Solved Calls going (almost) straight to voicemail

[Michael.Talbot]

One of the 3CX instances that we manage for our clients has just started misbehaving. When a user calls another user the called extension rings twice and then the call goes through to voicemail even though there is a 20 second ring time defined for the extension. External calls get put through to the call queue properly but, once there, the extensions on the queue ring twice before the queue goes on to the next defined behaviour.

The 3CX server is debian installed using the official ISO from 3CX and kept up-to-date (including the latest update) and this strange behaviour has only started happening in the last day or two. The exact same configuration (installed from a backup) on Windows does not have this problem.

Has anyone else seen this problem or have any ideas on how to fix it?

Thanks

Michael

 

[JohnS_3CX]

Hi Michael,


Was NTP installed when then issue appeared or was it installed later?

The timestamps will not be reliable during the period where NTP is not active.

Reference: https://www.3cx.com/docs/installing-microsoft-hyper-v/#h.cjj20zk71lns

 

[Michael.Talbot]

OK, just to clarify. Strictly speaking it isn't NTP that is installed but the Systemd TimeSyncd service which is installed by default during the initial installation. Also, since this is a VM on Hyper-V, the debian hyperv-daemons package is also installed. All of these services are running properly and the time is being properly synced with the correct timezone set.

As for the timestamps on the log, the server had been running for a good 10 minutes before I ran the tests so any time sync would have completed well before then (Linux performs an initial time sync on boot).

 

[SweetAction]

3CX iso or stock ISO and 3cx package?
Does the CPU graph have strange lines (i.e. showing it goes backwards)?

 

[Michael.Talbot]

The installation was done using the official 3CX ISO (Debian 9) and there were no strange (time reversal) lines on the graph when the tests were being done (I have seen them on occasions but they've always corrected quickly and never been associated with these issues).

 

[Mark Jones]

beleive it or not check the user webclient for any forwarding they may have enable.

 

[Michael.Talbot]

I've had a look at the webclient (and the admin console) and there are no forwarding rules other than those that have always been there. Certainly nothing has been changed since last week (when this problem wasn't happening).

 

[SweetAction]

For testing can you install NTP client as 3cx recommendations.

HyperV extensions on official ISO have not worked correctly for me in the past

 

[JohnS_3CX]

OK, just to clarify. Strictly speaking it isn't NTP that is installed but the Systemd TimeSyncd service which is installed by default during the initial installation. Also, since this is a VM on Hyper-V, the debian hyperv-daemons package is also installed. All of these services are running properly and the time is being properly synced with the correct timezone set.

As for the timestamps on the log, the server had been running for a good 10 minutes before I ran the tests so any time sync would have completed well before then (Linux performs an initial time sync on boot).

Hi Michael,

The NTP service must be installed when using Debian on Hyper-V as per our guide.

 

[Michael.Talbot]

OK, I installed the NTP service and restarted the server to make sure that it picks up everything properly.

I gave the server a few minutes to get settled and then ran some tests (extension to extension and incoming external calls) to see if the problem still exists. I'm pleased to report that, at the moment, the system is working properly and the voicemail diversion issue has been resolved.

I don't know when the NTP requirement was added but, last time I checked (earlier this year), there was nothing on the system requirements page saying that it needed to be installed. I hope that 3CX will add automatic installation of NTP to their Debian ISO so that other people don't get caught by this issue.

Thanks to everyone for their help getting this fixed.

 

[techcare]

This forum post helped me as well. Thanks guys. I've rolled out NTP on all my Debian 3CX versions hosted on HyperV now.

Just curiously do you NEED to restart the VM for the Hyper V integration settings to turn off?

 

[JohnS_3CX]

@Michael.Talbot It has been there since V15 at least that I remember and it also applies to Windows and Debian too.

The issue is not the OS, but rather the fact that you are running inside a VM (virtual hardware) so with a lack of real hardware timers, drift will occur sooner or later. The NTP service (for Debian in this case) takes care of this "keeping you in track".

Following the guides is critically important