- Joined
- Aug 6, 2018
- Messages
- 52
- Reaction score
- 7
I'm having issues with my WAN failover...
3CX connects to the Internet via Firewall that has Uplink redundancy, being connected to 2 Uplinks.
There's a routing rule defined on the Firewall that routes all Traffic from local 3CX to Internet via WAN1,
if WAN1 goes offline all traffic from local 3CX to Internet is routed via WAN2.
3CX is configured as "Dynamic public IP" and STUN resolution works fine. PRO license.
Today, WAN1 failed and we successfully switched to WAN2.
Shortly after this, WAN1 uplink was restored and new connections were served via WAN1 again.
Except for the 3CX SIP registration.
Eventually, there was a mismatch of PublicIP detected via STUN by 3CX, and PublicIP used in SIP/SDP...
The result was no audio in outbound calls. Manually re-registering the Trunk solved the issue.
Before re-register, my VoIP provider showed WAN2_IP for the active registration, 3CX resolved its public IP to WAN1_IP.
After the re-register, both provider and 3CX resolved to WAN1_IP.
I guess the STUN resolution used WAN1 again after it was restored, since those are new connections every time.
The SIP registration was kept alive and never switched back to using WAN1.
So my question is: How should WAN failover be handled?
My provider only offers SIP via TCP/TLS, so switching to UDP is not an option.
Configuring a "hard" switch once WAN1 goes up again on the firewall is not desirable while WAN2 still serves connections.
The Re-Register Timeout is set to 600 (seconds?) in the SIP-Trunk settings, but the issue persisted way longer.
Should the 3CX Host be connected to both Uplink networks so it figures out its own failover?
SIP-ALG after all?
3CX connects to the Internet via Firewall that has Uplink redundancy, being connected to 2 Uplinks.
There's a routing rule defined on the Firewall that routes all Traffic from local 3CX to Internet via WAN1,
if WAN1 goes offline all traffic from local 3CX to Internet is routed via WAN2.
3CX is configured as "Dynamic public IP" and STUN resolution works fine. PRO license.
Today, WAN1 failed and we successfully switched to WAN2.
Shortly after this, WAN1 uplink was restored and new connections were served via WAN1 again.
Except for the 3CX SIP registration.
Eventually, there was a mismatch of PublicIP detected via STUN by 3CX, and PublicIP used in SIP/SDP...
The result was no audio in outbound calls. Manually re-registering the Trunk solved the issue.
Before re-register, my VoIP provider showed WAN2_IP for the active registration, 3CX resolved its public IP to WAN1_IP.
After the re-register, both provider and 3CX resolved to WAN1_IP.
I guess the STUN resolution used WAN1 again after it was restored, since those are new connections every time.
The SIP registration was kept alive and never switched back to using WAN1.
So my question is: How should WAN failover be handled?
My provider only offers SIP via TCP/TLS, so switching to UDP is not an option.
Configuring a "hard" switch once WAN1 goes up again on the firewall is not desirable while WAN2 still serves connections.
The Re-Register Timeout is set to 600 (seconds?) in the SIP-Trunk settings, but the issue persisted way longer.
Should the 3CX Host be connected to both Uplink networks so it figures out its own failover?
SIP-ALG after all?