Manual dial stats corrupted
Posted: Wed Nov 09, 2016 7:23 pm
Hi Everyone,
First of all here is my installation and server:
- Single server installation
- Vicidial from scratch on Debian 7 (I know it's better use Vicibox, but it was customer decision)
- VERSION: 2.12-563a
- BUILD: 160801-2119
- Asterisk version: 1.8.23.0-vici
Server:
- 2x Intel(R) Xeon(R) CPU E5620 @ 2.40GHz
- 48GB RAM
- 4x SATA drives (Raid10)
- 2 campaigns set to autodial with inbound (up to 16 agents), which are using GSM gateway.
Everything was working correctly for 2 months with autodial - simple outbound, inbound.
Since Monday there are few single user campaigns set to manual/inbound dialing.
What we have discovered there is a problem with statistics for users who dial in manual mode.
E.g.
1.
- Information from "Agent performance detail": Agent 1: 9 Calls Customer Time: 13 minutes 23 seconds
- When I go to user stats and listen recordings I noticed that these 9 calls last over 2 hours. Manager confirmed this agent was almost all the time during the call.
2.
- Information from "Agent performance detail" before running AST_cleanup_agent_log.pl --last-24hours by cron : Agent 2: customer time: over 3,5 hours
- Information from "Agent performance detail" after running AST_cleanup_agent_log.pl --last-24hours by cron : Agent 2: Customer time: 5 seconds
- User was lagged once.
I was comparing this installation from scratch to Vicibox 7.0.3 and I can say all working screens are the same, all cron jobs are the same.
I didn't notice any overload on server. It was working at load average: ~2.0, disks read/write was lazy.
Does anyone faced similar issue in the past? Maybe someone knows what else I can check, or knows the answer what causes this discrepancy
First of all here is my installation and server:
- Single server installation
- Vicidial from scratch on Debian 7 (I know it's better use Vicibox, but it was customer decision)
- VERSION: 2.12-563a
- BUILD: 160801-2119
- Asterisk version: 1.8.23.0-vici
Server:
- 2x Intel(R) Xeon(R) CPU E5620 @ 2.40GHz
- 48GB RAM
- 4x SATA drives (Raid10)
- 2 campaigns set to autodial with inbound (up to 16 agents), which are using GSM gateway.
Everything was working correctly for 2 months with autodial - simple outbound, inbound.
Since Monday there are few single user campaigns set to manual/inbound dialing.
What we have discovered there is a problem with statistics for users who dial in manual mode.
E.g.
1.
- Information from "Agent performance detail": Agent 1: 9 Calls Customer Time: 13 minutes 23 seconds
- When I go to user stats and listen recordings I noticed that these 9 calls last over 2 hours. Manager confirmed this agent was almost all the time during the call.
2.
- Information from "Agent performance detail" before running AST_cleanup_agent_log.pl --last-24hours by cron : Agent 2: customer time: over 3,5 hours
- Information from "Agent performance detail" after running AST_cleanup_agent_log.pl --last-24hours by cron : Agent 2: Customer time: 5 seconds
- User was lagged once.
I was comparing this installation from scratch to Vicibox 7.0.3 and I can say all working screens are the same, all cron jobs are the same.
I didn't notice any overload on server. It was working at load average: ~2.0, disks read/write was lazy.
Does anyone faced similar issue in the past? Maybe someone knows what else I can check, or knows the answer what causes this discrepancy