Duplicated Events vicidial_agent_log

[gjorge]

Regards!!

I'm having problems with the table vicidial_agent_log, in the same I'm being duplicated precisely the event pause causing this changes in the times of the agents.

Specifically what happens is that when an agent is placed on pause and selects a descant substatus (not billable) it appears with the pause time at 0 and then another "fictitious" event appears with the time of pause in wait_time, this is one of the cases, another case is the same stage but the duplicate event shows the time in pause and the substatus null.

VERSION: 2.14-621a
BUILD: 170717-1444
© 2017 ViciDial Group
virtualized Cluster

[mflorell]

The vicidial code doesn't handle billable or non-billable pause time any different from each other, so that part doesn't make much sense to me.

How frequently does this issue occur?

Can you post actual examples of the log entry issues?

We have confirmed that there are several known logging issues that tend to happen on VICIdial systems that are on under-powered or over-utilized equipment. Virtualized systems are especially susceptible to those issues.

[gjorge]

Hi Matt!!

How can I post a log file?

can I share the agent's full day so you can see the detail?

[mflorell]

I was thinking you would just post the two duplicated log entries here in text.

[gjorge]

Here is an example of what happens with events:

event_time lead_id campaign_id pause_epoch pause_sec wait_epoch wait_sec talk_epoch talk_sec dispo_epoch dispo_sec sub_status
3/22/2018 14:09:06 NULL GEST0001 1521742146 0 1521742146 12 NULL 0 NULL 0 3
3/22/2018 14:09:06 NULL GEST0001 1521742158 2134 1521742158 0 NULL 0 NULL 0 NULL

explaining what happens, the agent selected the code 3 in pause code and the seconds belonging to this pause were counted in the next event where they must have belonged to the event of the pause code.

I tried to recreate these cases to see what is wrong but I could not find the cause, please if you can help me.

[mflorell]

How frequently does this issue occur?

[gjorge]

Hi Matt,

this happens daily with all the agents randomly.

I have been doing tests and I notice that when I select 'next call pause' an event is written in the database with NULL values in this event it is written about the pause code and when the agent finishes with the pause time then a following event is created with time pertaining to the code.

[mflorell]

Out of how many calls handled does this happen how many times in a day?

For instance: "Agents handled 53,000 calls yesterday and this happened 23 times...".

Can you reliably replicate this issue happening?

[gjorge]

Hi Matt!

Agents handled 1,902 records and made 4,135 calls yesterday and this happened 15 times.

I have noticed the following when the campaign is manual:
agent select next call pause in active call and when the call is dispo the box appears to select the pause code, the system creates a null event and over this it writes the pause time and in another event writes the pause code.

This does not happen when the campaign is predictive.

[mflorell]

Yes, pause entries work different in manual dial campaigns because they have to, there really is no "paused" state in manual dial mode because the agent is never "active/ready" either, so there had to be a way to denote pause codes and pause time for those agents.

Can you reliably replicate the duplicate vicidial_agent_log problem, or is it random?

[gjorge]

It's random Matt, I'm going to study the source code of writing logs to see their operation from the most intrinsic.

I am new working with vicidial and I still have a lot to learn.

thank you.

[gjorge]

Greetings team!

I am writing to tell you that I found the reason why some pause times were not written in their corresponding event in the vicidial_agent_log table, this is due to the JavaScript asynchrony!

It happens when agents click the pause button several times waiting for a response from the agent interface for some brief slowness that the platform may present, this scenario not only happens with pauses, it also happens with the buttons that commit to the database.

I solved it with promises.

[williamconley]

Greetings team!

I am writing to tell you that I found the reason why some pause times were not written in their corresponding event in the vicidial_agent_log table, this is due to the JavaScript asynchrony!

It happens when agents click the pause button several times waiting for a response from the agent interface for some brief slowness that the platform may present, this scenario not only happens with pauses, it also happens with the buttons that commit to the database.

I solved it with promises.

have you tested this against the latest svn code for Vicidial? If so, this sounds like something you should put in the Issue tracker. A single precise (button for button, including timing if that's important) example that will duplicate the flaw reliably and it can be fixed.

[Arjene]

We have the same issue


Version: 2.14b0.5
SVN Version: 3204
DB Schema Version: 1588

VERSION: 2.14-741a
BUILD: 200318-1330