|Summary:||ASTERISK-02396: iax2 show channels report too great lag|
|Date Opened:||2004-09-14 03:58:03||Date Closed:||2004-10-02 21:21:38|
|Description:||iax2 show channels give me :|
192.168.9.13 pbxau 00012/16385 00005/00006 269746266ms 0633ms 0000ms ILBC
****** ADDITIONAL INFORMATION ******
the call is trunked between too asterisk's (last CVS)
quality is good, but the reported lag is false.
|Comments:||By: lists (lists) 2004-09-14 04:05:33|
my trunkfreq is set to 30ms, when i set it back to 20ms there is no more problem.
By: zoa (zoa) 2004-09-14 05:59:36
hmm.. why did you change it?
And did you change it on both ends ?
By: lists (lists) 2004-09-14 07:54:49
I changed it to reduce the bandwith used by this trunk
(and of course changed it on both ends)
By: zoa (zoa) 2004-09-14 08:05:12
hmm.. i don't think that would make any difference in bandwidth...
By: Mark Spencer (markster) 2004-09-14 08:26:53
Changing trunkfreq could definitely break things because if you're sending 20ms packets but have a trunkfreq of 30ms then you're manually introducing jitter.
By: zoa (zoa) 2004-09-15 03:57:45
Just to get something straight, when using iLBC, will asterisk send 20ms packets or 30ms packets ?
From my measurements, with iLBC, asterisk will send 70pps for a while, then send 60pps then some more 70pps.
most other codecs use 100pps fixed. (except for : g723.1 - 60/70pps and lpc10 - random).
Does this mean that for iLBC we need a different trunk frequency ?
By: Brian West (bkw918) 2004-09-15 08:23:32
ilbc is a 30ms codec so I think that might need 30ms?
By: zoa (zoa) 2004-09-15 08:52:25
i asked kram before, and he claims thats not the case bkw.
i think we need a little diagram to understand all this :)
By: Brian West (bkw918) 2004-09-24 06:57:30
And how many times has kram said one thing to only say the exact reverse a few days later :P Its not really his fault... haha he is a very busy man.
By: adam (adam) 2004-09-24 08:30:20
*'s iLBC only allows 30ms frames, although the spec allows for 20ms too. So setting the trunk size to 30ms should work
By: Mark Spencer (markster) 2004-09-27 20:33:45
Is this still the case?
By: Mark Spencer (markster) 2004-10-02 10:09:11
Bug poster seems to have lost interest. Should be fixed by stevendavies patch earlier just before 1.0 anyway.