Exceptionally long voice queue length

Hi,

what could be causing these?
[2013-01-31 09:24:32] WARNING[874] channel.c: Exceptionally long voice queue length queuing to Local/307@from-internal-00000bc4;1

and… related errors:
[2013-02-14 15:29:57] ERROR[2002] chan_capi.c: Could not write to pipe for ISDN1#29 fd:57 errno:11

Author of chan_capi suggests that “channel status of your Asterisk is incorrect when this occurs.
Maybe a Asterisk-thread hangs.” More about it: lists.melware.net/pipermail/chan … 01345.html

It happens randomly, once or twice per week, usually under moderate load (10 concurrent calls+). The only known remedy for above is killing asterisk process and starting it afresh. Could it be the parameter pedantic=no?

We were experiencing that problem on asterisk 1.8.20.1, using Cisco phones 7960 with SIP firmware 8.12 (I’m testing 1.8.21 to see if it’s any better).

Any help or idea would be greatly appreciated.

Thanks
Andrej

Similar problem, suggesting CEL is the culprit:
issues.asterisk.org/jira/browse/ASTERISK-20335

In cel.conf I have parameter cel=yes commented, but issuing command modules show like cel shows:

asterisk2*CLI> module show like cel
Module Description Use Count
cel_odbc.so ODBC CEL backend 0
app_celgenuserevent.so Generate an User-Defined CEL event 0
cel_custom.so Customizable Comma Separated Values CEL 0
cel_manager.so Asterisk Manager Interface CEL Backend 0
4 modules loaded

Another posts suggests disabling pthread as timing source. I have both pthread and dahdi as timing sources loaded.

Would I need to use noload parameter in modules.conf to apply above suggestions?

Any other ideas?

Best regards
T