Trying to understand why our predictive routing campaigns consistently show a queue depth of zero in the Genesys Cloud dashboard, despite active SIP sessions flowing through our 15 BYOC trunks in the Singapore region. The calls are successfully registered and routed, yet the real-time analytics API returns empty arrays for active interactions within the specific outbound campaign queues. This discrepancy suggests a disconnect between the trunk-level signaling and the predictive routing engine’s state tracking.
The environment involves Genesys Cloud release 2403 with custom SIP headers added for carrier-specific failover logic. When inspecting the interaction logs via GET /api/v2/interactions, the status updates to ‘completed’ immediately upon SIP 200 OK receipt, bypassing the expected ‘queued’ state duration. The SDK version used for polling is 3.2.1 for Python. No specific error codes appear in the predictive routing campaign logs, but the WFM integration reports a 400 Bad Request when attempting to reconcile these phantom zero-depth queues with scheduled agent capacity.
Has the community seen similar latency or state-sync issues between BYOC trunk registration and predictive routing queue metrics in ap-southeast-1? The SBC logs confirm proper INVITE handling, so the issue likely resides in how Genesys Cloud maps the trunk connection to the predictive queue object. Detailed SIP trace logs are available if needed for correlation with the internal interaction IDs.