Phones Unregistering

Status
Not open for further replies.

telis

Premier Customer
Advanced Certified
Joined
Aug 13, 2007
Messages
238
Reaction score
65
We have been struggling with an issue with handsets unregistering on 3CX V15 running on Windows 2016 Server guest OS on a Windopws 2012R2 Hyper-V host for a couple of weeks. There are 127 handsets and a smaller number of 3CX clients in CTI mode.

We have previously though this was something to do with the May Cumulative Microsoft security update on the host (because that update caused other issues) but now we are not so sure - this may have been co-incidence. Unfortunately we didn't have the luxury of doing clean troubleshooting where we only changed one thing at a time, but the issue now seems to have gone away.

One of the changes we made was to give the 3CX server a separate Virtual Switch which only it used - effectively giving it a dedicated NIC.

Has anyone come across anything similar of does anyone have any comments that could help us shed some light on what happened?
 
We have been struggling with an issue with handsets unregistering on 3CX V15 running on Windows 2016 Server guest OS on a Windopws 2012R2 Hyper-V host for a couple of weeks. There are 127 handsets and a smaller number of 3CX clients in CTI mode.

We have previously though this was something to do with the May Cumulative Microsoft security update on the host (because that update caused other issues) but now we are not so sure - this may have been co-incidence. Unfortunately we didn't have the luxury of doing clean troubleshooting where we only changed one thing at a time, but the issue now seems to have gone away.

One of the changes we made was to give the 3CX server a separate Virtual Switch which only it used - effectively giving it a dedicated NIC.

Has anyone come across anything similar of does anyone have any comments that could help us shed some light on what happened?

An easy test to fix this might be to build a Linux instance and restore your config to that. We have had a lot of issues with 2016 server over the past few months in general. It is still early days for it. Linux is bulletproof.
 
Well as I say it is working now. I thought it was most likely an issue with the Hyper-V host, which we'd still use if we were using a linux guest.

I supposed I could create a Linux guest and try it without a dedicated NIC on the Hyper-V host and see what happens. It could be a problems in the windows NIC drivers for Hyper-V that's not present in linux NIC drivers for Hyper-V. In general we haven't had any trouble running 3CX on Windows server, apart from this issue of course, for several years.
 
Last edited:
Hello @telis,

Please note that in the required network configuration for the PBX, it is strongly recommended that you have a dedicated NIC for the system, which as you stated was one of the changes you made before the issue was resolved.

Thank you for sharing your resolution with us.
 
Obviously the guest always had a dedicated NIC....
 
Hello @telis

It is not a requirement, but a recommendation. It would depend on the traffic you have on the other virtual machines. Obviously, if all VM's have a lot of traffic, the PBX will need a dedicated NIC.
 
This documentation has nothing to do with what we're talking about.

The use of host machine in this context does not mean Virtual host.

The guest and always was set up with a single NIC inside the VM OS in the way described above.

What I changed did that *may* have fixed the issue was to dedicate a physical NIC on the Virtual Hyper-V to a single Virtual Switch and then make the 3CX guest VM be the only PC attached to the Virtual switch. That is the sense in which the physical NIC is dedicated to the guest on the host.

I now again suspect this is an OS issue and not a 3CX issue and it is related to a bug introduced in the networking stack by the May update -
https://support.microsoft.com/en-us/help/4019216/windows-server-2012-update-kb4019216 - since there were messages in the logs about ephemeral port exhaustion as experienced in widely differing scenarios by people here - https://www.reddit.com/r/sysadmin/comments/6bjtb1/ms_2012_r2_clusterfile_shares_serious_performance/
 
I am running my one hosts on 2016 however not on Hyper-V and dont have any issues, even under bulk load.

But the combo 2012 hyper/v and 2016 got me thinking.
https://technet.microsoft.com/library/dn792027.aspx
they do not state to support the OS over never updated the page as it ends with the tech preview.

in the hyper/v guide of 2016 the OS is listed as supported
https://docs.microsoft.com/en-us/wi...uest-operating-systems-for-hyper-v-on-windows

As you pointed already out that there is an port exhaustion, this can result from an incompatible Integration Service. But thats just a maybe
 
Status
Not open for further replies.