Active Status while still on a call

[pandzkilla]

Hi Everyone,

Need your expertise regarding this issue.

We have agents on ACTIVE STATUS while being on a call, do you know what caused this?

https://postimg.cc/vgs5ch63

Thanks

[jamiemurray]

Could be network issues, server load or agent workstation load. There is a chance if using chrome that Javascript throttling could be causing it too.

[martinch]

Hey, welcome to the forums!

Just seconding what jamiemurray said above...it's likely one of those things. A few things from my end;

- From your signature, it looks like you're using an older version of ViCi (2017?) so you may want to consider upgrading to access a more stable base system.
- If you're using Google Chrome, be aware that Google is driving for energy efficiency with it's browser and so JavaScript Timer Throttling is common across versions high than Chrome 85 (I think it was introduced in that one...happy to be corrected there). This means ViCiDial's heartbeat will be slowed at various intervals depending on your version causing agent and ViCiDial disconnections. It has been discussed at length here -> viewtopic.php?f=2&t=37098 If your version is from 2017, you'll need to upgrade to something newer in 2020 as Chrome Throttling counter-methods were introduced in that year. Also, be aware than Chromium-based browsers also may have their own version of JavaScript Throttling (Edge calls it "Energy Efficiency Mode") and Chromium itself throttles. ViCiDial currently has 3 unique counter-throttling measures and a further 2 measures have been proposed.
- If you're not using Google Chrome or not able to upgrade, go for a browser like Firefox. Feature rich and does not throttle as far as I know.
- In terms of network issues, run a ping between your ViCi server and some of your agents just to see if the ping is respectable. I've found ViCi operates well with pings less than 500ms. Anything above that, you may see this kind of inaccurate information with Agents View as well as other anomalies.
- In terms of the workstations, it may be worth checking them to see what the load is like, especially if the agents are using thin clients. You may have security software running or whatnot that could be swamping it. I used to work with thin clients with only 2GB of RAM back in the day and if you're using really, really thin clients like that these days, it may present this kind of problem.

The way the problem manifests on ViCiDial side is as follows;

- ViCiDial auto dial engine is constantly monitoring the database for changes of agent states as it requires that in order to calculate how many phone calls to place.
- If any of the above symptoms are experienced on the agent side for a period of more than 30 seconds, the auto dial engine will mark that agent as "gone" in the form of entering "LAGGED" into `vicidial_live_agents.sub_status`. You can verify in Admin Panel > Reports > LAGGED report.
- Not only is the sub status updated, the engine will alter `vicidial_live_agent.status` from `INCALL` (true status) to `PAUSED` (which is false). The call will still be happening as the call is being handled by Asterisk.
- The live agents status will remain paused until the call is dispositioned.
- All functionality with ViCiDial that queries the state of agents (using `vicidial_live_agents`) will be fooled into thinking that agent is paused and could potentially lead to wrong conclusions.

The problem is linked in with this problem reported by another user here (and many related posts) -> viewtopic.php?f=4&t=41269 albeit in a seperate part of the ViCiDial stack, the problem eminates from the same issue. I am going to be providing a patch for that problem and will include this problem in the same patch. Essentially, any part of the stack that is referring to `vicidial_live_agents` table status and giving agents information about other agent states. These ideally need to be accurate even in difficult operating conditions just so the customer experience isn't affected in a negative way. You can track the status of the patch here -> https://www.vicidial.org/VICIDIALmantis ... id=0001356

As for short term fix here, without any code changes, you may want to use the chat option so agents may confirm their availability with each other. That may be difficult to negotiate though but I'll have a think about ways to tighten up that part of the transfer screen. In theory, if you try to transfer to those agents, the transfer should be blocked. I'll test that out and file a patch on Mantis if required.

Hope this helps.