Need some troubleshooting help with a persistent timeout issue in our Architect flows involving agent scripting data actions. The specific flow triggers a lookup against our internal CRM via a Data Action when a call arrives on our Singapore BYOC trunks. The Data Action consistently fails with a 504 Gateway Timeout after exactly 5 seconds. This happens despite the underlying CRM API responding in under 200ms when tested directly via Postman. The environment is Genesys Cloud Edition 2023-12. The BYOC trunks are registered with a 100ms jitter buffer and show healthy SIP registration status. Packet loss is negligible at 0.01%. The issue appears isolated to calls routed through the SG region trunks, while US-based trunks process the same Data Action without errors. The Architect flow logs show the Data Action initiating but never receiving a response from the integration endpoint. We have verified that the firewall rules allow outbound traffic from the Genesys Cloud edge to our CRM endpoint. The error log in the Architect flow execution history cites a connection reset by peer. This suggests a network-level interruption rather than an application-layer failure. The SIP INVITE latency for these calls is within acceptable limits at 150ms. However, the subsequent Data Action lookup seems to hang until the timeout threshold is reached. We suspect a routing anomaly specific to the Singapore edge nodes handling BYOC traffic. Is there a known limitation with Data Action timeouts for BYOC trunks in the APAC region? We have tried increasing the timeout value in the Data Action configuration, but the platform enforces a hard 5-second limit for this type of lookup. The impact is significant as agents receive empty script variables, leading to manual lookups and increased handle time. The SIP CDRs show the call legs are established correctly. The issue persists across all 15 BYOC trunks in the SG region. We have checked for any recent changes in the carrier failover logic, but none were deployed in the last 30 days. Any insights into why the Data Action would timeout specifically for BYOC routed calls in this region would be appreciated.