Vodia PBX

One-way audio can also be caused by SRTP - if the MAC "signature" does not match then the PBX will discard the message. You'll see a message like "Dropped ... SRTP packets with wrong MAC" on log level 4 in the media log category.

White noise is usually caused by SRTP problems. Try turning it off on the phone and the PBX.

Hhm, you mean it uses an IP address that the box actually does not have? Maybe because the box runs behind a DMZ and the installation uses that trick to get that working? Maybe we need something that explicitly set the "https://<ipnr defaultgateway>" part - will be in the next version.

It says "Please enter the extension number"? Maybe is there a space behind the '*'? What version are you running?

IMHO using the outbound proxy is a must. Imagine that you really want to call a complete URI, then the phone should still send the call to the PBX which then locates the destination URL. Apart from that, it also makes sense to present also for inter-system calls routable telephone numbers, meaning the whole DID number for that extension. So it would be better to present the From-header "Fred Feuerstein" <sip:+44322342342@domain1.com> instead of "Fred Feuerstein" <sip:342@domain1.com>.

Well the caller-ID is transmitted between the first and the second ring. That's why analog phones with Caller-ID support first start ringing and then after that start showing the Caller-ID. Unfortunately, the genius that specified SIP did not envision that and wrote that the caller-ID cannot be changed after call setup (though there is a new RFC that now starts supporting that, but procatically all SIP phones not support that yet). That is why the CS410 FXO gateway waits for the second ring before it sends the call to the SIP phone.

Well, that is one reason why every dialog should eventually hang up - at the very maximum after two hours. If the call is still occupying the line after this there is some more fundamental trouble. Hangup detection is really a big problem in FXO. Every country does it differently. What country are you in?

At the moment it is all or nothing.

I would take the DMZ out of the game here first. If it does not help, only Wireshark will tell wat is really going on.

Hmm. Seems there is a problem with the Alert-Info header on the Aastra. I guess the PBX puts a header into the INVITE? You can register multiple phones, but you can automatically provision only one phone per extension.

In most cases a dongle really make sense. But at the moment we only support Windows dongles. Ask sales@pbxnsip.com for help... Not possible. One reason is obviously the license problem (create one virtual machine and you are "all set"). The other reason is that VM are not very good in keeping real time requirements. Maybe one day it will be better, but at the moment we are careful about this problem.

Yea, PnP is the key. Hopefully more stuff coming!

There is a lot of information on this topic on http://wiki.pbxnsip.com/index.php/Automatic_Provisioning. I attach the latest version of the pnp.xml file. If you need other files, let me know. pnp.xml

Well, the transfer happens when the other call is not connected yet. We have a patch for that, but it is not part of a released version yet. In Wireshark, enter as filter: sip.Call-ID == "998aeed0-22a00782-71608e9b@192.168.31.111" to see the unconnected call, and sip.Call-ID == "9d5c5862@pbx" to see the call that sends the REFER.

No, the MAC address binding is only used for the provisioning. We can rule out interference with calls here. Just try to remove the localhost alias (rename it) and see what happens, I am sure that this will make things clearer.

Well, you can always put a "custom" file into the html directory. What file are you looking at?

For the license check things are a little bit more complicated. The PBX counts the call lets that are primary call legs - so that the example with the forking the call would count only as one. So that means if your license says you can have 10 calls, then it might actually have 20 call objects. I know what you are saying... I think for the next version we should show the number of primary calls and the number of call objects. Agreed.

Try the following links: http://fox.snom.com/download/snom300-7.1.29-beta-SIP-f.bin http://fox.snom.com/download/snom320-7.1.29-beta-SIP-f.bin http://fox.snom.com/download/snom360-7.1.29-beta-SIP-f.bin http://fox.snom.com/download/snom370-7.1.29-beta-SIP-f.bin Obviously still beta...

In the file system it is quite difficult. Because there are many links between the tables. And if you want to restore it, you must choose different table entries and restore the links.

Usually those kind of problems come if people are not using an outbound proxy. Other problems are when phones place calls from received call lists where the contact is not clear. But it should be possible to locate the problem looking at the INVITE with the Request-URI. In your setting the line 1 looks a little bit suspicious because the domain name is a IP address. In multiple-domain environments it would be a little bit strange to have a alias name that is the IP address. Maybe you should try to temporarily remove the alias name "localhost" to really point out where the domain does not match in your setup.

Well, it counts the call objects. Because of the B2BUA architecture, a regular call has two legs. But a call to a IVR has just one leg. Forking calls can have more than 2 legs...

Currently we are investigating if we should just support CSTA. Seems that this supported by a coiple of other tools.

We are working on a domain backup/restore feature. Backup is already there but restore not yet.

We found a problem when people are doing attended transfer and transfer the call before the other side picks up. Could this be a problem here?

Agree. It originally was supposed to be a feature, but it is probably just confusing. We'll disable that in the next version (no setting).