Congestion signal is not taking effect on vicidial_autocalls

[omar]

Hello

After a congestion signal has been sent to my system and the call has been hang up, the system doesn't clear the vicidial_auto_calls table and the call appears in http://ip/vicidial/realtime_report.php as ringing for about one minute more.

The cli output for this kind of situation is:

Code: Select all

[Oct 23 14:31:16]   == Manager 'sendcron' logged on from 127.0.0.1
[Oct 23 14:31:16]     -- Executing [919047863858@default:1] AGI("Local/919047863858@default-43c7,2", "agi://127.0.0.1:4577/call_log") in new stack
[Oct 23 14:31:16]     -- AGI Script agi://127.0.0.1:4577/call_log completed, returning 0
[Oct 23 14:31:16]     -- Executing [919047863858@default:2] Dial("Local/919047863858@default-43c7,2", "SIP/9047863858@vicibox||tTor") in new stack
[Oct 23 14:31:16]     -- Called 9047863858@vicibox
[Oct 23 14:31:17]     -- Got SIP response 503 "Service Unavailable" back from 64.2.142.189
[Oct 23 14:31:17]     -- SIP/vicibox-0000001c is circuit-busy
[Oct 23 14:31:17]   == Everyone is busy/congested at this time (1:0/1/0)
[Oct 23 14:31:17]     -- Executing [919047863858@default:3] Hangup("Local/919047863858@default-43c7,2", "") in new stack
[Oct 23 14:31:17]   == Spawn extension (default, 919047863858, 3) exited non-zero on 'Local/919047863858@default-43c7,2'
[Oct 23 14:31:17]     -- Executing [h@default:1] DeadAGI("Local/919047863858@default-43c7,2", "agi://127.0.0.1:4577/call_log--HVcauses--PRI-----NODEBUG-----34-----CONGESTION----------") in new stack
[Oct 23 14:31:18]     -- AGI Script agi://127.0.0.1:4577/call_log--HVcauses--PRI-----NODEBUG-----34-----CONGESTION---------- completed, returning 0
[Oct 23 14:31:20]   == Manager 'sendcron' logged off from 127.0.0.1


And the vicidial_auto_calls table still has this register for about 1 minute more

Code: Select all

mysql> SELECT status, uniqueid, callerid, channel, phone_number, call_time, call_type, stage, last_update_time FROM `vicidial_auto_calls`;
+--------+----------+----------------------+---------+--------------+---------------------+-----------+-------+---------------------+
| status | uniqueid | callerid             | channel | phone_number | call_time           | call_type | stage | last_update_time    |
+--------+----------+----------------------+---------+--------------+---------------------+-----------+-------+---------------------+
| SENT   | NULL     | V0231431150000000170 | NULL    | 9047863858   | 2012-10-23 14:31:15 | OUT       | START | 2012-10-23 14:31:15 |
+--------+----------+----------------------+---------+--------------+---------------------+-----------+-------+---------------------+
1 row in set (0.00 sec)


This problem is causing a slow performance in autocalls, because we need to try inmediatly another lead if we have congestion, busy, etc statuses.


My environment is:


ViciBox v.3.1.15.x86_64
Build: 120831-1523
Version: 2.6b0.5
SVN Rev: 1863
asterisk 1.4.39.2-vici
dahdi-2.4.1.2-24.1.x86_64
wanpipe-3.5.20-13.17.x86_64
1xDB, 1xWeb, 2xDialer, 1x Archive


Thanks

[williamconley]

known bug/feature. upgrade (which may or may not fix it, i have not had anyone check recently ...).

or install sangoma cpd/netborder (which completes calls properly for every possible permutation).

or stop making calls that get congested (this is not really a normal part of dialing ... you should only get congestion if something is wrong). ordinarily 1% or less of calls should congest.

we've written a patch that will immediately clear these for a few clients, and it mimics the sangoma method ... but it should not really be necessary. and remember it is only "mimicing" the sangoma method, but it does result in the call record departing immediately as desired.

[DomeDan]

A upgrade (right now) wont change this feature, I've been running with my patch more than a year now http://www.vicidial.org/VICIDIALmantis/view.php?id=510

[omar]

A upgrade (right now) wont change this feature, I've been running with my patch more than a year now http://www.vicidial.org/VICIDIALmantis/view.php?id=510

Great job DomeDan, maybe i am going to use your patch.

However, I don't know if it is a feature or a bug of vicidial.

And if it is a bug if it is taken into account to solve the bug in a near feature.

Thanks

[williamconley]

with our version, we noticed that netborder resolves the issue ... but there's nothing in the Vicidial code that really does anything more than "take note" of the fact that "netborder says the call result is 'xx' so terminate it and file the result". So we mimic that for congestion calls and it seems to "blend right in" as if netborder were present and sending that signal. Not a lot of code involved.

I expected that to be fixed in the later versions.

[omar]

I expected that to be fixed in the later versions.

me too