Anyone know why the WebRTC softphone registration state drops to ‘unavailable’ immediately after a successful Data Action execution targeting ServiceNow? We are running an Architect flow that triggers a REST API call to create an Incident record via ServiceNow’s table_api.do endpoint. The payload construction is verified correct, and the response code is consistently 201 Created. However, the agent’s softphone client (Genesys Cloud Agent Desktop v10.5.2.0) disconnects from the media server roughly 200ms after the Data Action completes. The browser console logs indicate a WebSocket closure with code 1006, but no explicit error is thrown in the Architect flow logs. This suggests the issue might be related to resource contention or a threading lock within the client-side SDK when handling the async callback from the external API. The environment is EU-West-1, and we have ruled out firewall timeout issues as the latency is minimal. We need to ensure the softphone remains registered while the backend integration processes. Is there a known conflict between WebRTC media session persistence and high-latency Data Action callbacks in the current SDK release?