DEAD and INCALL at the same time

[ryanhs]

I have had an on-going issue with active calls showing dead in real-time report and 'call hungup' on agent screen. Because of this problem I cannot use "Dead Call Max Seconds" which I would like to use and it also causes problems with agent productivity reports. The calls will remain active with a dead status for any duration until the agent hangs up the call.

This was a manual outbound call without any transfer from 11:13 to 11:28 and the majority of the call was reporting as dead. The call appears normal in agent_log and vicidial_log

The Agent Debug Log looks like this during the call:
| 2028| 2021-06-24 11:28:10 | 2021-06-24 11:28:10 | 0.003239 | conf_exten_check.php | refresh| 0| INCALL|10794936| DEAD:M6241113110002173968|2173968|0|
| 2028| 2021-06-24 11:28:09 | 2021-06-24 11:28:09 | 0.003745 | conf_exten_check.php | refresh| 0| INCALL|10794936| DEAD:M6241113110002173968|2173968|0|
| 2028| 2021-06-24 11:28:09 | 2021-06-24 11:28:09 | 0.003591 | vdc_db_query.php | AGENTSview | 0| 0|
| 2028| 2021-06-24 11:28:08 | 2021-06-24 11:28:08 | 0.003182 | conf_exten_check.php | refresh| 0| INCALL|10794936| DEAD:M6241113110002173968|2173968|0|
| 2028| 2021-06-24 11:28:07 | 2021-06-24 11:28:07 | 0.003265 | vdc_db_query.php | AGENTSview | 0| 0|
| 2028| 2021-06-24 11:28:07 | 2021-06-24 11:28:07 | 0.003964 | conf_exten_check.php | refresh| 0| INCALL|10794936| DEAD:M6241113110002173968|2173968|0|
| 2028| 2021-06-24 11:28:06 | 2021-06-24 11:28:06 | 0.003748 | conf_exten_check.php | refresh| 0| INCALL|10794936| DEAD:M6241113110002173968|2173968|0|
| 2028| 2021-06-24 11:28:05 | 2021-06-24 11:28:05 | 0.003727 | vdc_db_query.php | AGENTSview| 0| 0|
| 2028| 2021-06-24 11:28:05 | 2021-06-24 11:28:05 | 0.003728 | conf_exten_check.php | refresh| 0| INCALL|10794936| DEAD:M6241113110002173968|2173968|0|
| 2028| 2021-06-24 11:28:04 | 2021-06-24 11:28:04 | 0.003424 | conf_exten_check.php | refresh| 0| INCALL|10794936| DEAD:M6241113110002173968|2173968|0|
| 2028| 2021-06-24 11:28:03 | 2021-06-24 11:28:03 | 0.003201 | vdc_db_query.php | AGENTSview| 0 | 0|

Then when its hung up:
| 2028| 2021-06-24 11:28:10 | 2021-06-24 11:28:10 | 0.012769 | vdc_db_query.php | manDiaLlogCaLL | 2173968 | end
| 2028| 2021-06-24 11:28:10 | 2021-06-24 11:28:10 | 0.00245400 | manager_send.php | Hangup| 0| CALLHANGUP IAX2/127.0.0.1:40569-14014 10.10.10.10|

Unfortuntely I didn't catch the begining of the DEAD status on this call because I enabled the Agent Debug while it was happening. i still wanted to post while I wait for another example with the Agent Debug turned on for user 2028. I figured the fact the call is INCALL and DEAD at the same time is enough to start with. When I capture the beginning of the events I will append it to the post if needed.

Additional info:
I have tried disabling an agents web proxy in windows 10 and I still have the same problem but here is what conf_exten_check.php looks like in my squid proxy log:
1624635758.816 14 192.168.14.223 TCP_MISS/200 727 POST http://pbx.mydomain.com/agc/conf_exten_check.php - HIER_DIRECT/10.10.10.10 text/html

TCP_MISS The requested object was not in the cache.*

For the time being I am wondering if someone can point me down the right path to what I should be looking at to diagnose this.
There are some similar posts in the past usually talking about dead calls on real time report during active calls but I wasn't able to find anything very helpful for my situation except something similar with 3-way calls.

Thank you in advance for any help with this


-----------------------------
VERSION: 2.14-694a
BUILD: 181005-1738

Cluster 1 core + 2 dialers
SuperMicro board and chassis

Core DB+HTTP Server:
openSUSE Leap 42.3
Intel Xeon CPU E5-2620 v4 @ 2.10GHz - 8 core 16 thread
16GB RAM
SSD RAID

Dialer1:
openSUSE Leap 42.3
Asterisk 13.21.1-vici
Intel Xeon CPU E3-1220 v3 @ 3.10GHz - 4 core 4 thread
8GB RAM
HDD RAID

Dialer2:
openSUSE Leap 42.3
Asterisk 13.21.1-vici
Intel Xeon CPU E3-1220 v3 @ 3.10GHz - 4 core 4 thread
8GB RAM
HDD RAID

LAN + Internet:
Mitel 6865i Phones (latest firmware)
Phones on separate VLAN than PC's
200 Mb fiber
bandwidth.com sip trunks
2ms ping from call center to vicidial servers in a colo
Agents are behind a Squid Proxy that is located on the LAN

Typically 10 active agents at a time on each dialer
Dialers reboot nightly, web/db reboots weekly

[GenXOutsourcing]

Is it ALL agents? a few agents? or 1 agent?

There is a connectivity issue, Vici does not see that your agent is still in the call....... only that the customer is still there. I will bet that you also have agents incall, and they show in Dispo.

From my experience, and was having the same issue with homebased agents, change to webphone and use Viciphone2.1.

Now, I have 0 issues.

[ryanhs]

It's intermittent with all agents.

Thanks for the recommendation to use the webphone.
I opted to use hardware phones at a time when the agents were allowed to freely browse the web and stream video. Using the hardware phone allowed me to tag the phone with a VLAN ID separate from the PC plugged into the switch on the phone. The voip traffic uses a different uplink and internet connection. Now we are using the proxy and agents cannot browse the web.

[GenXOutsourcing]

With the webphone, i see 90kbps per agent or less. These are homebased agents in Philippines connecting to servers in US.