vicidial.org

OK so here is the deal, I currently have multiple web/asterisk servers running off a single mysql server.

I have installed (full install) another vicidial to act as the asterisk server for the web servers.

I am now able to log into the agent interface though when I do I get a time synchronization error and my softphone does not ring.

I have created conferences in the web server for the asterisk server (192.168.1.100) and I have added the asterisk server to the web server and vice versa.

I have changed the telnet address for the servers accordingly.

for the cron password I have tried 1234 and test

I am also able to register my softphone to the asterisk server and can make calls through the softphone itself. The agent interface is where the problem is arising.

I also set the ntp clients in asterisk and web servers to the database server.

I have all servers running in a citrix virtual environment and making calls with the web/asterisk + DB server has been working well, Now it is at a point to cluster all three servers to further balance the load and allow more users.

Is there a question in there somewhere?

what instructions/method did you use for your cluster/multiserver setup?

please also describe the virtualness of the "environment"?

do you mean a virtual private network, or virtual servers or ...?

Sorry I tried to include everything I can think of..

The servers are virtual servers, inside the network the servers all communicate on a 10.10.11.X subnet while all internet traffic is designated to 10.10.10.X subnet. So the DB server does not have internet access.

I have been using the vicidial cluster manual http://www.poundteam.com/downloads/Vicidial%20Multi-Server%20Manual%20by%20Poundteam%20v1%201.pdf

We also have the paid version of the manager manual.

I guess my question is how to get asterisk to call my softphone when logging into the agent screen.
As soon as I log in I get the time synchronization error even though I have the three servers running off the DB time source.

Just want to be clear that running HTTP/Asterisk on one server and sql on another is working fine... But when I try to create a cluster with all three I get this error. And the asterisk server is a basic goautodial CE 2.1 fresh install

time sync error is a misnomer. it normally indicates that your agent screen and the web server are losing packets between them or that there is a configuration issue with your cluster.

which brings me to:

what instructions/method did you use for your cluster/multiserver setup?

Then, of course, you (hopefully) are simply doing this for a TEST (which is more easily accomplished within VMWare, but Citrix is fine, I guess). Because you CANNOT use vicidial in a virtual server environment in production (at least not for more than one user).

Right now we are currently hosting server for six or seven clients and will be expanding when we acquire our new servers. so far this system has been running very well... but of course resources are a major issue when going this route.. The reason I am trying this now is actually for a test. I plan on implementing a dedicated asterisk server and running HTTP / sql on the virtual servers so I may use a timing device.

The only real resource I have found is the one mentioned above (vicidial cluter.pdf) is there some better documentation you could point me to?

--EDIT---

I am an idiot... just saw the reference to the docs directory..

Some very valuable resources for everyone:

/usr/share/astguiclient has lots of scripts. each can be READ with nano AND can be executed with "--help" after it (perl scriptname.pl --help) to get a simple list of options.

/usr/src/astguiclient/VERSION/docs has documentation and is easily overlooked

http://wiki.vicidial.org/index.php/MultiServer has instructions for multiserver installation and there is also http://www.poundteam.com/downloads/Vici ... v1%201.pdf (written by Poundteam, of course!)

you will likely find that a dedicated sql server and web server are also required. LOL (You've been warned.) If you do go the "virtual" route, please share your experience WITH DETAILS (ie: Vicidial Cluster with virtual web server with XX agents on the web server ...). Useful virtual instances are not to be ignored, but so far WE haven't found a useful virtual system for any of our clients (except for the CRM, of course).

Right now we are running roughly 10 goautodial HTTP/Asterisk servers, each connects to a seperate mysql VM.

We are running Citrix XenServer on 3 physical machines and all run off a dedicated NFS server for storage. The one problem we have with expanding over 15 users or so is that the virtual machines do not run in PVM mode so they do not have xentools installed. When we attempted to create a seperate centos VM with asterisk we could not get the dahdi module to work correctly. This is definitely still a work in progress though I do believe we are onto a solution for many small call centers.

Today we are attempting to bring up the largest room yet which is 20+ seats and we have been working on the network (segregating traffic, aggregating links, and upgrading switches) so this is by far the best test yet.

I have read the resources and believe I have everything set up correctly though asterisk still is not calling my soft phone when I log in....

This may be a stupid question but... When I register my softphone do I register to the asterisk server or the HTTP server?

um ... 15 users on each server? with autodial? or 15 users across 10 servers?

HUGE difference! LOL

but ... if you have this working with all these users ... how is it that you do not know which server to register to? and i do not see a reference to your having db / web / dialer in sep servers ... just db separated. hm.


anyhow: phones register to dialers, not web, unless that machine is both.

log in and use asterisk -R to view registration attempts.

It is 15 users per server. And right now I have web/asterisk on one vm for each callroom. Those servers connect to the database server.

Right now I am trying to move asterisk to its own dedicated VM (to be moved to physical machine in production)

So far I have the web server communicating with the asterisk server. the asterisk server is communicating with the web and sql servers, and allows me to login. Though when I log into the agent interface I get time synchronization errors instantaneously.

I was hoping somebody with a physical cluster may have seen the same results and would know what I am missing in building the cluster.

I have read the poundteam PDF and the /usr/src/astguiclient/conf/LOAD_BALANCING.txt

still getting the same result.

Asterisk shows the softphone registering, and I am able to make phone calls straight through counterpath (eyebeam) with no problems.

time synchronization errors instantaneously

bad network connection, dropping packets (unless of course you actually don't have ntp set up, in which case you have a time sync error! LOL). it is also possible/likely that even with ntp these virtual servers cannot keep time to the degree necessary to support Vicidial.

But i have a question about this entire scenario:

if you have 15 users per server in production already ... how is it that you cannot log in without a time sync error ... ? obviously you do not have this issue with those other 15 users on all those other servers ...?

Yeah the other servers (Web/Asterisk + SQL) run just fine... its just for some reason that i cannot seem to get all three parts on different virtual servers.

BTW hardware is as below... just realized I am missing this from the description.

We are running on 4 servers with...

Intel(R) Xeon(R) CPU 5160 @ 3.00GHz
CPU Speed 2.99 GHz
Cache Size 4.00 MB
16GB memory
2X 1GB Network cards

One server does have 8GB of memory and that is mainly used for development (interface updates and such)

Our NFS server is also the same as above with 8GB of memory and 1.5TB of drive space on a mirrored raid.


Right now I have the web server and apache server on the cluster set up synchronizing time with the DB server. When I check time on all the servers it shows the same time and ntpq -p shows that all the servers are synchronizing.

I am at the point now where I do not know where/what else to look at, It would seem that if this can be done with physical machines I should at least be able to get them to connect in a virtual environment.

And as for the

Bad Network Connection

I had thought about that, I can ping all servers within the cluster and have under .5 response time and I have no dropped packets on any of the servers.

I really appreciate the help

OK: Lemme see if I have these pieces:

The servers are virtual servers

We are running Citrix XenServer on 3 physical machines and all run off a dedicated NFS server for storage.

Right now we are running roughly 10 goautodial HTTP/Asterisk servers, each connects to a seperate mysql VM

It is 15 users per server. And right now I have web/asterisk on one vm for each callroom. Those servers connect to the database server.

Yeah the other servers (Web/Asterisk + SQL) run just fine...

Intel(R) Xeon(R) CPU 5160 @ 3.00GHz
CPU Speed 2.99 GHz
Cache Size 4.00 MB
16GB memory
2X 1GB Network cards

Now what I do not see ... is which instruction set you used to 'cluster' these machines? (by these, i refer to the new 3 server virtual cluster you are attempting ...)

The three servers I am attempting to cluster are all VMs running in Citrix.

Right now I do not have any physical machines I am building a cluster with. I guess best way to describe it is that I want to get this done virtually before I attempt this on any number of physical machines... I will post a screenshot once I get to my windows PC that may shed some light on the situation

-----EDIT----
Ok here is the pic.. I am trying to cluster the three highlighted machines.. Even though this may not help resolve anything I hope it helps to at least describe our situation and show others that it is possible.... though it is a lot of work and we have learned that we obviously will not be able to buy TOO POWERFUL of machines to host the dialer for larger rooms...

The whole reason I attempted this is due to the fact that bossman wants to bring up a room in Cali with 200 + seats... So I figured I should get started on at least figuring out how to set a physical cluster

Now what I do not see ... is which instruction set you used to 'cluster' these machines?

also i note that one of my earlier questions with a technical point went unanswered as well: are the 15 agents in each of the virtual goautodial servers on Outbound campaigns or Inbound?

williamconley wrote:

Now what I do not see ... is which instruction set you used to 'cluster' these machines?

also i note that one of my earlier questions with a technical point went unanswered as well: are the 15 agents in each of the virtual goautodial servers on Outbound campaigns or Inbound?

I have been using the vicidial cluster manual http://www.poundteam.com/downloads/Vici ... v1%201.pdf

I also used /usr/src/astguiclient/conf/LOAD_BALANCING.txt which you had pointed out earlier.

The agents are Inbound, Outbound, and Blended. Though most of the agents are outbound auto dialing, with forced recording on all calls.

knotbeerdan wrote:

I have been using the vicidial cluster manual http://www.poundteam.com/downloads/Vici ... v1%201.pdf

I also used /usr/src/astguiclient/conf/LOAD_BALANCING.txt which you had pointed out earlier.

The agents are Inbound, Outbound, and Blended. Though most of the agents are outbound auto dialing, with forced recording on all calls.

excellent, thanks for the information. now back to the time sync error.

the error technically stems from the lack of the update of the live agents table's random number field. check a "working" against a "not-working".

The agent screen sends an ajax data request every second. the answering php file (in addition to sending the response) also updates a value in the live agents table to signify "still live!". Without this, your system fails.

That is my guess of your problem. Easily verified, of course, since you have both working and non-working available to you.