SIP 408 Request Timeout on BYOC Trunk Failover in Singapore Region

Context:
Environment: Genesys Cloud EU, NICE CXone Hybrid
Region: Asia/Singapore
Trunks: 15 BYOC trunks managed via Terraform
SDK: Python 3.9

We are experiencing intermittent SIP 408 Request Timeout errors specifically during the failover sequence of our BYOC trunks. The primary carrier in the Singapore region drops registrations, triggering the failover logic to the secondary carrier. While the registration fails over correctly, approximately 15% of outbound calls initiated within the first 30 seconds of failover receive a 408 from the Genesys Cloud edge before being routed to the secondary SIP URI. The logs indicate the INVITE is sent but no 100 Trying is received within the configured timeout window. This behavior is consistent across all 15 trunks in this region, suggesting a potential edge processing delay rather than a carrier-specific issue.

Question:
Can anyone clarify the expected behavior of the SIP session timer and retransmission logic during an active BYOC trunk failover? We have adjusted the SIP transport timeout in the trunk configuration, but the 408 persists. Is there a specific setting in the outbound routing or the trunk SIP credentials that influences the initial handshake latency during a state change? We need to determine if this is a platform limitation or a configuration error in our failover script logic.

The documentation actually says… check the sip_timeout in your trunk resource. Set it to 30000. The default is too low for Singapore latency during failover. Also verify register_timeout. Update your HCL and run terraform apply. This usually fixes the 408 spikes.