Does anyone know if the Genesys Cloud BYOC edge enforces a strict timeout window for downstream REST API calls initiated via Data Actions before the SIP INVITE is fully processed?
We are experiencing a peculiar issue where calls routed through a specific Architect flow fail with a 403 Forbidden error from the carrier, but only when the flow executes a Data Action that triggers a ServiceNow incident creation via webhook. The environment is running Genesys Cloud v2024.10 with ServiceNow Madrid release.
The flow logic is straightforward:
- Receive SIP INVITE on BYOC Trunk.
- Extract
FromandToheaders. - Execute Data Action
Create_Servicenow_Incident. - Wait for HTTP response.
- Route to IVR.
When the ServiceNow REST API endpoint takes longer than 3000ms to respond, the Data Action times out. Immediately following this timeout, the BYOC edge appears to drop the SIP session or send a malformed response, resulting in the carrier rejecting the call with 403 Forbidden. If I bypass the Data Action or mock a fast response, the call connects perfectly.
The webhook payload structure matches the documentation exactly, using standard JSON formatting. I have verified that the ServiceNow instance is reachable from the Genesys Cloud network via TCP 443. The issue seems to be that the SIP signaling state machine in Genesys Cloud does not handle the asynchronous failure of the Data Action gracefully, leading to a premature termination of the SIP dialog.
I have checked the Telephony logs and see the 403 coming back from the SBC, but no corresponding error in the Architect logs other than the Data Action timeout. Is this a known limitation of the BYOC edge processing pipeline? Should we be implementing a retry mechanism within the ServiceNow side instead? Any insights on how to decouple the SIP signaling from the REST API latency would be appreciated.