Now putting this oddity aside, I still have a strange behaviour since the ACD doesn't seem to handle de DCFs correctly, and I'm wondering if this 'odd' registration of the endpoints could be the cause.
Here is a small scenario showing this :
( memo : endpoint is registered with RCF|192.168.0.70:1720|17177:dialedDigits=17177:h323_ID=17177:dialedDigits=17177:h323_ID|terminal|5427_endp; )
- Both aliases are available to the waiting queue :
Agent list
17177:dialedDigits (5427_endp): logged in state: AVAILABLE last call: 15:12:40
17177:h323_ID (5427_endp): logged in state: AVAILABLE last call: 15:12:40
Pending list (0 entries)
- I make a call to 7177, everything goes ok, then I hangup :
DEBUG: RouteRequest|192.168.0.55:1720|5439_endp|5248|7177|7190:dialedDigits=7190:dialedDigits;
Routing call for 7177 to 17177:dialedDigits(5427_endp)
DEBUG: ACD->GK:RouteToAlias 17177 5439_endp 5248
DEBUG: ACF|192.168.0.55:1720|5439_endp|5248|17177:dialedDigits|7190:dialedDigits=7190:dialedDigits|false;
DEBUG: ACF|192.168.0.70:1720|5427_endp|4309|7177:dialedDigits=7177:dialedDigits|Loic Le Bel:h323_ID|true;
DEBUG: CDR|181|06 02 4b 97 0a 50 d8 11 a6 7d 00 50 da de ba 15|17|Mon, 26 Jan 2004 15:12:36 +0100|Mon, 26 Jan 2004 15:12:53 +0100|192.168.0.55:1720|5439_endp|192.168.0.70:1720|5427_endp|17177:dialedDigits|7190:dialedDigits=7190:dialedDigits|LinkeoGK;
DEBUG: DCF|192.168.0.55|5439_endp|5248|normalDrop;
DEBUG: DCF|192.168.0.70|5427_endp|4309|normalDrop;
DEBUG: HTTP:GET /status HTTP/1.1
DEBUG: Command accepted: Status
Agent list
17177:dialedDigits (5427_endp): logged in state: TALKING last call: 15:12:57
17177:h323_ID (5427_endp): logged in state: AVAILABLE last call: 15:12:40
Pending list (0 entries)
The DCFs are looking good to me, but the ACD is still considering 17177:dialedDigits in talking state, which seems wrong
- Now making a second call to 7177, the call is ok since there is a second alias still available for the endpoint, then I hangup :
DEBUG: RouteRequest|192.168.0.55:1720|5439_endp|5249|7177|7190:dialedDigits=7190:dialedDigits;
Routing call for 7177 to 17177:h323_ID(5427_endp)
DEBUG: ACD->GK:RouteToAlias 17177 5439_endp 5249
DEBUG: ACF|192.168.0.55:1720|5439_endp|5249|17177:dialedDigits|7190:dialedDigits=7190:dialedDigits|false;
DEBUG: ACF|192.168.0.70:1720|5427_endp|4310|7177:dialedDigits=7177:dialedDigits|Loic Le Bel:h323_ID|true;
DEBUG: CDR|182|08 02 4b 97 0a 50 d8 11 a6 7d 00 50 da de ba 15|10|Mon, 26 Jan 2004 15:13:21 +0100|Mon, 26 Jan 2004 15:13:31 +0100|192.168.0.55:1720|5439_endp|192.168.0.70:1720|5427_endp|17177:dialedDigits|7190:dialedDigits=7190:dialedDigits|LinkeoGK;
DEBUG: DCF|192.168.0.55|5439_endp|5249|normalDrop;
DEBUG: DCF|192.168.0.70|5427_endp|4310|normalDrop;
DEBUG: HTTP:GET /status HTTP/1.1
DEBUG: Command accepted: Status
Agent list
17177:dialedDigits (5427_endp): logged in state: TALKING last call: 15:12:57
17177:h323_ID (5427_endp): logged in state: TALKING last call: 15:13:42
Pending list (0 entries)
Once again the DCFs are looking good, but now both aliases are considered in talking state (remember both call where properly terminated) and are only reset after the next sanity check.
So is this behaviour truly a bug or could the inital RCF from the endpoint causing trouble (potential problem with the parsing of the aliases) ?
Any hint is welcome before I look further in the code of the ACD :-)
(For info, I'm using GnuGK 2.0.7, ACD version is 1.3, tested under Linux with both sun 1.3.1 and 1.4.1 jre)
Best Regards, Loic
Jan Willamowius wrote:
Hi,
the first version of the ACD had a model of on endpoint == one agent. That was pretty much broken considering that many endpoints can handle many simultaneous calls. The current version treas one alias == one agent and the endpoint may register many aliases, but must send for all of them in the ACF if it can't handle simultaneous calls. The only endpoint I had to test that with was netmeeting and it did that correctly (don't sdtart a flamewar here about all the other things it doesn't do correctly). ;-)
- Jan
Loic LE BEL wrote:
Hi folks,
I'm wondering what is the correct behaviour of endpoints registered with multiple aliases in the ACD, because I'm experiencing some availability conflicts on the waiting queues. Here is the situation :
- My virtual queue in GnuGK is definied by : [CTI::Agents] VirtualQueuePrefixes=7177
- The agent list in ACDAgentConfig.properties for this queue is defined by :# Virtual queue 7177 17177=7177
- The endpoint (Siemens LP5100 ip phone) is registered on the gatekeeper with 17177 as E.164 number : RCF|192.168.0.70:1720|17177:dialedDigits=17177:h323_ID=17177:dialedDi gits=17177:h323_ID|terminal|5427_endp;
When looking at the agent list with the /status http request, we can see that the endpoint is apearing twice in the list :
Agent list 17177:dialedDigits (5427_endp): logged in state: AVAILABLE last call: 11:57:26 17177:h323_ID (5427_endp): logged in state: AVAILABLE last call: 11:57:26
Pending list (0 entries)
So the problem is that when this endpoint is busy, only one of the two
instances in the list is considered in talking state, thus giving
'agent 17177' the avaible state although it's already busy.
Shouldn't all instances of an aliases be considered in talking state when receiving a call or is my endpoint incorrectly registering on the
gk with two aliases instead of only one ? (bad rcf ?)
Best regards,
Loic
------------------------------------------------------- The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. http://www.eclipsecon.org/osdn _______________________________________________ List: Openh323gk-users@lists.sourceforge.net Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549 Homepage: http://www.gnugk.org/
