Vodia PBX

Eh you mean RHEL4? That might take a day or two......

Well the problem is when the user decides to change the username you cannot generate a new hash. If we are using the hash method we have to request users to re-enter their passwords.

Did you try "If the caller already waited longer than (s) ... redirect to the destination:"?

Maybe for the sake of finding if that is the problem try "always".

I think we fixed the problem already... Will be included in the next release. If you want a preview, let me know what OS you prefer.

That is a kind of Linksys classical problem. The default dial plan of the phone makes it hard to dial star codes on the PBX. See http://wiki.pbxnsip.com/index.php/Linksys for some dial plan examples.

What is your setting "Generate passwords" (admin/settings/ports)? Maybe the PBX would not generate the passwords for the device?

Ehh. Looks like a bug to me. Workaround could be to use the form "<accountname@domainname.org>; <accountname2@domainname.org>".

Very good. Maybe you increase the log level and look for messages like "Trunk xxx sends call to xxx" (log level 5)? Then we can see what the PBX tries to do with the INVITE.

Maybe you have to turn "strict RTP" routing on. I believe that test tools don't have to be very NAT-friendly, so the PBX gets one-way audio.

Well, at least Broadsoft is using SIP, and I guess with Asterisk they are also using SIP so I would say it works. But only the real device will tell how much exactly works.

We thought about that also. However, when you change a domain name or change one of the alias names than a hash would not work any more. Encrypting it with a hardcoded key only "obscures" the passwords (until someone gets the bit secret out of the code). Encrypting it using the private key of the PBX (used for TLS) would be a possibility. At least the sys admin login uses a hash for the password! And of course, file system access should be strict. This is not a public area.

Well, you have to set up the the password on the Polycom phone (see http://wiki.pbxnsip.com/index.php/Polycom). Did you do that?

Oh there is probably an old PBX running still (don't reboot the box until the problem is solved!). You can find the process with netstat (e.g. netstat -anp|grep 5060). Kill it.

You mean that cell phone forks should use a different dial plan that regular calls or other call redirections for that extension? What is the purpose? Caller-ID presentation?

You see this screen only if you log in as a user. The domain or sys admin cannot set up conferences.

Whow, we don't support that yet but it looks interesting. Do you have a pointer to the documentation?

Well if they keep their software it should work like to other models...

We used WinSIP and had it run for a million calls. No problem.

That is really hard to say. But lets assume 50 calls that you be 1000 calls for the whole box. Nice!

The [2-9]11 is essentially for 911. Maybe you just write 911; maybe also include 411 if someone wants to dial it. This way you can take the 811 out. In the end, you need to test test test. IMHO life would be easier if they can move the 1xx extensions out of the way or just accept that they have to dial 9 for an outbound call (to "seize" a line, haha). If you have to tailor them their own dialplan that will mean some work. Or they get over the whole thing and press Dial after entering a number. Thanks to the several billion cell phones out there that is not such an crazy idea!

We tested Cyberdata (see http://www.cyberdata.net/products/voip/voip-speaker.html). Works nicely. Available for example from http://www.abptech.com/products/cyberdata.

Last resort is to use the PCAP feature on the phone to get a Wireshark trace of what the phone tries to send out.

Ouch. The white edition does not have the features that you need to run version 3. Check the /etc/network/interfaces file. Make sure that there is no "eth2".

Linksys says that is a feature! There was a discussion some time ago about this and at that time there was no option. The only way to stop it is to set the number of lines to "1" - however then call waiting also becomes impossible.