Time for hang-up

[mtancoff]

When an agent is in an inbound call, after the customer has hang-hup, vici takes thirty seconds to send the hang-up to agent.

Is this on purpose so the agent can fill all data?

[mflorell]

Do you have a Wrapup time defined in the campaign detail screen?

If not there is most likely a problem somewhere.

What is your server loadavg when this occurs?

What Linux distro are you using?

[mtancoff]

No wrap-up time defined for this campaign

We´re using centOS 4.3

How do I see loadavg?

[SuPrSluG]

type top at a command line and it's right at the top of the list

[mtancoff]

top - 16:43:51 up 5:05, 4 users, load average: 0.29, 0.33, 0.34
Tasks: 92 total, 1 running, 91 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.8% us, 0.3% sy, 2.2% ni, 96.7% id, 0.0% wa, 0.0% hi, 0.0% si
Mem: 1001904k total, 438780k used, 563124k free, 90840k buffers
Swap: 779144k total, 0k used, 779144k free, 132760k cached


This is it, nothing seems wrong.

Everything looks fine, but vici takes about 30 seconds to send the hang-up to customer, counting from the moment when asterisk -gcvvvvvvvvvvvvv shows:


-- Executing DeadAGI("Zap/1-1", "VD_hangup.agi|PRI-----NODEBUG-----0---------------") in new stack
-- Launched AGI Script /var/lib/asterisk/agi-bin/VD_hangup.agi
2006-09-06 16:53:14|VD_hangup.agi|Perl Environment Dump:
2006-09-06 16:53:14|VD_hangup.agi|0|PRI-----NODEBUG-----0---------------
2006-09-06 16:53:14|VD_hangup.agi|AGI Environment Dump:
2006-09-06 16:53:14|VD_hangup.agi| -- accountcode =
2006-09-06 16:53:14|VD_hangup.agi| -- callerid = unknown
2006-09-06 16:53:14|VD_hangup.agi| -- calleridname = Y0906165307000000052
2006-09-06 16:53:14|VD_hangup.agi| -- callingani2 = 0
2006-09-06 16:53:14|VD_hangup.agi| -- callingpres = 0
2006-09-06 16:53:14|VD_hangup.agi| -- callingtns = 0
2006-09-06 16:53:14|VD_hangup.agi| -- callington = 0
2006-09-06 16:53:14|VD_hangup.agi| -- channel = Zap/1-1
2006-09-06 16:53:14|VD_hangup.agi| -- context = default
2006-09-06 16:53:14|VD_hangup.agi| -- dnid = unknown
2006-09-06 16:53:14|VD_hangup.agi| -- enhanced = 0.0
2006-09-06 16:53:14|VD_hangup.agi| -- extension = h
2006-09-06 16:53:14|VD_hangup.agi| -- language = en
2006-09-06 16:53:14|VD_hangup.agi| -- priority = 2
2006-09-06 16:53:14|VD_hangup.agi| -- rdnis = unknown
2006-09-06 16:53:14|VD_hangup.agi| -- request = VD_hangup.agi
2006-09-06 16:53:14|VD_hangup.agi| -- type = Zap
2006-09-06 16:53:14|VD_hangup.agi| -- uniqueid = 1157575982.3
2006-09-06 16:53:14|VD_hangup.agi|AGI Variables: |1157575982.3|Zap/1-1|h|Zap|Y0906165307000000052|
2006-09-06 16:53:14|VD_hangup.agi|DEBUG: NODEBUG
2006-09-06 16:53:14|VD_hangup.agi|VD_hangup : Y0906165307000000052 Zap/1-1 2 52
2006-09-06 16:53:14|VD_hangup.agi||SELECT lead_id,callerid FROM vicidial_auto_calls where uniqueid = '1157575982.3' limit 1;|
2006-09-06 16:53:14|VD_hangup.agi|-- VDAC record deleted: |1| |52|1157575982.3|Y0906165307000000052|127.0.0.1
2006-09-06 16:53:14|VD_hangup.agi||SELECT start_epoch,status FROM vicidial_log where uniqueid='1157575982.3' and lead_id='52' limit 1;|
2006-09-06 16:53:14|VD_hangup.agi|no VDL record found: 1157575982.3 Y0906165307000000052 52 1157575982.3
2006-09-06 16:53:14|VD_hangup.agi||SELECT start_epoch,status,closecallid FROM vicidial_closer_log where lead_id = '52' and call_date > "2006-09-06 10:53:14" order by call_date desc limit 1;|
2006-09-06 16:53:14|VD_hangup.agi|no VDL or VDCL record found: 1157575982.3 Y0906165307000000052 52 1157575982.3
-- AGI Script VD_hangup.agi completed, returning 0
-- Hungup 'Zap/1-1'

Since this hung-up Zap/1-1, agent console delays thirty seconds to show the "Customer has hang up" message.

[mflorell]

well load is not the issue, try looking at the action_full.2006-XX-XX log file and compare the datestamp of the Hangup Action with when it was actually pressed.

Do you by chance have CentOS process prioritization enabled?

This has caused your exact issue for others in the past.

[mtancoff]

mmmm you killed me there!

I´ll try to find out

[mtancoff]

Matt I don't know if i expressed myself good.

When the agent is in an inbound or oubund call and the customer hangs-up, everygthing goes fine, I can see the asterisk debug executing VD_hangup.agi almost inmediately, that would mean that asterisk is receiving the hang-up, and executes the agi pretty fast.

But the agent keeps the call data in his interface, taking about half a minute to show the message "Customer has hang up, finish and disposition call"

I've been looking for process prioritization but coul'nt fin anything about in goolge, redhat or centos.

Thanx!

PD: We are going into production tomorrow, I'll post in the list when that happens.

[mflorell]

OK, that is a very different problem. All you have to do to get the display of the "customer hungup" message to display faster is to edit the vicidial.php script:
around line 4779, just change the "-30" to another value:
if ( (custchannellive < -30) && (lastcustchannel.length > 3) ) {CustomerChanneLGone();}


That is the counter that is used to determin when to display the customer gone screen.

[mtancoff]

thanx a lot!

[artimus]

After a customer hangs up on an agent we drop about 3 calls before the agent gets the message saying the customer hung up. The agent can see the other caller drop out but then has to manually click hangup.

I tried to change custchannellive -30 to -10 in vicidial.php. That didn't have much effect.

[mattyou1985]

After a customer hangs up on an agent we drop about 3 calls before the agent gets the message saying the customer hung up. The agent can see the other caller drop out but then has to manually click hangup.



I tried to change custchannellive -30 to -10 in vicidial.php. That didn't have much effect.

wils i did I tried to change custchannellive -30 to 1 in vicidial.php. and that had a better effect but not the result i was looking for ide like it so when a client hangs up we get a blep tone of sorts just like the goautodial 3.3 version as that duz it right a way and post a box on the screen saying you need to disbo the call go back and hang up <<< that for me is just ausom and very disappointed that VERSION: 2.12-557a BUILD: 160517-1927 has not got that and now i see that ime having to hard code it in,
if eney one nows of a way that can help us get that result we would be very happy

[mattyou1985]

ok after 2 reboots and a nother test call i can confirm that changing

( (custchannellive < -30) && (lastcustchannel.length > 3) ) {CustomerChanneLGone();}

to

( (custchannellive < 1) && (lastcustchannel.length > 3) ) {CustomerChanneLGone();}

i all so changed this just to see what hapends so ither or both changes have given me the results i need

(custchannellive < customer_gone_seconds)

to this

(custchannellive < 1)


has the affect of 1 sekend after custamer hangs up a sound is played to let you no that thay have gone now i am happy