Is it possible to replicate the seamless click-to-call functionality we had in Zendesk Talk using the Genesys Cloud WebRTC softphone without encountering persistent 408 Request Timeout errors? We are currently migrating our digital support team from Zendesk to Genesys Cloud, and while the ticket-to-interaction mapping is working, the real-time voice component is failing. In Zendesk, the agent just clicked a button, and the call connected. Here, the softphone loads, but the outbound call never rings the destination.
The environment is a standard EU-1 deployment. We are using the Genesys Cloud Web SDK version 2.5.1 within a custom React wrapper that mimics our old Zendesk widget. The error logs show a WebSocket connection failed followed by the 408 timeout on the /api/v2/telephony/phone/connections endpoint. We have verified that the user’s phone number is correctly assigned in the admin console and that the user is online.
Here is the relevant configuration snippet we are using to initialize the softphone client:
softphone_config:
tenant_id: "eu1-abc123"
auth_token: "{{ACCESS_TOKEN}}"
options:
enable_rtc: true
debug_mode: true
logging_level: "verbose"
routing:
default_skill: "Support-English"
queue: "General-Queue"
We noticed that in Zendesk, the underlying SIP signaling was abstracted away completely. In Genesys Cloud, we are seeing the WebRTC handshake initiate, but it stalls at the ICE candidate gathering phase. Is there a specific firewall rule or STUN/TURN server configuration we are missing that differs from the default Zendesk setup? The agents are behind a corporate proxy that allows standard HTTPS traffic. We need the softphone to work reliably before we cut over the final batch of agents next week. Any guidance on debugging the ICE failure or mapping the Zendesk click-to-call logic to the GC WebRTC flow would be appreciated.