Toggle sidebar Toggle sidebar

Intermittent audio loss during calls

Status
Not open for further replies.

interactive7

Customer
Joined
Jun 11, 2015
Messages
13
Reaction score
1
I've been running 3CX on a Windows Server 2012 VM for a few years with no real issues.
I recently switched over to a Linux VM due to the the Windows OS requirement change for v16.5 concerning the APN issue.

Now, calls are intermittently losing audio for up to 10 - 15 seconds (both ends). Internal and external calls have the issue.

Bandwidth is not an issue.
The PBX is an on premise installation and extensions are local on a 1GB LAN.
Same hypervisor as the previous Windows VM.

Has anyone else had this experience?
 
Changes made. Not sure if it works yet. Collecting feedback from users.
 
  • Like
Reactions: JohnS_3CX
Hi,

What hypervisor are you using?
 
The NTP changes did not solve the issue. Calls are still experiencing the same intermittent audio loss. Not all calls.

Call recordings capture the audio loss also. I can hear the pauses in the recordings.
 
cobaltit said:
Have you seen this? https://www.3cx.com/docs/installing-microsoft-hyper-v/

Of particular note (besides NTP) would be the VM generation and the Broadcom/VMQ issue.
Click to expand...

I followed this guide concerning the NTP settings.
I did not set a Static MAC address on the Debian VM since no static MAC was set on the previous Windows VM that was used. Note again, the previous Windows VM did not have this issue. It's also the same hypervisor, only change is Windows VM to Debian VM for 3CX.

I will set a static MAC and see if this helps.
 
There's another issue, not sure if it may be related or if I should post another topic..

After a reboot of the PBX VM:
  1. No audio on calls.
  2. Softphone (3CX iOS - latest version) shows "connecting" but does not connect. I can however see the status of extension on the softphone (it's getting presence information).
  3. If I restart the 3CX Media service, the softphone connects and I have audio for calls.
 
Last edited:
interactive7 said:
I followed this guide concerning the NTP settings.
I did not set a Static MAC address on the Debian VM since no static MAC was set on the previous Windows VM that was used. Note again, the previous Windows VM did not have this issue. It's also the same hypervisor, only change is Windows VM to Debian VM for 3CX.

I will set a static MAC and see if this helps.
Click to expand...

I completed a new install, followed guides for NTP and static MAC.
New install has old static local IP.

Same intermittent audio issue.

At this point I may be forced to update to Windows 2016 and revert to a Windows install.
 
Hi @interactive7

Is there any chance that since you are in a VM there is a "noisy neighbor" effect when the Host is busy?

Or is the Broadcom NetXtreme being used?
 
JohnS_3CX said:
Hi @interactive7

Is there any chance that since you are in a VM there is a "noisy neighbor" effect when the Host is busy?

Or is the Broadcom NetXtreme being used?
Click to expand...

The Broadcom NetXtreme is not used.
I will try to move to a standalone machine to verify if resources is an issue.
Currently allocated 2GB RAM.
3CX dashboard shows over 1GB available most times and CPU usage is negligible.

However, I've encountered another issue that may point to resources being a problem.

interactive7 said:
There's another issue, not sure if it may be related or if I should post another topic..

After a reboot of the PBX VM:
  1. No audio on calls.
  2. Softphone (3CX iOS - latest version) shows "connecting" but does not connect. I can however see the status of extension on the softphone (it's getting presence information).
  3. If I restart the 3CX Media service, the softphone connects and I have audio for calls.
Click to expand...

I did some more investigating on the above and it seems if I restart 3CX services after the VM has booted, things work as expected (barring the call audio issue).

There are two interfaces on the VM. The first is the default GW for the LAN.
The second is dedicated to a SIP trunk connection. (ISP delivers the SIP trunk on a dedicated switch port. This switch is separate from the LAN switch.)

When the VM boots, the SIP off of the secondary NIC does not work. Other trunks (via the default NIC) work. I can ping the SIP, the outgoing/incoming calls connect, but there is no audio.
Restarting the 3CX media service restores audio on this SIP trunk.

Not sure if there is a race condition happening on VM bootup between the 3CX services and the initialization of the secondary interface.

I will try to move to a dedicated box and continue testing.
 
JohnS_3CX said:
Ah I see, you might have an unsupported setup currently, which can be addressed with some quick settings.

The second adapter (for SIP) must NOT have a gateway defined as per our guides.

Multiple internet gateways (used by ISPs that offer VoIP Services)
https://www.3cx.com/blog/docs/network-configurations-supported-3cx-phone-system/
Click to expand...

This is my network config


# The primary network interface
allow-hotplug eth0
iface eth0 inet static
address 192.168.253.4/24
gateway 192.168.253.3
# dns-* options are implemented by the resolvconf package, if installed
dns-nameservers 192.168.253.28 192.168.253.1 192.168.253.2


allow-hotplug eth1
iface eth1 inet static
address 10.180.100.30/30
post-up ip route add 10.180.100.28/30 dev eth1 src 10.180.100.30 table cwsip
post-up ip route add default via 10.180.100.29 dev eth1 table cwsip
post-up ip rule add from 10.180.100.30/32 table cwsip
post-up ip rule add to 10.180.100.30/32 table cwsip
post-up ip route add 10.160.16.5/32 via 10.180.100.29 dev eth1


I suppose removing the line
" post-up ip route add default via 10.180.100.29 dev eth1 table cwsip"
should resolve the issue?
I admit I am not very familiar with Debian so it may well be a config issue on my part.

I will try removing and continue monitoring.
 
The SIP network settings was supplied by the ISP. After consulting with them I was able to do a simpler config and still reach the SIP server.

New network configs are

auto eth0
iface eth0 inet static
address 192.168.253.4/24
gateway 192.168.253.3
dns-nameservers 192.168.253.28 192.168.253.1 192.168.253.2

auto eth1
iface eth1 inet static
address 10.180.100.30/8

SIP is eth1 and the SIP server is reachable.

Audio issue is still persists with the new network settings. Not isolated to SIP trunk calls. Experienced with internal calls also and calls on other trunks that run on the main network.

I placed the VM on a dedicated host for the PBX to try to rule out resource contention on the host.
VM allocated 4GB, 4 cores, only VM on the host.
Host is now 2016 Hyper-V Server. VM is a new install of 3CX-Debin with restored settings from backup.
Will have to monitor now for the audio issue again.

Even on the new host, I still have to restart the " SIP/RTP Tunneling Proxy " service after bootup to get mobile extensions to connect...
 
have you tried with 10.180.100.30/24 instead of /8 ?
 
JohnS_3CX said:
have you tried with 10.180.100.30/24 instead of /8 ?
Click to expand...

Reaching the SIP server is not the issue. Also, that subnet will not work. ISP gave a network 10.180.100.30/30 to be used.GW being 10.180.100.29. Since I am unable to set two gateways as per 3CX, the alternative was to use 10.180.100.30/8 which works. Using /24 puts it on a different subnet, which does not reach the SIP at 10.160.16.5
 
The issue seems to have been related to a CA certificate.
On Windows, the VM was part of a domain network with an internally trusted CA issued certificate for 3CX.

Moving to Debian, the CA was not trusted.

Adding the internal CA as a trusted CA on the Debian VM seems to have solved the issues.
There were no complaints about the intermittent audio loss today, after I made the changes.
Also there was no need to restart 3CX SIP/RTP Tunneling Proxy and Media Services after bootup.
 
Status
Not open for further replies.

Latest Posts

Forum statistics

Threads
111,953
Messages
589,914
Members
164,850
Latest member
masvty
3CX Logo Footer Copyright © Sitemap Privacy
Top Bottom