RE: RE: Globalcall stress test
"Julien Chavanton" <[email protected]>
| Newsgroups | gmane.comp.gnu.bayonne.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi made some modification and the server receiving call is working just fine after 150 000 really fast calls. Often I receive an Offered while still in release state and the small modification are working. However my test script on the answering server was only doing sleep. Now there is another problem with the voice dx_ and dt_ function call I think we must modify dt_listen() by gc_listen() since it is not good to make lower level R2 call while using Globalcall. I will modify this and make some play, and after conference to test the sc_bus and dx_ function again since currently there is a problem during theses tests. Julien LINE ISDN STATUS CALLEDID CALLERID CALLS DURATION 0 RELEASED 6566 1 RELEASED 6577 2 RELEASED 6590 3 RELEASED 6595 4 RELEASED 6603 5 RELEASED 6607 6 RELEASED 6614 7 RELEASED 6991 8 RELEASED 6995 9 RELEASED 6998 10 RELEASED 7005 11 RELEASED 7014 12 RELEASED 6643 13 RELEASED 6646 14 RELEASED 6654 15 RELEASED 6661 16 RELEASED 6672 17 RELEASED 6679 18 RELEASED 6683 19 RELEASED 6690 20 RELEASED 6699 21 RELEASED 6706 22 RELEASED 7093 _____ From: Julien Chavanton Sent: August 18, 2005 11:17 AM To: Julien Chavanton; 'Etoile Dièse'; '[email protected]' Subject: RE: [Bayonne-devel] RE: Note that this modification is really important since this situation currently result in port being blocked since "idle" does not support disconnected. _____ From: Julien Chavanton Sent: August 18, 2005 11:15 AM To: Julien Chavanton; Etoile Dièse; [email protected] Subject: RE: [Bayonne-devel] RE: The only Bayonne state where there can be more then one simultaneous call is "release" so I think we can modify this state to detect an offered and notify the next "idle". Later we shall go further and have a real support for multiple calls per line but for now this is not required since as far as I know the Telco does not offer call on hold. Julien _____ From: bayonne-devel-bounces+j.chavanton=oneworldinteractive.com@gnu.org [mailto:bayonne-devel-bounces+j.chavanton=oneworldinteractive.com@gnu.org] On Behalf Of Julien Chavanton Sent: August 18, 2005 10:39 AM To: Etoile Dièse; [email protected] Subject: RE: [Bayonne-devel] RE: The problem is a bit more complex then expected: There can be more then one simultaneous call on a line device (For example ISDN call on hold). And when between drop and release we can receive an offered. So Bayonne Globalcall shall be modified to support more then one CRN on each line. I have to evaluate the complexity of such modification. Here is an example of 2 simultaneous CRN on one line device. dx(12): GCEV_DISCONNECTED HDL:31 CRN:33616924 dx(12): step 2 exit() dx(12): exit dx(12): script exiting dx(12): HANGUP TRUNK_ENTER_ STATE reset_timer:1090604252 hangup_timer:100 dx(12): hangup dx(12): gc_DropCall(33616924) == SUCCESS dx(12): detach script dx(12): GCEV_DROPCALL HDL:31 CRN:33616924 dx(12): HANGUP TRUNK_CALL_DISCONNECT:31 CRN:33616924 dx(12): release dx(12): gc_ReleaseCallEx(33616924) == SUCCESS dx(12): GCEV_OFFERED HDL:31 CRN:33617948 dx(12): GCEV_RELEASECALL HDL:31 CRN:33616924 dx(12): ENTERING IDLE STATE:31 CRN:-1 dx(12): idle dx(12): GCEV_DISCONNECTED HDL:31 CRN:33617948 -----Original Message----- From: bayonne-devel-bounces+j.chavanton=oneworldinteractive.com@gnu.org [mailto:bayonne-devel-bounces+j.chavanton=oneworldinteractive.com@gnu.org] On Behalf Of Etoile Dièse Sent: August 17, 2005 5:06 PM To: [email protected] Subject: Re: [Bayonne-devel] RE: Cool, May we have this modif ? Le mercredi 17 Août 2005 21:17, Julien Chavanton a écrit : > I have made modification to Bayonne Globalcall to remove CRN > modification from outside driver.cpp (the Dialogic event handler) > > This way the CRN is always correct. > > > > I am stressing the port as much as I can. > > > > I have good result. > > > > > > > > > > _____ > > From: Julien Chavanton > Sent: August 17, 2005 10:25 AM > To: '[email protected]' > Subject: > > > > While debugging and stressing Bayonne1 / Globalcall I have found another > problem that take place under heavy stress > > Here we can see a call clearing and a new call coming with almost no > delay between them > > > > Bayonne is in idle while the new call as been offered. > > > > The problem is that Bayonne idle clear the CRN but in this case the CRN > of the new call is cleared ! > > > > I think only driver.cpp should be able to clear CRN since it is the > driver event handler. > > > > I am testing the modification. > > > > Aug 27 08:37:18 localhost bayonne: dx(16): GCEV_DISCONNECTED HDL:39 > > Aug 27 08:37:18 localhost bayonne: dx(16): step 2 exit() > > Aug 27 08:37:18 localhost bayonne: dx(16): exit > > Aug 27 08:37:18 localhost bayonne: dx(16): script exiting > > Aug 27 08:37:18 localhost bayonne: dx(16): hangup > > Aug 27 08:37:18 localhost bayonne: dx(16): detach script > > Aug 27 08:37:18 localhost bayonne: dx(16): GCEV_DROPCALL HDL:39 > > Aug 27 08:37:18 localhost bayonne: dx(16): release > > Aug 27 08:37:19 localhost bayonne: dx(16): GCEV_RELEASECALL HDL:39 > > Aug 27 08:37:19 localhost bayonne: dx(16): GCEV_OFFERED HDL:39 > CRN:33604644 > > Aug 27 08:37:19 localhost bayonne: dx(16): idle > > Aug 27 08:37:19 localhost bayonne: dx(16): gc_AcceptCall error HDL:39 > CRN:-1 > > Aug 27 08:37:19 localhost bayonne: dx(16): ring > > Aug 27 08:37:28 localhost bayonne: dx(16): GCEV_DISCONNECTED HDL:39 > > Aug 27 08:37:29 localhost bayonne: dx(16): hangup > > Aug 27 08:37:29 localhost bayonne: dx(16): gc_DropCall 66 Invalid call > reference number has been used > > Aug 27 08:37:29 localhost bayonne: dx(16): idle -- Etoile Dièse _______________________________________________ Bayonne-devel mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bayonne-devel _______________________________________________ Bayonne-devel mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bayonne-devel