Does anyone know why our Bring Your Own Cloud (BYOC) edge instance is dropping connections precisely when the WFM schedule publish job completes? We are running Genesys Cloud version 2024-2.0 with a custom BYOC deployment on AWS us-east-1. The edge cluster consists of three nodes, all reporting healthy status via the health check endpoint before the publish window. However, immediately after the WFM module pushes the new schedule to the agent desktops, we see a spike in WebSocket disconnections on the edge.
The issue seems tied to the surge in presence updates. When agents log in to view their newly published shifts, the edge receives a burst of WebSocket handshake requests. Instead of establishing the connection, the edge returns a 408 Request Timeout. This happens consistently every Friday at 4:00 PM Central Time, which is our standard publish window. We have increased the thread pool size on the edge container to 500 threads, but the timeouts persist.
Here is the error snippet from the edge container logs:
2024-05-10T16:00:05.123Z ERROR [wss-handler] Connection failed for agent_id: 98765432
Error: net::ERR_CONNECTION_TIMED_OUT
Context: WebSocket handshake interrupted during WFM sync event
Source: BYOC-Edge-Node-03
We verified that the load balancer in front of the edge nodes is not rejecting traffic. The health checks pass, but the application-level WebSocket layer seems to choke on the concurrent load generated by the schedule publish. We suspect the WFM publish webhook is triggering too many simultaneous presence updates, overwhelming the edge’s ability to maintain state for new connections. Has anyone configured rate limiting on the WFM publish event to stagger these updates? We need to ensure agents can log in without hitting these timeouts.