Call Quality Issues

[JohnS_3CX]

Hi @kriley87

Please ensure/confirm the following applies:

• Hyper-V VM has a Static MAC configured and Static IP
• Make sure you are not using the "Legacy Network Adapter" in Hyper-V
• Install NTP service by running the Debian shell command sudo apt-get install ntp

 

[kriley87]

Hi @kriley87

Please ensure/confirm the following applies:

• Hyper-V VM has a Static MAC configured and Static IP
• Make sure you are not using the "Legacy Network Adapter" in Hyper-V
• Install NTP service by running the Debian shell command sudo apt-get install ntp

@JohnS_3CX
Hi John-

This issue has been resolved by moving this deployment onto dedicated hardware. It was later brought to my attention that our other 3CX customer running on the same virtual host has recently begun experiencing the same problems and we are about to move them over to dedicated hardware as well to resolve this issue as quickly as possible. I'm not sure why this is happening, this other customer has been running on this Hyper-V host now for 6 months without any issues until recently.

I checked the Hyper-V settings you recommended and confirmed the following:
1.) The VM had a static IP address configured
2.) We were not using a legacy network adapter. We were using the normal network adapter option which is configured in a NIC team.
3.) I installed the NTP service but that didn't solve the issue.

Anymore suggestions? I have confirmed that the Hyper-V host has plenty of available resources when it comes to processing power, RAM, disk space, etc. I would love to be able to deploy 3CX in a virtual environment as deploying on dedicated hardware each and every time just doesn't scale well from a providers standpoint.

Thanks!

 

[LariFari]

With hyper-v "time synchronization" integration service active the cpu load graph on the 3cx dashboard will often draw loops, as if time went backwards for a while.
To avoid timing issues with the hyper-v guest, in "Integration Services" disable hyper-v "time synchronization", rely on using ntp instead.
You could also try reducing the cpu count to one and give 100% of declared cpu resources to the 3cx machine.

 

[JohnS_3CX]

Obviously you witness an effect, and there is as always an underlying cause. So don't be so fast to rule out anything. There is a long path between your extension dialling a number, and the voice finally reaching the other end, a long path indeed:

1. VM
2. OS
3. NIC Driver
4. Virtual NIC
5. Bridge on Host
6. Host machine
7. Host OS
8. Host Driver
9. Host NIC
10. .... and in all the above we have not even reached the local network yet, not a single packet has yet to leave the door and nevermind the firewalls, ISPs, physical link layer etc that come afterwards..

As you see it not so easy to pinpoint, things are getting extremely complex when you want to get into details. Since you built a dedicated machine, connect it to the exact same network as the previous VM Host and see if all is good. At least you can work you way back the from Host machine upwards to the VM. Captures can also reveal network stats if you want to get more into WireShark, you might be able to see jitter, latency, out of order packets etc