Users need the ability to query (using system tables) instrumentation data captured for various system level events. Actual data we capture for these events can be arbitrary as these are specific to each event.

eg.

select * from system.snapshot_fetch should return event_time, event_data at a high level. A sample output for such a query is below,

event_time                |  event_data
2024-11-01T10:24:00.000   |  { start_time: "2024-11-01T10:23:59.500", time_taken_ms: "500", total_fetched: 430 }
2024-11-01T10:35:00.000   |  { start_time: "2024-11-01T10:34:59.500", time_taken_ms: "500", total_fetched: 12 }

These events will,

• not be persisted (any server restarts will lose these events)
• bring back up to N items (N will default to 1000)
• have arbitrary event_data (use Maparray internally)