Phones fail to Register - Setting to TCP registers instantly.

[Po3]

A hosted 3CX system installed on Linux (NOT OnPrem)
Version - 15.5.15502.6

Hello,

I am in the process of rolling out 3CX to our 400ish some employees across 15 ish sites. The initial rollout of some 60 phones went off without a hitch located at 3-4 offices and connected via Cisco Site to Site Tunnels with CISCO 5508X Firewalls with the current firmware version. The switches are a mix of Cisco 9500 and Ubiquiti edge switches. All phones (Yealink T42S, T48s, CP920, CP960) were provisioned via STUN with the default templates and firmware compliant with 3CX.

But as more phones are provisioned and registered to the system, I'm seeing more phones lose their register and fail to re-register. I found a few posts here suggesting changing the transport to the Server Host from UDP to TCP and that actually fixed the problem for the phones that would suddenly lose the service. But that still does not explain why phones that were previously provisioned and registered and WORKING for weeks would suddenly just fail. I came in yesterday to phone calls and emails saying all the phones at 3 offices were showing OFFLINE when they had been working fine for the last 2 months with no issues and nothing was done to the network over the weekend that would have changed anything.

My colleague has been in contact with Cisco Support to see if there is anything that would be limiting or blocking UDP but so far any changes we've made have not worked effectively. I do not have access to firewall or routers in the hosted environment so Im looking for any other possible things that could be caused of this.

Thanks in Advance.

 

[cobaltit]

Are the sites connected via S2S but you aren't connected to the 3CX system via S2S? If you have a VPN tunnel to your 3CX instance then you should be provisioning as local LAN and not STUN. And if you don't, you should be using SBCs, Since you have first hand knowledge of the infrastructure but don't have access to the equipment at the hosted environment and can't figure it out our guesses here aren't going to be much better. The fact that switching to TCP points to some issue with traffic tranversing an edge device. Could be running out of ephemeral ports. I'd strongly suggesting using SBCs and consulting with an experienced 3CX partner.