asterisk users - Oct 2009 - Origin of "Exceptionally long voice queue length queuing to IAX2/blahblah" messages

Hello,

I'm using asterisk for a quite long period, i integrated a lot of stuff to
make it behave like any carrier class system, so users can:

manage Call forward on busy
manage Call forward on no answer
manage unconditionnal call forward
call back missed calls


and a lot of such services that can be both configured by menus and dtmf or by
web interfaces (PHP+SQL), in fact i recreated services everybody use to get with
their classical operator.

All of theses functions are achieved throught dialplan commands, first MYSQL
commands, conditions on results etc..

It worked as a charm but got wrong with the growing number of users, i began
having "asterisk network connectivity loss", in fact qualified IAX2
friends became "UNREACHABLE" for seconds, connected calls were
disturbed when not simply cut.

I first thought of IAX problems on charge, moved to SIP, but got the same
problems, i then moved my MYSQL commands to ODBC functions, optimized my macros,
it worked for a short while, i changed dialplan to reduce as much as possible
LOCAL channels, it worked for a while but now the problem comes back with its
IAX messages:

[Oct 16 17:55:40] WARNING[2514]: channel.c:951 __ast_queue_frame: Exceptionally
long voice queue length queuing to IAX2/blahblah
[Oct 16 17:55:40] WARNING[2511]: channel.c:951 __ast_queue_frame: Exceptionally
long voice queue length queuing to IAX2/blahblah
[Oct 16 17:57:38] NOTICE[2513]: chan_iax2.c:9186 __iax2_poke_noanswer: Peer
'XXX' is now UNREACHABLE! Time: 9
[Oct 16 17:57:38] NOTICE[2505]: chan_iax2.c:9186 __iax2_poke_noanswer: Peer
'YYY' is now UNREACHABLE! Time: 9
[Oct 16 17:57:55] WARNING[2505]: chan_iax2.c:804 jb_warning_output: Resyncing
the jb. last_delay 0, this delay 22334, threshold 1000, new offset -22334
[Oct 16 17:57:57] WARNING[2510]: chan_iax2.c:804 jb_warning_output: Resyncing
the jb. last_delay -135, this delay 1529, threshold 1000, new offset -1529
[Oct 16 17:58:05] NOTICE[2507]: chan_iax2.c:8264 socket_process: Peer
'XXX' is now REACHABLE! Time: 4
[Oct 16 17:58:07] NOTICE[2511]: chan_iax2.c:8264 socket_process: Peer
'YYY' is now REACHABLE! Time: 11

On some servers i used less advanced dialplan, with ODBC functions but geeting
rid of LOCAL channels, and, until now, i didn't got any problem on them...
so i suspect LOCAL channels, without proof of it, but what else? And for my
advanced services, it will be hard to get rid of them without scripts.... so,
i'm actually thinking of moving my dialplan commands to AGI script(s), so i
may get rid of LOCAL channels and manage the processes more easily, but reading
wikis and mailing lists, some of them consider dialplan commands better than
scripts for performances... but if so, why are AGI used by so much projects
(asterbilling or ccard systems), which seems to support very large amount of
users....

So, if somebody on the list has any idea or advice... please let me know...


Regards

Daren
 		 	   		  
_________________________________________________________________
A la recherche de bons plans pour une rentr?e pas ch?re ? Bing ! Trouvez !
http://www.bing.com/search?q=bons+plans+rentr%C3%A9e&form=MVDE6
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.digium.com/pipermail/asterisk-users/attachments/20091016/99f4f3ae/attachment.htm