Incorrect Registration Port After Internet Interruption

Status
Not open for further replies.

technomojo

Bronze Partner
Joined
Jul 10, 2015
Messages
22
Reaction score
2
After loss of Internet connection for whatever reason (client has had 3-5 outages a month over the last 3 months), 3CX registers on port 1024 (according to SIP provider Internode in Australia) instead of 5060 and fails.

Telephony works sporadically (somehow) until deregistering 3CX for 20 minutes, allowing the active registration to timeout at the SIP provider, reregistering, then all is fine until the next outage.

The best I can determine is that the original registration remains active after an interruption until it times out after 15 minutes or so. For some reason, the 3CX system doesn't reregister to that port, but to port 1024. What I'm trying to figure out is: why 1024? Is this a 3CX thing? Or maybe a SIP provider (Internode) thing? I can't find any documentation about 3CX utilising port 1024 whatsoever.

I have another client that uses the same Internode NodePhone product but doesn't have this issue. The only other variable in the equation I can think of is that the client with the problem also uses Internode for their internet connection, whereas the client without problems uses iiNet (Internode's parent company). Ookla's Speedtest actually identifies the Internode connection as iiNet and the iiNet connection as TPG (iiNet's parent company). Bottom line is, they probably use significantly different infrastructure. Not sure if it's a factor.

Thanks in advance.
 
ISP is also providing you their modem/router or not?
 
Despite being in the UK, I have been involved in a few installs for clients in Australia.

Firstly to get it out the way, Internode as a provider are not supported as a provider for 3CX. We have used Vocus and Twilio in the past. Is there anything that stands out about this provide in the way that the setup is done ?

3CX registers on port 1024 (according to SIP provider Internode in Australia) instead of 5060 and fails.

Have you tested this yourself ? run a PCAP trace from the PBX: https://www.3cx.com/docs/capture-network-traffic/ just out of interest I guess you made no changes to the default ports on PBX install ?

The randomness of the registration (and unknowing of ports for SIP) do make me thing that SIP ALG/Helper/Transformation maybe configured on your 3CX - local firewall, have you checked ?
 
The why #1024 is simple, it is the lowest free usable port according to standard. Below 1024 it is commonly known as reserved ports (for ftp http smtp etc).

The why the router has given it as dynamic nat port over a configured nat ports is more complicated. Many routers assign you a dynamic nat port if they have not a configured nat pair (sourceIP:source:Port) and in 99% of all firewalls, I have seen, you will not get the source port as the masquerading port you are sending from.

I can speak here for WatchGuard as a vendor, if you first install 3CX and then configure the NAT, you have to clear the masquerading table as the router holds a dynamic nat for this IP:Port pair already and does not use the configured NAT. May stop 3CX service. reboot the router. start 3CX service. May this fixes the situation. If not review the IPs in your system and the configuration in your router.
 
Status
Not open for further replies.

Forum statistics

Threads
111,935
Messages
589,823
Members
164,817
Latest member
Innovative Advisory