vicidial.org

Hi,

I am building a dialer with multiple clusters, we need to achieve 2000 concurrent calls. everything looks like it's working but i don't know if i am missing anything... the dialers are not spreading the laod evenly. vicidial seems like dialing on Some Nods and capping them out instead of spreading the traffic on all the dialers. at the moment this is the setup

1 DB machine
1 Web
4 Dialers ( i know this will not hit the 2k channels but should at least do 1200)

the problem i am having is that i get 2 dialers capping out and 2 more not really doing much. is there a config i missed somewhere ? i do have load balancing set to yes

The way i have it setup is that all agents login to 1 NOD ( at the moment 32 agents) and i put Max trunk on this nod low so it does not do much dialing.


VERSION: 2.7-401a
BUILD: 130508-2256

thanks
Mike

1) Welcome to the Party!

2) As you are obviously new here, I have some suggestions to help us all help you:

When you post, please post your entire configuration including (but not limited to) your installation method and vicidial version with build.

This IS a requirement for posting along with reading the stickies (at the top of each forum) and the manager's manual (available on EFLO.net, both free and paid versions)

You should also post: Asterisk version, telephony hardware (model number is helpful here), cluster information if you have one, and whether any other software is installed in the box. If your installation method is "from scratch" you must post your operating system and should also post the .iso version from which you installed your original operating system. If your installation is "Hosted" list the site name of the host.

If this is a "Cloud" or "Virtual" server, please note the technology involved along with the version of that techology (ie: VMware Server Version 2.0.2). If it is not, merely stating the Motherboard model # and CPU would be helpful.

Similar to This:

Vicibox X.X from .iso | Vicidial X.X.X-XXX Build XXXXXX-XXXX | Asterisk X.X.X | Single Server | No Digium/Sangoma Hardware | No Extra Software After Installation | Intel DG35EC | Core2Quad Q6600

3) Yes, the above is a "standard request", and I note that you did include the minimum required info (vicidial version with build). But this is a multi-server cluster and you're having load issues. I'd say it stands to reason you would want to supply all relevant data to get a good response.

4) Here is some more relevant data you may want to supply (in addition to the standard information, not merely the minimum, requested above ... all of which is possibly relevant!):
* From the top right area of the Campaign's Real Time Screen: Short/Fill
* From the Admin->Servers page for each dialer: Max Trunks / Max Calls Per Second / Asterisk Version (ensure it matches "asterisk -V" output from each server!) / ALL Vicidial Balance Settings
* Admin -> System Settings: Enhanced Disconnect Logging
* % of call in each category when "Show Carrier Stats" = YES in the Campaign Realtime View "Choose Report Display Options" (ie: Is there a lot of congestion?)
* Campaign settings for dial ratio and how many agents were in Ready Status
* Campaign Dial Prefix / Dialplan Entry that is in use on the carrier for each dialer (which would of course be activated by the aforementioned dial prefix)
Obviously you may consider playing with some of these settings if you have not yet ... but be sure to record what they were before you changed them. LOL

5) It is not a bad idea to have an outside Vicidial person look at your system.

6) Seriously consider spreading your agents among the dialers evenly to cause even dialing by having them register their phones to all four dialers (spread all phone registrations out 25% on each dialer). This will also allow each dialer to prefer to send an answered call to a local agent whenever possible which is handy. But mostly it will cause each dialer to want to dial out.

7) Log in to each dialer yourself and verify that they can all actually dial out when an agent is logged in directly. A dialer that does not work cannot be a "balance" dialer in the campaign for other dialers.

Happy Hunting!

Generally Diallers will prefer to make calls for agents who's phones are logged in to them directly so as William says, make sure you've evenly distributed your phone logins between your dialling servers. Also make sure you have set 'Balance Dialing' to 'Y' for each server in Admin -> Servers.

VERSION: 2.7-401a
BUILD: 130508-2256
Scratch Install on CentOS 5.10

Web Server: 4 x 3.6 GHz, 16 GB RAM, 70GB 15k raid1
DB server(5.0.95 x64 version): 8 x 3.00GHz, 32 GB RAM,300GB SSD raid1
4 x dialers(asterisk 1.8.21.0-vici-beta)

the rest of the info i will send them tomorrow as we had a outage on the carrier and the real time report might be off

Silly question: Why Scratch on CentOS instead of GoAutoDial?

There's the rest of the info:

Short/Fill = 352 / 352
Max Trunks / Max Calls Per Second = 250 / 15cps
Asterisk 1.8.21.0-vici-beta
Enhanced Disconnect Logging = 0
carrier congestion at 2.1%
16 agents logged in for this test period at 50 channels per agent running survey, getting 620 concurrent calls. most of the time 1/2 of them are on ready.
the idea of separating the agents on different dialers is not a option. the company is offshore it's not going to be possible to seperate every 8 agents on a dialer as lots of things change and the management of it going to be a pain in the ass. this is a upgrade to vicidial from freeswitch and if we need to do that then it's more of a downgrade.

thanks
Mike

williamconley wrote:Silly question: Why Scratch on CentOS instead of GoAutoDial?

is Goautodial install comes on centos OS ?

mikes wrote:

williamconley wrote:Silly question: Why Scratch on CentOS instead of GoAutoDial?

is Goautodial install comes on centos OS ?

! Obviously you are unaware of the GoAutoDial history. But in summary: YES.

Not only that, but there are instructions for goautodial being installed AFTER the OS (ie: Install stock CentOS and then use the goautodial system to get everything up and running). This is available (if I recall) on the goautodial Wiki.

But if you have the ability to install from .ISO, you should use Vicibox.com's .iso instead. Goautodial is excellent, however, if you have a CentOS requirement of any sort. Very stable, arguably as stable as the Vicibox OpenSuSE install.

Short/Fill = 0/0

This indicates that the system is NOT in need of Balance dialing. In other words: If short/fill are 0/0 the dialers without agents on them have no reason to dial. If there are no dialers which are "short" of trunks, the other dialers have no reason to "fill" that shortage by dialing for them. Reducing the available channels on the server with the agents (Admin->Server->Max Trunks) will cause that server to be "short" and request other dialers to "fill" for it.

the idea of separating the agents on different dialers is not a option. the company is offshore it's not going to be possible to seperate every 8 agents on a dialer as lots of things change and the management of it going to be a pain in the ass. this is a upgrade to vicidial from freeswitch and if we need to do that then it's more of a downgrade.

Properly configured with load balanced agents, Vicidial will handle the "agent X on dialer Y" portion for you. It requires special setup of the phones when you begin. You'll use "phone aliases" (Admin->Phones->top menu area for aliases) and create phones on 1-4 dialers for each agent. When they log in with the alias instead of a phone, the system will call them from the phone on the server with the least agents. Each agent phone must be capable of registering to multiple dialers. It is not necessary to have all agents registered on all dialers for this to be effective. Having them registered to two per agent will still help to balance.

Also worthy of note: In the over 1000 simultaneous channel range, you'll need lots of calls per second to keep the dialers up to speed. Your carrier may frown on this, so you should check to find what your Max Calls Per Second is before jumping. I've had systems in need of 200 CPS per dialer to be able to maintain 1500 lines with high-volume dialing. Many carriers will limit you to 5CPS or 30CPS depending on their internal restrictions unless/until you discuss it with them.

Ok that is a good option to split the agent if it does not require for the agent to change the IP on his eyebeam. i will look into that.

also for the short/fill i had edited the value it is at 352/352 i had typed 0/0 by mistake

as for the vicidial vs goautodial, i asked and basically it was just decision to do it this way ...

williamconley wrote:Also worthy of note: In the over 1000 simultaneous channel range, you'll need lots of calls per second to keep the dialers up to speed. Your carrier may frown on this, so you should check to find what your Max Calls Per Second is before jumping. I've had systems in need of 200 CPS per dialer to be able to maintain 1500 lines with high-volume dialing. Many carriers will limit you to 5CPS or 30CPS depending on their internal restrictions unless/until you discuss it with them.

The carriers i am using have me set for 100 cps but i can increase with no problem if i need to. now let me ask you if i have 4 dialers at 20cps for each... does that mean it's 80cps? or 20 in total?

Yep. Here's the HOWTO: http://goautodial.org/projects/goautodialce/wiki/64bit. GOautodial CE is based from CentOS since the very beginning (circa 2006 as VicidialNOW).

Not only that, but there are instructions for goautodial being installed AFTER the OS (ie: Install stock CentOS and then use the goautodial system to get everything up and running). This is available (if I recall) on the goautodial Wiki.

mikes wrote:Ok that is a good option to split the agent if it does not require for the agent to change the IP on his eyebeam. i will look into that.
also for the short/fill i had edited the value it is at 352/352 i had typed 0/0 by mistake

Agents would require registering their eyebeam one time to all servers for load balancing (after you created the phone alias setup for them). But when they log in, they just log in. No decisions. Vicidial chooses for them. After the setup there's nothing to do. But: If one of the dialers goes offline, those agents can log right back in and go to work. Not even the manager has to do anything! Very handy. (Dead dialer can sit there being dead until IT gets back from lunch ... reduced overall "dialout" capacity is the only side-effect. And if you have over-supply of that, it may not impact dialing at all!)

If short/fill is 352/352 then the system is running optimally. It stopped dialing when it filled the short trunks. Didn't need another dialer to dial more channels, so they didn't.

gardo wrote:Yep. Here's the HOWTO: http://goautodial.org/projects/goautodialce/wiki/64bit. GOautodial CE is based from CentOS since the very beginning (circa 2006 as VicidialNOW).

Not only that, but there are instructions for goautodial being installed AFTER the OS (ie: Install stock CentOS and then use the goautodial system to get everything up and running). This is available (if I recall) on the goautodial Wiki.

Undeniably a better idea than a scratch install. But a step above that is the Goautodial ISO. And a step above that is the Vicibox ISO! (gardo and I disagree on the order of those last two, but that's expected as we're both prejudiced, LOL)

mikes wrote:

williamconley wrote:Also worthy of note: In the over 1000 simultaneous channel range, you'll need lots of calls per second to keep the dialers up to speed. Your carrier may frown on this, so you should check to find what your Max Calls Per Second is before jumping. I've had systems in need of 200 CPS per dialer to be able to maintain 1500 lines with high-volume dialing. Many carriers will limit you to 5CPS or 30CPS depending on their internal restrictions unless/until you discuss it with them.

The carriers i am using have me set for 100 cps but i can increase with no problem if i need to. now let me ask you if i have 4 dialers at 20cps for each... does that mean it's 80cps? or 20 in total?

Four at 20cps is 80cps total. This setting is applied Per Server. The servers do not take each other's calls per second into account. No method to do so.