WebSocket 1006: abnormal closure at /v2/websockets/rtc/agent
The outbound predictive campaign just locked the pacing multiplier at 1.9x the second the WebRTC softphone sessions started dropping. Running the 2024-12-10 tenant patch on US-East, and the admin UI completely freezes when trying to adjust the pacing rules under Outbound > Campaigns > Pacing. Console spits out a 422 UNPROCESSABLE_ENTITY on POST /api/v2/outbound/campaigns/{id}/pacing/multiplier every time the value touches 2.0x. Agents are sitting on idle screens while the dialer keeps firing calls into the eastern callable window. Abandonment rates already jumped to 14% before compliance had to pause the list.
Softphone settings look fine in the admin UI. TURN servers are configured, STUN hits the public endpoints, and the desktop client works perfectly. Switching to the browser-based WebRTC softphone triggers the drop. Network traces show the ICE candidates gathering, then the SDP exchange just hangs. The dialer thinks the agent is available, so it keeps routing predictive calls. Once the WebSocket drops, the pacing engine floors to 1.9x and refuses to budge. DNC sync runs clean in the background, so it’s not a compliance block.
Cleared the browser cache, disabled all extensions, even spun up a fresh Chrome profile. Same 422 on the pacing endpoint. The abort rate is eating into the daily production targets hard. The softphone media path seems to be colliding with the predictive pacing engine whenever the multiplier crosses that 2.0x threshold. Mic stays hot but the call never bridges. Doing jack all on the queue while the pacing rule locks up.
curl -X POST https://mycompany.genesyscloud.com/api/v2/outbound/campaigns/{id}/pacing/multiplier
-H “Content-Type: application/json”
-d ‘{“multiplier”: 2.2}’