Summary:ASTERISK-02846: Dropping voice to exceptionally long queue
Reporter:outtolunc (outtolunc)Labels:
Date Opened:2004-11-18 00:41:38.000-0600Date Closed:2004-11-25 09:59:28.000-0600
Versions:Frequency of
Description:i have a T400P in a dev enviroment.  currently i'm using IAX2 over nufone for outbound pstn.  everything was fine a week ago.  then updated to current CVS.  (i've completely wiped asterisk zaptel libpri (many times, many variations). The reason i put this under app_queue is that is that is were i can replicate everytime.


is a shorten version of the debug output. (one it hits, i'll get that error in the thousands (until i kill -9 asterisk))

even shorter output
Nov 16 21:06:45 DEBUG[7387]: Dropping voice to exceptionally long queue on
Nov 16 21:06:45 DEBUG[7387]: Immediately destroying 4, having received
Nov 16 21:07:12 DEBUG[7387]: Sending VNAK
Nov 16 21:07:35 DEBUG[7387]: Raw Hangup, src=2, dst=504
Nov 16 21:07:35 DEBUG[7387]: Raw Hangup, src=2, dst=504
Nov 16 21:07:35 DEBUG[7387]: Raw Hangup, src=2, dst=504
Nov 16 21:08:24 DEBUG[7387]: Immediately destroying 1, having received INVAL
Nov 16 21:09:12 DEBUG[7366]: Avoiding initial deadlock for 'Agent/301'

the only other bugs i've seen mention 'Dropping voice' are ASTERISK-354 and ASTERISK-1302  they are similar, not same.

i can load asterisk, login agent (agentlogin method), listen to MOH, gen a call, connect, toss into queue, it finds the agent, and as soon as it bridges, each end of the call can hear themselves, they can not hear the other end.  starts spewing the 'Dropping voice' errors.  the systems 'acts' like it hangs, obviously thats just a reaction.

i'm usually on IRC from 9am-4:30pm PST if you want to discuss this or jump in for a hands on look.
Comments:By: Mark Spencer (markster) 2004-11-20 11:48:03.000-0600

If you can find me on IRC today I'd like to take a look at this while it is happening. Also please confirm you have *not* patched your Asterisk from CVS

By: outtolunc (outtolunc) 2004-11-21 14:14:14.000-0600

well i was being abused by my buddy hanging drywall yesterday, and i won't be in the office till monday.  maybe i should have said ...
i'm usually on IRC from 9am-4:30pm PST >> WEEKDAYS << ,if you want to discuss this or jump in for a hands on look. <G>

and no, no patches.  i must have did 30+ cvs checkouts of various combinations asterisk libpri zaptel to try and narrow down the problem tree/source date.

first thing tomorrow morning i'll
cd /usr/src; rm -rf asterisk libpri zaptel; login to cvs; cvs co asterisk libpri zaptel; cd /usr/src/zaptel; make clean; make install; cd /usr/src/libpri; make clean; make install; cd /usr/src/asterisk; make clean; make update; make install;

(scrolling back at each make looking for errors) re-test, and post the results. and make myself available on IRC.  

side note:  i did notice that if i did a make valgrind vice make install on asterisk, that when the agent/queue_call bridged they were able to hear themselves and the other ends (working normal).  so it's probably in the optimazations somewhere.  

gotta love it when freaks, you attempt to debug and in that mode it works <G>

By: outtolunc (outtolunc) 2004-11-22 13:17:34.000-0600

well it's monday morning, and i completed all the tasks listed in th previous bugnote.  (Asterisk CVS-HEAD-11/22/04-15:14:07) and retested.

first call worked, second ... they bridge and they can only hear themselves, not the other ends. (same issue)

side note: after a failed attempt it seems to effect the zaptel device, i have to reboot to get back to one attempt working.

By: outtolunc (outtolunc) 2004-11-23 01:25:29.000-0600

received te410p, reinstalled dev system with Fedora Core 2 (latest yum updates), and the te410p.  moved configs over and retested.  

All is well with the world again.

thank you

you can close this bug.

By: Mark Spencer (markster) 2004-11-23 01:28:08.000-0600

RedHat 8.0 issue