These two applications both use VM’s in the cloud:
I’ve run into this twice already. The first time, I just replaced the P-Serices phones with S-Series phones, and manually configured the buttons. That seems to have worked fine.
I am now addressing this same issue at a second site. I’ve studied the possible remedies that might allow me to keep using P-Series phones and DPMA. From what I’ve ready, it seems possible, but more than I want to tackle for a 3 phone system.
I’m at a loss trying to understand your questions or the two applications you are referencing. Would you provide more specifics?
Thanks for bringing that to my attention Alan, I can see the deficiencies in my original posts.
CGNAT breaks the commuications on telephone using DPMA, the phones may work for days, and then will suddenly fail, lose their registration, and will not pull down their configuration files. Sometimes Call Logs will not work on the phones usiing DPMA and this is telling becasue Call Logs are updated on the fly rather than being stored in the phone, the CAll Log data stream is sent to the phone upon request.
I have one deployment where P-330 phones just stopped working, and this deployment is in the cloud but is accessed by the DPMA enabled P-330 Phones via Starlink. This deployment, and another cloud deployment using Starlink with DPMA failed on the same day. My solution for the first deployment was to replace the P-330 phones with S-705 phones, and manually configure them-the customer was happy with this solution. So, I drove to the second site ready to replace the P-330 phones with S-705 phones, and the problem resolved itself enroute, so I didn’t do anything and those phones have worked for a while since then.
This latest (Third issue) is another FreePBX17 cloud deployment, where the DPMA enabled P-330 phones are connected to a cellular wifi router that uses CGNAT. I have sporadic issues with registration, calls logs not working, and phones provisioning from this Internet connection, not consistantly, but too much to be acceptable. I can take the phones to my shop, or to my home, where I have unfiltered Cox Internet without any NAT, and the phones work perfectly, and will sometimes work for a while on the cellular wifie (that uses CGNAT) after being provisioned elsewhere.
I’ve done quite a bit of research on this issue, this is a recent search describing what I’ve been dealing with: Google Search .
It wasn’t my first or only search about this topic. I’ve been led to possible solutions that adjust timers (Keep Alives), using third party VPN, using a proxy, or using VPN from the phone to the PBX, among others. I’m trying to avoid re-engineering the wheel to solve this issue with this one site now that only has two affected phones.
And, I’d like to be better prepared in the future with applications like this. I told the client NOT to get cellular Internet, but she did it anyway, and here is what that caused. IF there is a simple fix-cool. It seemed that one of the solutions, using VPN, may solve the issue as this may keep a P2P connection open at alls times, but I’ve been very unlucky trying to get the builtin VPN to work on the best of connections, and I’m not particularly excited about getting buried in that for the likes of two phones.
I believe I’ve tried modifying default ports, these are the current settings:
Service Insecure Port Secure Port
HTTP Provisioning 84 not forced to 1443
Rest/GraphQLAPI 83 not forced to 2443
Restful Phone Apps disabled 6443 (I believe I modified this from default)
I hope this better explains the issue. When the phones work, the audio is acceptable, service is acceptable. But then… It’s not a throughput issue, but relates to issues specifically attributable to CGNAT.
Curious, Dejected, Enthusiastically Pressed for Time )
Chad Fournet
…when I wrote “at my office without any NAT”, I specifically meant without any Carrier NAT. In my office and Home I have static IPs with Sonicwall Routers, and thos connections work great…
If they are on cellular you will have to get a business plan with static ip to avoid double nat / CGNAT. On tmobile I think a static ip is $5mo. With Starlink, you can get a public ip on the right plan, enable that and it should resolve your issue.
This is an excellent summary. Our Sangoma P320 behaves erratically on Starlink. It’s only sometimes, it’s certainly not every day, and I’m not on one of their routers either. It’s not an essential phone so I don’t care that much when it’s idly fetching. By the time I can be bothered looking at it properly, the space invaders have fixed it.
It’s not on VPN, I should try moving it on to…
Never mind, it’s working again. This particular phone has a relatively straightforward connection to a Vultr instance with a static public IP.
We shifted a whopping 600GB a month off our fibre connections by moving 4x mobiles, my son’s Xbox, and our TVs onto Starlink. It’s unstable several times a day, but then we are asking a lot of it at 100Mbps.
Edit: Responsive firewall is enabled. I have now put the phone on VPN and will report back.
P320 connected over on-board VPN all night. Starlink IP didn’t change. I restarted the phone, it was then stuck on Fetching. Disabled the FreePBX firewall and hit Retry. Phone registered almost straight away, still on VPN.
Does the P-Series expose the network IP during boot?
Seems to me this has nothing to do with Starlink itself. It’s an issue with Responsive Firewall not entering the boot sequence soon enough and allowing an untrusted, dynamic IP.
Frequent changes to the dynamic public IP, coupled with a phone restart of some description, indicates a timing/state problem during boot rather than Starlink causing the failure.
Does the P-Series expose the network IP during boot?
Yes. During bootup, possibly after receiving network config from DHCP, P-Series reaches out to get a configuration, which will [hopefully] contain further registration information.
If Responsive Firewall isn’t familiar with the phone’s IP anymore (for whatever reason), I suspect it could be blocked.
That’s consistent with the fw whitelist entry being tied to an active registration — reboot drops registration, entry lapses (or was never promoted past the 90-second window), and the phone’s provisioning fetch arrives from a now-unknown IP.
Thanks to everyone who replied, and confirmed the issue.
I used an OVH trunk for few months without issue. So the VoIP with Starlink can works.
I don’t know if it’s specific for P series.