 17:05:48.600,214
144369,17362,0.06,0.05,,Wrong sequence nr.,05/17/2005 17:05:48.600,214
144383,17363,0.019990,0.05,,[ Ok ],05/17/2005 17:05:48.620,214
144401,17364,0.019995,0.05,,[ Ok ],05/17/2005 17:05:48.640,214
144417,17365,0.020001,0.05,,[ Ok ],05/17/2005 17:05:48.660,214
144432,17366,0.019994,0.05,,[ Ok ],05/17/2005 17:05:48.680,214
144452,17367,0.019997,0.05,,[ Ok ],05/17/2005 17:05:48.700,214
144469,17368,0.020002,0.05,,[ Ok ],05/17/2005 17:05:48.720,214
144486,17369,0.019998,0.05,,[ Ok ],05/17/2005 17:05:48.740,214
144487,17369,0.08,0.05,,Wrong sequence nr.,05/17/2005 17:05:48.740,214
144506,17370,0.019985,0.05,,[ Ok ],05/17/2005 17:05:48.760,214
144523,17371,0.01,0.05,,[ Ok ],05/17/2005 17:05:48.780,214
144540,17372,0.020005,0.05,,[ Ok ],05/17/2005 17:05:48.800,214
144559,17373,0.019992,0.05,,[ Ok ],05/17/2005 17:05:48.820,214
144560,17373,0.08,0.05,,Wrong sequence nr.,05/17/2005 17:05:48.820,214
144576,17374,0.019988,0.06,,[ Ok ],05/17/2005 17:05:48.840,214
144577,17374,0.07,0.06,,Wrong sequence nr.,05/17/2005 17:05:48.840,214
144592,17375,0.019992,0.06,,[ Ok ],05/17/2005 17:05:48.860,214
144593,17375,0.08,0.06,,Wrong sequence nr.,05/17/2005 17:05:48.860,214
144610,17376,0.019987,0.07,,[ Ok ],05/17/2005 17:05:48.880,214
144627,17377,0.019997,0.06,,[ Ok ],05/17/2005 17:05:48.900,214
144628,17377,0.08,0.06,,Wrong sequence nr.,05/17/2005 17:05:48.900,214
144642,17378,0.019989,0.07,,[ Ok ],05/17/2005 17:05:48.920,214
144661,17379,0.019997,0.07,,[ Ok ],05/17/2005 17:05:48.940,214
144678,17380,0.020004,0.06,,[ Ok ],05/17/2005 17:05:48.960,214
144679,17380,0.09,0.07,,Wrong sequence nr.,05/17/2005 17:05:48.960,214
144695,17381,0.019988,0.07,,[ Ok ],05/17/2005 17:05:48.980,214
144696,17381,0.07,0.07,,Wrong sequence nr.,05/17/2005 17:05:48.980,214
144713,17382,0.019982,0.08,,[ Ok ],05/17/2005 17:05:49.000,214
144732,17383,0.020002,0.07,,[ Ok ],05/17/2005 17:05:49.020,214
144733,17383,0.08,0.07,,Wrong sequence

[Asterisk-Users] Re: Audio flutter on OH323 output?

2005-05-22 Thread Tony Mountifield 
In article [EMAIL PROTECTED], I wrote:
 In article [EMAIL PROTECTED],
 Michael Manousos [EMAIL PROTECTED] wrote:
  
  Can you get an ethereal trace on a call with that problem?
  Run an RTP analysis on the captured stream (Tools Menu) and save
  the contents of the RTP packets in audio files. Then check if
  the playback of these files is normal or not.
 
 Aha, sounds like Ethereal has even more clever features - I didn't
 know it could do that.
 
 Fortunately, I've been running a continuous tcpdump capture on the
 Asterisk box of the traffic between it and the switch. I'll see
 what Ethereal thinks of it.

Hi Michael,

I've now done some RTP analysis of affected streams, and have found that
the times when we get the audio flutter correspond with parts of the
analysis showing repeated packets in the RTP stream. This only happens
on the outgoing stream; the incoming stream from the switch is perfect.

When a packet is repeated, it is identical to the previous and occurs
about 7 to 14 microsec after, as if something in the software has decided
to send the packet twice in immediate succession.

I've pasted below an extract from the CVS file saved from the RTP
stream analysis. I also saved the payload (pity it converts it to uLaw
instead of saving it in the stream's native format) and listened to it
to confirm the periods of distortion.

I would assume the problem is somewhere in the depths of openh323, but
any pointers in the right direction would be appreciated!

Cheers
Tony

Here is the CSV extract:

Packet,Sequence,Delay (s),Jitter (s),Marker,Status,Date,Length
144209,17353,0.02,0.03,,[ Ok ],05/17/2005 17:05:48.420,214
144225,17354,0.020006,0.03,,[ Ok ],05/17/2005 17:05:48.440,214
144226,17354,0.07,0.03,,Wrong sequence nr.,05/17/2005 17:05:48.440,214
144244,17355,0.019987,0.04,,[ Ok ],05/17/2005 17:05:48.460,214
144260,17356,0.019998,0.04,,[ Ok ],05/17/2005 17:05:48.480,214
144261,17356,0.08,0.04,,Wrong sequence nr.,05/17/2005 17:05:48.480,214
144276,17357,0.019991,0.04,,[ Ok ],05/17/2005 17:05:48.500,214
144296,17358,0.019992,0.04,,[ Ok ],05/17/2005 17:05:48.520,214
144298,17358,0.09,0.05,,Wrong sequence nr.,05/17/2005 17:05:48.520,214
144313,17359,0.019990,0.05,,[ Ok ],05/17/2005 17:05:48.540,214
144329,17360,0.02,0.05,,[ Ok ],05/17/2005 17:05:48.560,214
144349,17361,0.019992,0.05,,[ Ok ],05/17/2005 17:05:48.580,214
144368,17362,0.020003,0.05,,[ Ok ],05/17/2005 17:05:48.600,214
144369,17362,0.06,0.05,,Wrong sequence nr.,05/17/2005 17:05:48.600,214
144383,17363,0.019990,0.05,,[ Ok ],05/17/2005 17:05:48.620,214
144401,17364,0.019995,0.05,,[ Ok ],05/17/2005 17:05:48.640,214
144417,17365,0.020001,0.05,,[ Ok ],05/17/2005 17:05:48.660,214
144432,17366,0.019994,0.05,,[ Ok ],05/17/2005 17:05:48.680,214
144452,17367,0.019997,0.05,,[ Ok ],05/17/2005 17:05:48.700,214
144469,17368,0.020002,0.05,,[ Ok ],05/17/2005 17:05:48.720,214
144486,17369,0.019998,0.05,,[ Ok ],05/17/2005 17:05:48.740,214
144487,17369,0.08,0.05,,Wrong sequence nr.,05/17/2005 17:05:48.740,214
144506,17370,0.019985,0.05,,[ Ok ],05/17/2005 17:05:48.760,214
144523,17371,0.01,0.05,,[ Ok ],05/17/2005 17:05:48.780,214
144540,17372,0.020005,0.05,,[ Ok ],05/17/2005 17:05:48.800,214
144559,17373,0.019992,0.05,,[ Ok ],05/17/2005 17:05:48.820,214
144560,17373,0.08,0.05,,Wrong sequence nr.,05/17/2005 17:05:48.820,214
144576,17374,0.019988,0.06,,[ Ok ],05/17/2005 17:05:48.840,214
144577,17374,0.07,0.06,,Wrong sequence nr.,05/17/2005 17:05:48.840,214
144592,17375,0.019992,0.06,,[ Ok ],05/17/2005 17:05:48.860,214
144593,17375,0.08,0.06,,Wrong sequence nr.,05/17/2005 17:05:48.860,214
144610,17376,0.019987,0.07,,[ Ok ],05/17/2005 17:05:48.880,214
144627,17377,0.019997,0.06,,[ Ok ],05/17/2005 17:05:48.900,214
144628,17377,0.08,0.06,,Wrong sequence nr.,05/17/2005 17:05:48.900,214
144642,17378,0.019989,0.07,,[ Ok ],05/17/2005 17:05:48.920,214
144661,17379,0.019997,0.07,,[ Ok ],05/17/2005 17:05:48.940,214
144678,17380,0.020004,0.06,,[ Ok ],05/17/2005 17:05:48.960,214
144679,17380,0.09,0.07,,Wrong sequence nr.,05/17/2005 17:05:48.960,214
144695,17381,0.019988,0.07,,[ Ok ],05/17/2005 17:05:48.980,214
144696,17381,0.07,0.07,,Wrong sequence nr.,05/17/2005 17:05:48.980,214
144713,17382,0.019982,0.08,,[ Ok ],05/17/2005 17:05:49.000,214
144732,17383,0.020002,0.07,,[ Ok ],05/17/2005 17:05:49.020,214
144733,17383,0.08,0.07,,Wrong sequence nr.,05/17/2005 17:05:49.020,214
144747,17384,0.019989,0.08,,[ Ok ],05/17/2005 17:05:49.040,214
144768,17385,0.02,0.07,,[ Ok ],05/17/2005 17:05:49.060,214
144769,17385,0.07,0.07,,Wrong sequence nr.,05/17/2005 17:05:49.060,214
144784,17386,0.019986,0.07,,[ Ok ],05/17/2005 17:05:49.080,214
144801,17387,0.020001,0.07,,[ Ok ],05/17/2005 17:05:49.100,214
144802,17387,0.08,0.07,,Wrong

[Asterisk-Users] Re: Audio flutter on OH323 output?

2005-05-22 Thread Tony Mountifield 
In article [EMAIL PROTECTED], I wrote:
 
 I've now done some RTP analysis of affected streams, and have found that
 the times when we get the audio flutter correspond with parts of the
 analysis showing repeated packets in the RTP stream. This only happens
 on the outgoing stream; the incoming stream from the switch is perfect.
 
 When a packet is repeated, it is identical to the previous and occurs
 about 7 to 14 microsec after, as if something in the software has decided
 to send the packet twice in immediate succession.

I've been trying to read through the openh323 and pwlib code related to
writing the RTC stream. It's like a cross between peeling an onion and
wading through treacle :-(

I've focussed on the function RTP_UDP::WriteData() in openh323/src/rtp.cxx
where dataSocket-WriteTo() is called in a loop. Is it possible that there
are occasions where the packet did get sent but for some reason WriteTo()
returned an error?

I'm clutching at straws really, but without solving this, the system is
not good enough to deploy.

How can I turn on the PTRACE output from openh323, and where would it
be sent to?

Cheers
Tony
-- 
Tony Mountifield
Work: [EMAIL PROTECTED] - http://www.softins.co.uk
Play: [EMAIL PROTECTED] - http://tony.mountifield.org
___
Asterisk-Users mailing list
Asterisk-Users@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

[Asterisk-Users] Re: Audio flutter on OH323 output?

2005-05-18 Thread Tony Mountifield 
In article [EMAIL PROTECTED],
Michael Manousos [EMAIL PROTECTED] wrote:
 
 Can you get an ethereal trace on a call with that problem?
 Run an RTP analysis on the captured stream (Tools Menu) and save
 the contents of the RTP packets in audio files. Then check if
 the playback of these files is normal or not.

Aha, sounds like Ethereal has even more clever features - I didn't
know it could do that.

Fortunately, I've been running a continuous tcpdump capture on the
Asterisk box of the traffic between it and the switch. I'll see
what Ethereal thinks of it.

Cheers
Tony
-- 
Tony Mountifield
Work: [EMAIL PROTECTED] - http://www.softins.co.uk
Play: [EMAIL PROTECTED] - http://tony.mountifield.org
___
Asterisk-Users mailing list
Asterisk-Users@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