SBC restarting on inbound calls

[RossPCS]

We have a remote office with 17 phones all Yealink 12 - T27g and 5 - T21P_E2 phones. We started with a raspberry pi 3b+ as the controller and it seemed to work fine during testing. But when we set up the locations rules things have not worked as well. They want all 17 phones to ring when an inbound call comes in. I thought this is easy and added all the phones. Well, when we do this the SBC restarts and the BLFs on the phones start blinking and you can't answer the call. In the SIP Trunks window of the SBC will say Register Sent will be 0 and start again. We lowered the number of extensions in the list down to 6 extensions and that seems to help but still restarts. We also moved the SBC to an i5 windows box to see if that would help and it did not. We left if on the windows box to see if we could possibly troubleshoot it better. We also lowered all the BLFs to less than 10 on each phone included the park lines. Any help would be appreciated.

Netgear Firewall
The location has a Comcast cable modem with 95 Mbits down and 16 up.
Server version Windows 16.0.622

 

[RossPCS]

We are at 1 day without any issues. We also increased the number of phones in the ring group and it is working. Seems like problem solved. I think the TCP setting corrected it.

 

[JohnS_3CX]

Ok that's very good. Is there any chance that UDP traffic for 5090 is blocked at the SBC site from a firewall?

 

[cobaltit]

Or is it a 'smart' firewall that does some sort of packet inspection/IDS/IPS and that is getting triggered when calls come in?

 

[JohnS_3CX]

You can also try to update your SBC to 16.2.24 which is our latest beta and a close release candidate.

 

[Nick B]

We are now to 2.5 days without interruption or "re-registering" of SBC. We have TLS for the security setting in the SBC settings.

My issue appears to have been a time sync issue with our v-host. We moved it to another hypervisor on the same network and all is still going strong. No matter what settings we tried on the other host, the live graph was still jumping around and events happening out of order. A call was recorded as missed before it ever rang. That is the only change that was made. Firewall, SBC version, all other settings stayed the same. Never underestimate the impact of clock issues.