Closed
Description
Problem
Jupyter doesn't have a way to capture enterprise gateway's client and kernel failures to be able to log them as events/metrics in the internal telemetry system. We would like to know
- How many gateway client failures took place in the last few days corresponding to number of users
- How many kernel failure took place.
- Error/exceptions from Enterprise gateway.
Proposed Solution
By leveraging jupyter_events, we can emit events for both GatewayClient and GatewayMappingKernelManger. On the receiver side, having a server extension that listens all the events and log them to internal telemetry system.