Cluster Performace, Looking for advice

[covarrubiasgg]

I just made a small Cluster with the following specs:

Cluster Setup

50 Agents - Adapt AVG Max Level 3, Available Only Tally Threshold on 2 NON Pauses Agents, Dial Level Difference Target: 2 Agents Waiting for call, Adapt Intensity Modifier: -5 (With this setup we always have Dial Level between 1.5 - 2.5)

Hardware
DB/Web: Xeon E3 CPU E3-1220, 8GB ECC RAM , Raid 10 w/ 4 x 128GB SSD 6.0 GB/s using a LSI MegaRaid 9240-4i
2 x Telephony: Core i5 , 4 GB RAM, 128 GB SSD 6.0 , no hardware timer

INSTALLATION METHOD: VICIBOX 6.0.2 | VICIDIAL VERSION: 2.10-451a BUILD: 140902-0816 | Asterisk 1.8.31.0-vici

Phone Aliasing Both Servers and using AMD. Full Recoring + compression

This is me more seeking for advice then reporting an issue.

This setup is working almost perfect, everything seems to be running great, but we have seen some minor issues with Agents getting calls and Vicidial not detecting ¨LIVE CALL¨ and some times they get 2 call in the same conference, Some others when the agent transfer the Call to a Closer, more than one Closer get the same Calls. The process we are using for transfer is Agent to Ingroup Consultative, So, they pick the ingroup, they check consultative and press Dial with customer, after the Closer have the call they Leave 3 Way.

This is minimal, we are dialing about 22,000 Calls per day and around 7,000 answered calls. They reported me about 7 calls a day with this issues, so 7/7,000 is a ridiculous small rate, BUT we intend to get another 2 Telephony servers and a dedicated web server in order to increase our campaign up-to 100 Agents. And i want to be prepare.

Why this kind of problems is happening? Is it the lack of a real timer on the dialer? Is it the lack of caching memory on the raid controller?

Just to let you know, the server load is minimum, all servers are working below 1 (0.50 to be more accurate)

The agent stations are time synchronized with the server

I hope you can help with you fine comments as always, thank you very much

[dspaan]

See this thread.
viewtopic.php?f=4&t=23656

I've also encountered it and never figured it out. I'm thinking it might also be workstation issues. Everything in your network has to be perfect, NTP, VLAN's, QoS, etc.

[covarrubiasgg]

Already check that the NTP is synchronized between the Windows Domain Controller and the Cluster Server.

We still facing the same problem, 2 Closer getting the same Call.

But now i realize that when the agents choose MANUAL DIAL, they go automatic on PRECAL PAUSE but sometimes they get an Auto Call in the green screen, but they don’t have enable the call controls and even worse if they press Go Back, they will be available and they get 2 Calls on the same session. Is there any way to prevent this?

[dspaan]

Are you using pause codes? What i've seen is that this can happen if you click pause and then want to click the pause code but a call comes in. I recommend disabling manual dial, it will only give you a headache with duplicate leads or leads getting lost or not having the proper data.