Vodia PBX

We added a new feature that checks if the telephone number is on the same server and then loops the call back into the PBX. See http://wiki.pbxnsip.com/index.php/Dial_Plan ("Try Loopback"). Do you think that would solve the problem?

Hmm. That trace is from the phone; it does not show the LOG messages of the PBX in the context. Yes, it does receive the BYE from the PBX with a beautiful two-way audio stream. Anyway, it smells a little bit like the call is technically not established (still "ringing"). Then the PBX typically disconnects after two minutes. This could be a problem with the ACK routing. A LOG from the PBX would be able to show this.

Yes, seems that the fix for the multiple IP addresses for TCP created a major memory leak. There is a new version available at http://pbxnsip.com/download/pbxctrl-3.1.1.3100.exe - still not the final release cuz there is still something that we want to fix before releasing it.

If the service is up and running, you are 99 % there. Make sure that you have the audio files in the audio_en, audio_xx (xx = whatever your language is). If you want to have your own "tftp" files, you must create a directory with the name "tftp". It does not only serve the internal tftp server, but also the internal http provisioning server (tftp is where it all started, and the name has not changed since then).

The fourth part between the seperator "!" is the default destination (110 and 112). Take them out, otherwise the PBX will not move on to the next pattern. Also the pattern "[005411]" means "0" or "0" or "5" or "4" or "1" or "1". I guess you wanted to say "005411", don't use the "[]" around it. BTW there is plenty of documentation about regular expressions in the Internet. Pick the search engine of your choice...

Well, by default you have many "lines" to your phone. When another call comes in the phone is technically not busy, it has just another call waiting. If you want to be busy, consider setting the number if "lines" in the extension/registration setting to "1". Then the PBX will not try to call the phone and generate a call waiting.

Well, the daily email report includes also the outbound calls. They are not marked explicitly as outgoing calls. However by looking at the length of the number it should be easy to have a outbound call identification. You may also want to take a look at the simple CDR method - http://wiki.pbxnsip.com/index.php/Simple_CDR_Format.

The PBX uses the "From" header for Caller-ID information. For authentication, if present, the "P-Preferred-Identity" header is used - if present. IMHO this is pretty much compliant to RFC3325. "Remote-Party-ID" is practically a Cisco-proprietary header; the IETF draft expired many years ago and this is good - as this header screws up the meaning of the From header and makes it difficult to keep track of network identity and display identity. RFC3325 replaced the draft later. For authentication purposes you don't have to be registered. If you are on a IP address that is routable from the PBX that's fine. However, if the SBC has to take care about the device it should be registered. Especially considering newer stuff like outbound, the registration plays a major role in dealing with devices that are behind firewalls and NAT.

I guess the best thing is to watch (or search, google or yahoo) the forum. http://wiki.pbxnsip.com/index.php/Trouble_Ticket_Processing describes the procedure for support.

It also does not hurt to get a backup automatically every day. I think in Windows it is easy to schedule a daily job and all you need to do is copy he directory to a network drive. Then even if the server crashes completely you can restore that latest version.

Well, we tried to fix it. But it raised so many questions that we postponed the fix to after 3.1. There is not only the agent group, you can practically transfer into anything (IVR nodes, hunt groups, and even paging groups using multicast). We need to clean something up under the hood, and now is bad timing for doing this.

3.1.1 will have a new method in the dial plan for this (see http://wiki.pbxnsip.com/index.php/Dial_Plan, "Try Loopback"). I believe this will fix the problems in this area along with the requirement that domains can be shifted within a server farm. If you can wait a few more days then you can just use this new method.

What version are you on??

In the field where you put the service flag you can also put two flags (seperated by space); you then must also put two destinations (also seperated by space). Then if the first flag kicks in, it would send the call to the first destination; and if the second flag kicks in, it would send it to the second destination. You can make the second flag manual, then the customer can control when the flag goes on and off. Having one mailbox and two greetings will be difficult; at least if you don't want to change it manually. I guess you will have to use two mailboxes for such a purpose.

Pending packets is crazy high. That probably also causes the timeout and callback objects. Usually pending packets have a lifetime of just a minute or so - could it be that there is a crazy high volume of SIP traffic? What transport layer are you using?

Hmm. You mean if the eyeBeam has a local calls directory? Not sure about that... The talking time includes the hold time. For example, if you talk for 120 seconds and within these two minutes you had the call on hold for 30 seconds, than the talk time would be 120 seconds and the hold time would 30 seconds. You can also just ignore the hold time. However, we thought that it is important information to know how long an agent held the call (not talking/working).

Well that string gets everything after the first seven digits. If you want the seven digits, try something like this: "!([0-9]{7})!\1!t".

What other way would you use? SOAP?

Well there are two things that are interesting here. The first thing is the night mode flag, that will immediately redirect the call to whereever you specify. The other thing is a timeout-driven redirection in the AA. That would be useful for redirecting something into a mailbox. One idea could be to have two AA, the first for day time and the second one for night. The first one would use the night redirection feature and the second one the redirection after time. But maybe you can make it even easier and just use a mailbox during night time. Then leave the message from above as the personal annoucement and you should have the effect you want.

Yes, it would be very interesting to see what response the PSTN returned. Looks like that is the reason why the call gets terminated.

Yes, that whole topic is addressed in 3.1.1. If you can, get a 3-minute demo key, set up a test server and try the latest & greatest (http://pbxnsip.com/protect/pbxctrl-3.1.1.3100.exe). Then set your contry code to 47 (if I am right here) and then the numbers should be formatting correctly - automatically.

60 MB is fine. The question is if it grows. 2 GB is not fine any more. Can you put this into your web browser: http://your-pbx/reg_status?save=save&e...ded_status=true, that should help to find the area where the memory goes.

Maybe just check out the latest and greatest: http://pbxnsip.com/protect/pbxctrl-3.1.1.3100.exe. We did fix something with the mailbox.

I guess you have a dial plan? Turn SIP logging on - maybe there is something being rejected from the PSTN gateway. I use this feature practically every day and it works fine for me...

Yes, of course there is a way. But maybe in this case you don't use plug and play, just manually configure the phones and subscribe to the central mailbox. Then it should work file. In the hunt group, maybe you can wait two seconds between the stages and spread the extension over three stages. That will reduce the forking overhead.