[asterisk-dev] Devicestate change is very slow updating

[Wolfgang Pichler]

Hi all,

i am at time working to improve the queue system we are using here - 
following setup

asterisk 1.4.21.1 with state_interface patch

Realtime Queue and queue members - each queue member does register as 
Local/AGENTNUM at callAgent/n with an attached state interface which does 
point to the real SIP device

I have had some troubles of agents getting a second call  - so i have 
enabled some more debuging and found out that the device state change 
information does come to slow - about 10 seconds after the real device 
state change - so the queue does assume the member status is still not 
in use - and does place a second call on the member.

Here some logs


[Oct 13 10:11:51] VERBOSE[5647] logger.c:     -- Executing 
[1012 at callAgent:12] Dial("Local/1012 at callAgent-0f3b,2", 
"SIP/spa941_0002||M(callAgentConnect^15^1223885511.8^SIP/spa941_0002)") 
in new stack
[Oct 13 10:11:51] VERBOSE[5647] logger.c:     -- Called spa941_0002
[Oct 13 10:11:51] VERBOSE[5647] logger.c:     -- 
SIP/spa941_0002-08419978 is ringing
[Oct 13 10:11:51] VERBOSE[5645] logger.c:     -- 
Local/1012 at callAgent-0f3b,1 is ringing
[Oct 13 10:11:53] VERBOSE[5647] logger.c:     -- 
SIP/spa941_0002-08419978 answered Local/1012 at callAgent-0f3b,2

[Oct 13 10:11:53] WARNING[5645] app_queue.c: The device state of this 
queue member, 1012, is still 'Not in Use' when it probably should not 
be! Please check UPGRADE.txt for correct configuration settings.

[Oct 13 10:12:01] DEBUG[1746] app_queue.c: Device 'SIP/spa941_0002' 
changed to state '3' (Busy).               <--- Here the app_queue does 
then do the update_status for the queue member - but as you can see - 
much too late !


And the ringing device state change was even not fired...

Here the relevant config settings in sip.conf

notifyringing = yes
limitonpeers=yes
rtcachefriends=yes
rtupdate=yes

SIP Devices are defined as type=friend and busy-level=1 and call-limit=1

How could it be that the state change is so slow ?

regards,
Wolfgang Pichler