Vicidial Taking Over all Calling when Dialing

[wizzbangca]

I have my vicidalnow box iax trunked to a trixbox iax extension. Beofre I start letting people use the auto dialer, incomming and outgoing calls work great in trixbox. Once the auto dialer starts, all inbound calls and manually dialed outbound calls using trixbox do nothing but ring. Trixbox logs show no incomming or outgoing during vicidialnow dialing. Vicidial logs show dialing out and people picking up. Our pri line has 23 lines. Even if vicidial is using 4 lines, no one can dial in and none of the extensions can dial out, except for vicidial.

Any ideas?

Thanks again everyone.

[marcin]

limit number of channels in server setup from default 96 to number you want

[wizzbangca]

Thanks for the reply. I have been playing around with vicidial and trixbox trying to track down where this problem is coming from. I set the number of available lines in the trunk to 16, the server max lines is 16, and the max dials per second in 12 for both. Still the same problem.

When I run just trixbox and not vicidial, the lines seems to stay available and incomming and outgoing calls clear the line when hung up. This may actually be a vicidialnow problem somewhere, just not in the number of available lines in the trunk or server.

[williamconley]

Have you tried setting the max lines to ONE and running a test campaign?

VicidialNOW should not be able to hammer the Trixbox engine like that, BUT the overhead in Trixbox is extremely high.

I dumped my first trixbox (very educational, excellent software!) when my first situation like this occurred. This was because Digium (with whom I had a service contract at the time) refused to touch the box.

I now use FreePBX on many machines ... some in Vicidial boxes and some "connected" as you are. Without Trixbox, these problems cease to exist in the extreme (FreePBX still has some overhead, but nothing compared to Trixbox).

[wizzbangca]

This is definitely a trixbox issue. But, asterisk is reporting the pri gets congested at random times, requiring the box to be restarted. This could be an asterisk problem or a trixbox problem. I have tried asterisk 1.2 and 1.4 and still get the same problem. As far as I can tell, asterisk is not passing any errors the pri sends to the box. I'm up for guess and try until i figure out what is causing the problem. Freepbx will have the same problem if asterisk doesn't know how to prevent congestion.

[williamconley]

when i dumped trixbox, it was for a similar issue. turned out to be in the trixbox code somewhere as the issue went away (and never returned) when i dumped trixbox. i manually installed freepbx and sugar (and have since installed a2billing as well on a few) and have never experienced the oddities that came with trixbox. on any other machine. and i've built a lot of them.

and of course, i only dumped trixbox because digium paid me $1000 to do it. (they gave back the service contract money because they do not support trixbox ... but after i got rid of trixbox, i never needed digium again, so i never got another contract just kept the refund ... go figure)

[wizzbangca]

I tried installing freepbx from scratch since a lot of the ISOs floating around aren't only asterisk and freebpx. The installation documentation on freepbx's site has severe problems with it and is definitely not being maintained. I ran into 3 problems before I even got a chance to install freepbx itself. So, I have to scrap that idea and stick with trixbox. Also, features that we need that come in trixbox aren't even easily available in freepbx, or may not be available. Trixbox has the most features out of the box ready to go, that is just a plan pbx. There's no fluff like crm or other utilities that aren't needed.

[williamconley]

Not to turn this into a babbling idiot conversation, but I am curious: what part(s) of trixbox that are NOT FreePBX do you find that you actually need?

Seriously, as we integrate boxes all day long and I would like to know if I'm missing something that call centers ... need.

[wizzbangca]

I have been looking into why vicidial stops dialing after few calls. Vicidial is set to dial 5 calls per second with a max of 16 lines. That isn't where the problem lies. I can't try freepbx right now, DAHDI drivers for Rhino cards aren't exactly easy to install. I dug into live monitoring while emulating a dialing environment and found out the pri is going congested with the d-channel going down before a bunch of SAMBE mesasges. I have been spending the last 3 weeks trying to get rid of the congestion. No luck. AT&T, not surprisingly, doesn't want to help figure out where the congestion is being caused. Rhino, the manufacturer of the t1 card, looked into what was going on. They setup the zaptel configs to what they use for AT&T voice pri. Still get congested. Rhino is at a loss since all their known working settings aren't working for me. Since asterisk is coming up with congestion, even freepbx may end up with the same problem. For now, I am sticking with trixbox.

This project is close to being scrapped, along with my job. Countless hours of searching have turned up nothing for the correct settings to connect to an AT&T pri. I have smashed into a brick wall and have no idea what to do.

[mcargile]

Which system has the PRI plugged into it? The Vicidial system or the Tribox system. If it is the Trixbox system this is a very bad idea and most likely your problem.

When Vicidial dials out it goes through roughly 3 or 4 steps in it's dialplan before sending the call to the provider. Trixbox (mainly the FreePBX on it) goes through somewhere between 50 and 150 steps to place a single outbound call. Each one of these steps creates load. The more load on a Trixbox system the wackier it behaves. So if you are placing 12 calls per second your Trixbox could be going through as many as 1800 dialplan steps per second. Asterisk generates the most load during call setup and tear down. When asterisk is over loaded it can cause all kinds of problems including issues with keeping PRIs in sync.

It is far better to send the calls from the Trixbox system to the Vicidial system and then out to the provider.

[wizzbangca]

Another member sent me a pm with the same suggestion. I consider this a bandaid to the way the system should be: dialer to pbx. If the bandaid gets me up and going, leaving time to work on dialer to pbx, I'm all for it.

Thanks for the suggestion. It's amazing what pressure and stress will do to someone.

[mcargile]

I really do not consider this a bandaid at all.

Trixbox is not designed to be anything but a SMB pbx system. It is designed for a dozen or two physical person to be manually dialing a phone number once in awhile.

Vicidial on the other hand is designed specifically for this. It has been designed and optimized to be able to automatically place 20 calls a second with as many as 150 outbound lines, on a single server. Routing traffic from a Trixbox system into a Vicidial system will barely register as a blip on Vicidial's load.

We are building more and more PBX functionality into Vicidial. One of our goals for the future is for Vicidial to be able to operates as an Enterprise quality PBX system, most of the core features of this are already built into our development tree and should be in the next release (2.2.0).

[williamconley]

another alternate is to reduce your calls per second to 1 or 2 and see if this in any way affects your congestion issue. if it does, you have a place to begin (and by the way i have centers with under 10 agents that prefer it that way for reduction of dropped calls, but that's beside the point).

If indeed the problem is overloading the trixbox and not an "incompatibility" ... then 1 call per second SHOULD reduce the effect and at least delay the lockup, if not avoiding it entirely.

do you have ANY scenario where the PRI works properly that you can use as a target to try to tunnel all the calls through?

[wizzbangca]

I am attempting to create some type of working pri connection using an asterisk based distro of some kind. Trixbox failed miserably as we know. VicidialNOW is in the process of being updated, then the rhino drivers installed. AsteriskNOW is a major failure for me for only one reason: rhino drivers do not compile. The rhino setup script says there is no "ANOW" drivers rpm. Manual driver installs fail with a pile of errors. My next shot at a pbx is FreePBX manual install on CentOS. Never got that working, but if vicidialNOW works on the pri by itself, I now have a lot more breathing room to get a pbx running.

547 updates sure do take a while.

[mflorell]

You could also try ViciBox which has compilation tools and zaptel/asterisk source code included with it.

[williamconley]

definitely take a shot at vicibox (which is freakin marvelous, by the way) and vicidialnow (if you like centos) and if all else fails, try just plain "asterisk" and see if you can get it up. who knows, maybe it will turn out that you just need to have a sep box for the pri and pass calls through it (for now).

asterisk itself will run on most distros worthy of note. i use gentoo ("emerge asterisk"), and there are several other distros that will install it directly from designed packages. if you end up with a generic old p4 with naked asterisk running only the pri and all pri calls passing through it ... at least you're running.

[wizzbangca]

I really appreciate everyone's suggestions. vicibox is being downloaded for consideration. The Rhino card has been isntalled on Vicidialnow. For those who have a rhino card, setup-rhino is being modified for vicidialnow. I have started trunking trixbox to vicidialnow. Trixbox logs into vicidialnow through an iax trunk. When I call through vicidial from trixbox, I get the following on vididial showing the number i dialed through trixbox:

NOTICE[2942]: chan_iax2.c:7663 socket_read: Rejected connect attempt from 10.48.10.8, request '1XXXXXXXXXX@default' does not exist

the extensions.conf has the following:
TRUNK=Zap/g1

[default]
; dial a long distance outbound number
; This 'o' Dial flag is VERY important for VICIDIAL on outbound calls
exten => 1NXXNXXXXXX,1,AGI(agi://127.0.0.1:4577/call_log)
exten => 1NXXNXXXXXX,2,Dial(${TRUNK}/${EXTEN:1},,To)
exten => 1NXXNXXXXXX,3,Hangup


I am guessing the problem has to do with setting up the trunk properly in vicidial. Going to keep looking into this.

[mcargile]

Code: Select all

[default]
; dial a long distance outbound number
; This 'o' Dial flag is VERY important for VICIDIAL on outbound calls
exten => 1NXXNXXXXXX,1,AGI(agi://127.0.0.1:4577/call_log)
exten => 1NXXNXXXXXX,2,Dial(${TRUNK}/${EXTEN:1},,To)
exten => 1NXXNXXXXXX,3,Hangup


Should be:

Code: Select all

[default]
; dial a long distance outbound number
; This 'o' Dial flag is VERY important for VICIDIAL on outbound calls
exten => _1NXXNXXXXXX,1,AGI(agi://127.0.0.1:4577/call_log)
exten => _1NXXNXXXXXX,2,Dial(${TRUNK}/${EXTEN:1},,To)
exten => _1NXXNXXXXXX,3,Hangup


Note the _ at the beginning of the dial strings. That denotes that you are doing a pattern match. Before the VicidialNOW system was literally looking for a dial string with N's and X's in it rather than 2-9 and 0-9 respectively.

[wizzbangca]

Thanks mcargile, I now get the expected reaction from asterisk on the vicidialnow server. Now I need to configure incomming to go to the IAX trunk trixbox is on. My guess is, and what I have now, is below. Can't test this until I get the PRI for testing.

Code: Select all

; Extension 7275551212 - Inbound local number from PRI with 10 digit delivery
exten => NXXNXXXXXX,1,Ringing
exten => NXXNXXXXXX,2,Wait(1)
exten => NXXNXXXXXX,3,AGI(agi://127.0.0.1:4577/call_log--fullCID--${EXTEN}-----${CALLERID}-----${CALLERIDNUM}-----${CALLERIDNAME})
exten => NXXNXXXXXX,4,Answer
exten => NXXNXXXXXX,5,Dial(iax2/2300,30,To)
exten => NXXNXXXXXX,6,Voicemail,u2300


Next problem I just figured out is with the usage of lines on the pri. Max lines I want to use for dialing is 16. The rest I want to use for incomming/outgoing on the trixbox. I set Max VICIDIAL Trunks= 16. The help on the setting page seems confirm this is how I set the autodial max lines. What if I wanted to run multiple autodial campaigns, is there a way to set max lines on a campaign basis? This would be for future reference. I have looked around and haven't found anything on this.

Thanks.

[mcargile]

Set the Max Vicidial Trunks in the server section to 16 and that will limit Vicidial to 16 trucks across all campaigns. As far as your inbound is concerned you are once again forgetting the underscore at the beginning of the pattern match.

I would however suggest actually using the DID routing in Vicidial to send the calls to the Trixbox. Create an extension like this:

Code: Select all

exten => 2300,1,Dial(iax2/2300,30,To)
exten => 2300,n,Voicemail(u2300)
exten => 2300,n,Hangup()


And then when you setup the DIDs in Vicidial set the routing to EXTEN and put 2300 for the extension. By doing it this way if you decide in the future to do inbound on the Vicidial in the future you do not have to change the dial plan just the DID routing.

[wizzbangca]

You jumped ahead of me on the DID. For now, I need all incoming to go to trixbox for the IVR on trixbox. Before I got a chance to modify for the "_", you caught me. Thanks for verifying the line settings for the dialer, much appreciated. I get to test in an hour. Can't wait to see this thing work.

[mcargile]

I understand that you want the calls to end up in the Trixbox systems IVR, but I am suggesting you use the Vicidial DID route system to get them there. The reason being is that it will make it easier to setup inbound later in Vicidial.

[wizzbangca]

Testing went really really really really bad. Every call that comes is ends up with this:

Extension 'AAAAAAA' in context 'from-pstn' from 'XXXXXXXXXX' does not exist. Rejecting call on channel 0/2, span 1

Where AAAAAAA is the 7 digit DID seen in asterisk from the number people dial to get to us, XXXXXXXXXX is the number of the person calling in.

This has something to do with the outbound setup, I believe. That is piddly shit compare to the PRI is STILL going congested, which really really really really pisses me off to now end:

Jun 9 09:19:21 NOTICE[2806]: chan_zap.c:8365 pri_dchannel: PRI got event: Alarm (4) on Primary D-channel of span 1
Jun 9 09:19:21 WARNING[2806]: chan_zap.c:2441 pri_find_dchan: No D-channels available! Using Primary channel 24 as D-channel anyway!
-- AGI Script agi://127.0.0.1:4577/call_log--HVcauses ... ---------- completed, returning 0
== Manager 'sendcron' logged off from 127.0.0.1
== Primary D-Channel on span 1 down
Jun 9 09:19:26 WARNING[2806]: chan_zap.c:2441 pri_find_dchan: No D-channels available! Using Primary channel 24 as D-channel anyway!
== Parsing '/etc/asterisk/manager.conf': Found
== Manager 'sendcron' logged on from 127.0.0.1
-- Executing AGI("Local/1XXXXXXXXXX@default-2380,2", "agi://127.0.0.1:4577/call_log") in new stack
-- AGI Script agi://127.0.0.1:4577/call_log completed, returning 0
-- Executing Dial("Local/1XXXXXXXXXX@default-2380,2", "Zap/g1/11XXXXXXXXXX||To") in new stack
Jun 9 09:19:39 NOTICE[5423]: app_dial.c:1076 dial_exec_full: Unable to create channel of type 'Zap' (cause 34 - Circuit/channel congestion)
== Everyone is busy/congested at this time (1:0/1/0)
-- Executing Hangup("Local/1XXXXXXXXXX@default-2380,2", "") in new stack
== Spawn extension (default, 1XXXXXXXXXX, 3) exited non-zero on 'Local/1XXXXXXXXXX@default-2380,2'
-- Executing DeadAGI("Local/1XXXXXXXXXX@default-2380,2", "agi://127.0.0.1:4577/call_log--HVcauses--PRI-----NODEBUG-----34-----CONGESTION----------") in new stack
-- AGI Script agi://127.0.0.1:4577/call_log--HVcauses ... ---------- completed, returning 0


The only way to get rid of the congestion is to restart the vicidialnow server. The problem that caused this thread in the first place is still THE problem needing to be solved. The first and foremost question that must be answered: How the hell do I keep asterisk from going congested on the PRI? This can't be blamed on trixbox, vicidialnow is doing the same thing. Don't tell me to move to a different asterisk based platform, I just did.

[williamconley]

perhaps a simple detailed description of the hardware / software setup and ONE specific problem being experienced would be in order.

pieces from here and parts from there do not make it possible to have lots of eyes look at the issue and comprehend what is going on, and then help to resolve it.

i'd try to "summarize" the issue myself, but i've read all of it and i'm not sure which issue we are talking about.

i'd also appreciate it if you drop the expletives. take a deep breath, and remember how much we get paid for this help. but that's me. i'm like that.

[wizzbangca]

This is still going on:

I have been looking into why vicidial stops dialing after few calls. Vicidial is set to dial 5 calls per second with a max of 16 lines. That isn't where the problem lies. I can't try freepbx right now, DAHDI drivers for Rhino cards aren't exactly easy to install. I dug into live monitoring while emulating a dialing environment and found out the pri is going congested with the d-channel going down before a bunch of SAMBE mesasges. I have been spending the last 3 weeks trying to get rid of the congestion. No luck. AT&T, not surprisingly, doesn't want to help figure out where the congestion is being caused. Rhino, the manufacturer of the t1 card, looked into what was going on. They setup the zaptel configs to what they use for AT&T voice pri. Still get congested. Rhino is at a loss since all their known working settings aren't working for me. Since asterisk is coming up with congestion, even freepbx may end up with the same problem. For now, I am sticking with trixbox.

This project is close to being scrapped, along with my job. Countless hours of searching have turned up nothing for the correct settings to connect to an AT&T pri. I have smashed into a brick wall and have no idea what to do.

The congestion was what brought out this post in the first place, and is the underlying problem. VicidialNOW is not immune. The PRI still goes congested. Pointing the finger at trixbox or saying I should try another distro to solve this problem won't work. The only thing that will solve this are the appropriate settings for AT&T. Until someone decides to come forth with the appropriate settings, the project is considered over.

I have tried getting AT&T to rebuild the circuit, and test their equipment on the circuit. But, since the legacy pbx works 100% with no congestion, why would they do anything more than what they have already done? I wouldn't.

[mflorell]

I would suggest trying several different settings(5ESS, NATIONAL, DMS100, etc...) for the switchtype. We have run into many occasions where the telco said they were delivering one thing, only to be delivering another in reality.

Tyr all of these:
; national: National ISDN 2 (default)
; dms100: Nortel DMS100
; 4ess: AT&T 4ESS
; 5ess: Lucent 5ESS
; ni1: Old National ISDN 1

remember to do a full power off(unplug the server even) between tests to make sure the card resets properly.

[mcargile]

As Matt suggested I would try different settings for the signaling. This may very well be your D channel issue. The other possibility is the Rhino card may be malfunctioning. The normal way of checking this is to try a different card.

There are several dialplan configuration issues as well.

First off you have the calls from the pri coming into the context [from-pstn] in the dialplan. Does this context even exist, it is not part of the standard Vicidial dialplan? If not you will want to create this and put the dialplan section that sends the calls over to the trixbox there.

Second off you are dialing a 12 digit phone number from the looks of things.

Code: Select all

-- Executing Dial("Local/1XXXXXXXXXX@default-2380,2", "Zap/g1/11XXXXXXXXXX||To") in new stack


Count the number of digits after the "Zap/g1". I know of no carrier that accepts 12 digit numbers in the US. This could very well be why your calls are coming up congestion. 11 and 10 are the norm. I think you are adding an extra 1 at the beginning.

Based off the number of errors that you have had with your configuration so far (and I am betting there are quite a number more), might I suggest actually hiring a qualified Vicidial consultant? Particularly one who has experience with PRIs.

[williamconley]

So our issue is that the PRI settings are TESTED and WORKING (with other equipment) but will not work with Vicidial (or TrixBox? or FreePBX?)? Or is it that the Vicidial calls are going through the TrixBox and not direct?

If the issue is that neither Vicidial nor Trixbox work ... you should consider skipping both of them and installing pure asterisk, perhaps with a different distro (or two) in case there is a simple issue with an underlying utility.

Then the ball falls into the court of Rhino. If the card is designed to work with asterisk and there is no other software running ... it either runs or it goes back.

Let Rhino tell you exactly how to set up your server, I'd even let them choose the OS to be sure. AFTER that works, you will have a baseline upon which to build. IF that never works, send the card back.

By the way, this isn't just a suggestion, this is what I had to do with my FIRST digium card (which was defective!). The one they replaced it with worked perfectly out of the box. Bad week though.