Had an issue recently with a client configured on a debian install of 3cx.
The client was getting constant "network disconnected" messages, particularly when they attempted to take a call.
They used a combination of softphone and handsets, with some using the Chrome client and others using the windows client.
Looking at the logs, extensions appeared to be registering and deregistering every few minutes, and calls were showing as failed in the activity log.
firewall checker was showing random ports as failing the Full cone test. Watchguard firewall had all ports required open as per the documentation for 3CX.
The activity graph on the dashboard was the giveaway though, appeared to have "looping" ie multiple activities over the same time span.
The 3CX install was setup as a VM on a Hyper-V server.
And here we found the error. The Hyper-V server was providing time sync to the VMs and when the time on the Hyper-V server was checked against a known good source it was found to be incorrect. To remedy this, the alternatives were to:
1. install an NTP service on the Debian client, and ensure that the Hyper-V was not providing time sync services (under settings for the client server on the Hyper-V manager, under Integration services, un-tick the time synchronization service) ,
2. Better still make sure that the Hyper-V service was retrieving time from a known good source.
To complete option 2, run the following commands on the hyper-V server
net stop w32time
w32tm /config /syncfromflags:manual /manualpeerlist:"0.pool.ntp.org 1.pool.ntp.org 2.pool.ntp.org 3.pool.ntp.org"
w32tm /config /reliable:yes
net start w32time
This stops the windows time service, configures it to check with the pool.ntp.org servers ( please set these server names to the appropriate ones for your zone e.g. the ones above are for North America but other zones can be found at https://www.ntppool.org/en/)
It then sets your hyper-v server as a reliable source and finally restarts the windows time service to put the changes into effect.
With this in place, our 3cx server was able to pass the firewall test with no errors, and we think this has also prevented the issues with client disconnects.
So an additional step installing new servers should be to ensure that time sources are accurate and that your 3cx server is referencing these.
The client was getting constant "network disconnected" messages, particularly when they attempted to take a call.
They used a combination of softphone and handsets, with some using the Chrome client and others using the windows client.
Looking at the logs, extensions appeared to be registering and deregistering every few minutes, and calls were showing as failed in the activity log.
firewall checker was showing random ports as failing the Full cone test. Watchguard firewall had all ports required open as per the documentation for 3CX.
The activity graph on the dashboard was the giveaway though, appeared to have "looping" ie multiple activities over the same time span.
The 3CX install was setup as a VM on a Hyper-V server.
And here we found the error. The Hyper-V server was providing time sync to the VMs and when the time on the Hyper-V server was checked against a known good source it was found to be incorrect. To remedy this, the alternatives were to:
1. install an NTP service on the Debian client, and ensure that the Hyper-V was not providing time sync services (under settings for the client server on the Hyper-V manager, under Integration services, un-tick the time synchronization service) ,
2. Better still make sure that the Hyper-V service was retrieving time from a known good source.
To complete option 2, run the following commands on the hyper-V server
net stop w32time
w32tm /config /syncfromflags:manual /manualpeerlist:"0.pool.ntp.org 1.pool.ntp.org 2.pool.ntp.org 3.pool.ntp.org"
w32tm /config /reliable:yes
net start w32time
This stops the windows time service, configures it to check with the pool.ntp.org servers ( please set these server names to the appropriate ones for your zone e.g. the ones above are for North America but other zones can be found at https://www.ntppool.org/en/)
It then sets your hyper-v server as a reliable source and finally restarts the windows time service to put the changes into effect.
With this in place, our 3cx server was able to pass the firewall test with no errors, and we think this has also prevented the issues with client disconnects.
So an additional step installing new servers should be to ensure that time sources are accurate and that your 3cx server is referencing these.
