Currently, when a Fabric SQL Endpoint or Warehouse hits the 2,048 max connection limit due to application leaks (e.g., JDBC/TDS), the workspace enters a "hard deadlock."
The Critical Gap in P-SKU: Unlike F-SKU (Fabric) users who can at least restart their capacity via the Azure Portal, P-SKU (Power BI Premium) customers lack a capacity restart button. If a rogue application saturates the gateway:
T-SQL is blocked: Admins cannot connect to run KILL commands.
No Self-Service Reset: There is no UI-based mechanism to "reboot" the endpoint or flush sessions.
Support Dependency: We are forced to wait hours for 1st-level support to escalate to SRE teams just to clear a session table. It took 24 hrs to resolve and all admin & users not able to login and consume data.
The Proposal: I am calling for a REST-API or Portal-based "Emergency Valve" in the Fabric/Power BI Admin portal:
"Force-Disconnect All Sessions": A UI trigger that instructs the Gateway to drop all active TDS handles for a specific Lakehouse/Warehouse without requiring a T-SQL handshake.
"Emergency Single-User Mode": A toggle to block all non-admin incoming connections while we remediate rogue application code.
Why this is urgent: As we migrate enterprise workloads to Fabric, a single coding error in a JDBC-connected portal (internal app, external app) shouldn't be able to paralyze a P-SKU capacity with no path for admin recovery.
