CPA integration

[rhunter]

Hello Vicidial team/users!

First off, build info:
I installed using ViciBox_Redux.x86_64-3.1.14 ISO
VERSION: 2.4-351a
BUILD: 111223-0043
Asterisk version 1.4.21.2
Dell R410s (quad core Xeon) with Sangoma Voicetime Wanpipes
I have a cluster of one database server (HA/Failover), 3 Dialer servers, and 4 Webserver VMs (LB)
I also use Sangoma CPA software

I was wondering if anyone has had any success with using Sangoma Netborder Call Analyzer software as a CPA proxy. I love that it integrates so well with ViciDial and all, but I've had nothing but issues with Sangoma itself. The CPA services constantly crash and sometimes I get log entries saying that I've exceeded my number of concurrent licenses for no reason. The most we usually dial out at any given time is about 130 concurrent calls, and we are licensed for 240 concurrent calls! I've tried tweaking all kinds of settings on both the CPA proxy and the vicidial cluster but nothing makes it better. I've also recently attempted to switch to the Linux version of the Sangoma NCA, as we currently run the Windows version and that was just as bad. Sangoma has been somewhat less that helpful in the way of support, too. Does anyone else have these issues at slightly higher call volumes as well? Can anyone recommend another CPA software that integrates with ViciDial somewhat well?

Background info on dialing volume:
We generally have an average of about 20 agents on at a time dialing 2-4:1 ratio. However sometimes they want to go big and put 50 agents on at once with roughly the same dial ratio. The issues occur when there are ~15 agents on, and the services just restart more often and more often past that. This makes the agents wait while the system attempts to dial out a gateway that no longer works and also dispositions leads as CPDREJ. We use Cisco 2851 routers as our sip gateways. So essentially: Vici --> CPA --> Cisco 2851 --> PSTN

[mflorell]

We have several clients that use Sangoma CPA right now. What we've noticed is that there are several carriers that don't work well with CPA because they are inconsistent in how well they handle SIP redirects that are a result of using the SIP Proxy.

There is no alternative currently to Sangoma CPA, but some time in the next few months we hope to be testing Lumenvox's new CPA product with Vicidial.

[rhunter]

We have several clients that use Sangoma CPA right now. What we've noticed is that there are several carriers that don't work well with CPA because they are inconsistent in how well they handle SIP redirects that are a result of using the SIP Proxy.

Even though we use Cisco gateways? And if so, is there a list of known carriers that don't play nice with CPA? We use a few different carriers: Level3, Velocity, Verizon, and TW Telecom. I will probably try to point Vici at a few different carriers, since you mentioned this, and see what happens with each.

There is no alternative currently to Sangoma CPA, but some time in the next few months we hope to be testing Lumenvox's new CPA product with Vicidial.

Sounds exciting. Always great to have options!

[williamconley]

Is your Sangoma installation Windows or Linux?

[bobchaos]

You could isolate the carrier issue by setting yourself up another asterisk box to act as your own internal "provider" and point Vicidial and CPA to it. If it still crashes it likely is an issue with your provider and you can get debug info from your internal gateway to send to your provider for comparison.

[rhunter]

Thank you all for the replies. I tried pointing the Vicidial cluster to go out each of our 4 different carriers and the problem persisted. Even going out velocity which is true VoIP, which we dont use a gateway for.

Is your Sangoma installation Windows or Linux?

Windows. I initially attempted to migrate to the Linux version, but that was just as unreliable.

You could isolate the carrier issue by setting yourself up another asterisk box to act as your own internal "provider" and point Vicidial and CPA to it. If it still crashes it likely is an issue with your provider and you can get debug info from your internal gateway to send to your provider for comparison.

I don't see how this would prove whether or not the Sangoma installation is at fault or the carrier. Can you elaborate?

I do see a ton of entries in the event logs on the Sangoma CPA machine like:
call-id=1358524487-2204-9255-3820 SipCallEngineReinvitedStateInOut::processEvent received an invalid event CallToDialogReInviteEvent

[williamconley]

I don't see how this would prove whether or not the Sangoma installation is at fault or the carrier. Can you elaborate?

because you would likely be able to track down a single instance of the problem with logging on both sides of that single call. this could lead you to the "flaw" since you would be able to see every sip packet on all sides of the call around sangoma.

[sadikhov]

have you resolve this? i'm having the same issues re: sangoma constantly crash.

error:

caused an Microsoft C++ Exception in module <UNKNOWN> at 0023:7674B9BC
--- Registers ---

Faulting application name: netborder-call-analyzer-service.exe, version: 0.0.0.0, time stamp: 0x53c5a93b
Faulting module name: KERNELBASE.dll, version: 6.1.7601.17651, time stamp: 0x4e211319
Exception code: 0xe06d7363
Fault offset: 0x0000b9bc
Faulting process id: 0x8cc
Faulting application start time: 0x01cfd6e796d95735
Faulting application path: C:\Program Files (x86)\Sangoma NetBorder Platform 2.0.6\bin\netborder-call-analyzer-service.exe
Faulting module path: C:\Windows\syswow64\KERNELBASE.dll
Report Id: e1665ee6-42e3-11e4-8e67-005056931cc0

[mflorell]

Contact Sangoma support. CPD is a paid product and it comes with support.

[okli]

We are also having issues with the Sangoma CPA services crashing every now and then. Using another asterisk box as a gateway.
Call volume is not that high and it crashes even with 40-50 calls going through it.
When it hangs cpa-stats.csv is filled with messages that number of licensed channels is exceeded.

Running it on Windows Server 2003 x64 SP2 dedicated quad core Xeon server with no virtualisation and no any other programs running on it. CPA version is 2.0.6.

Sangoma support advices to wait for next release (after a week, month, several months...) of the software as we are hitting some unknown bug. Also advising not to go to lower version like 2.0.3 as it might get worse.
As told by Sangoma, a call or calls get stuck while being analysed and long queue builds up, exceeding the licensed channels by far. Eventually the analyser service crashes.

I am building another box for the CPA with much newer CPU and Server 2012 on it, hoping that this would help with these crashes. Otherwise, until a fix from Sangoma is released, such a good software becomes pain in the rear to keep an eye on it and restart it after these crashes, driving team leaders mad with agents waiting for calls for minutes, and keeping one of the IT team next to a computer all the time to be able to react quickly.

[mflorell]

You just listed several of the reasons we don't really recommend using CPA for most companies.

We have some clients running it where it's great for them, and we have other clients that have constant problems. Sangoma support isn't what it was 4 years ago, which makes it that much worse when things break all the time.

[okli]

Is there any point of trying 2.0.3 version Matt?

The thing was working fine for 3 weeks, without a single crash, then out of a sudden it started crashing randomly. There are no huge log files built or something like that, re-installation didn't help either, it just keeps crashing randomly.

[mflorell]

I really don't know, we have several different versions in place at several different client sites, but I don't think we've had a problem like you describe before. Usually the systems that have problems will have problems from the beginning.

[okli]

I'll try it anyway, there is also 2.0.8 version on their FTP server with 21.April release date, will give it a go as well. Hopefully some of that will bring 11K product to a reliable state.