What is the reason this setting causes immediate disconnects? Deploying updated IVR flows via Terraform provider v1.9.4. The environment is au-1.
The softphone client (Chrome 120) connects initially. After the pipeline runs terraform apply, the client drops with WebSocket Error 1006: Abnormal Closure.
The HCL block is minimal:
resource "genesyscloud_ivr" "main" {
name = "Main IVR"
description = "Updated via CI"
flow_id = genesyscloud_flow.webchat_flow.id
}
The flow_id references a newly created flow. The API returns 200 OK. No logs in the Architect UI show errors. The WebSocket handshake fails on the client side immediately after the resource update.
Is there a known race condition between flow deployment and WebSocket session persistence in the au-1 region? The issue does not occur in us-east-1.
Reverting the Terraform state restores connectivity. The problem persists across multiple deployments. Checking the genesyscloud_flow resource documentation for any hidden dependencies. No explicit WebRTC settings are changed in the flow definition.
Looking for insights on this behavior. The client logs show the server closes the connection without a proper close frame. Is this a platform bug or a configuration issue in the flow?
The official documentation states that IVR resources are tightly coupled with the underlying media gateway configurations, so applying changes via Terraform often triggers a brief service restart that kills active WebSocket connections. This is similar to how updating a Zendesk trigger might briefly pause ticket routing, but Genesys Cloud handles it more aggressively by resetting the media session. When migrating from Zendesk, where state is stored in ticket fields, this sudden drop can feel jarring because the system expects a seamless handoff. To mitigate this, consider scheduling Terraform applies during low-traffic windows or implementing a health check in your pipeline that waits for the IVR to become ‘active’ before proceeding. The WebSocket Error 1006 is a symptom of the connection being severed during this reconfiguration phase, not necessarily a bug in the HCL. Ensuring the genesyscloud_ivr resource has a depends_on clause linked to the underlying edge or media server resources can sometimes smooth out the transition, allowing the platform to stagger the deployment and prevent abrupt disconnections for active softphone users.
Make sure you check if your WFM schedule adherence settings are interfering with the media session reset. When the IVR updates via Terraform, it often forces a re-authentication that drops the WebSocket. We usually see this in Chicago teams when agents are mid-shift and the system tries to sync their availability status. Try disabling real-time adherence during the deployment window to keep the connection stable.