AWS VICIdial Cluster Configuration.

I have a multiserver vicidial cluster setup on AWS which encounters a handful of issues once upwards of 50+ simultaneous users are logged in and autodialing 100+ lines. In fact, the odd part is this is the only point at which the system begins to demonstrate the error (the reason i'm making this post).

Our current configuration is the following:

1 - DB instance

1 - Primary calling server A (which is configured to include 2 load balancing only asterisk instances. 1/3 of agents register softphones and use this as a webserver)

2 - Primary server A load balancing dialer instances ( agent webphones are registered on calling server A. These two instances do not have registered SIP softphones for agents.)

1- Secondary calling server B (which does not part of the load balancing. 1/3 of agents register soft phones and use as web server)

1 - Third calling server C (which does not part of the load balancing. 1/3 of agents register soft phones and use as web server)

all using:

VERSION: 2.14-715a

BUILD: 190705-1012

© 2019 ViciDial Group

To describe the issue as best as possible: inbound/outbound blended calling is seamless with 40 users registered and taking calls while the asterisk is ratio calling between 70 to 100 lines simultaneously. The issue is triggered by the following: 20 additional users are registered and ready for calls and 120 to 150 lines are dialing. There is a point which agent wait time begins to increase from seconds to minutes between ACD. The UI will display 50 agents ready with dialing 150 numbers and yet, a handful of calls are served every minute translating to 3 to 5 minutes between call distribution.

Consider the following changes and still encounter the same issue:

- I have tried using multiple SIP carriers

- increased and decreased trunks per server and calls per second

- optimized the DB settings for high load dialing

-had all agents logged into a single and several dialer instances with the same issue

-increased server size ( we are using T3-2xlarge on all instances and doesnt even max at 15% utilization )

-Tried variations of Max FILL grouping and trunk caps with and without LB

Hopefully i can find somebody whos more capable with cluster systems using vicidial. I have no other ideas on how to trouble shoot this.

Amazon Web Services Asterisk PBX System Admin Linux

Project ID: #23914369

About the project