Troubleshooting scenarios
Move from a symptom to the evidence, then confirm the outcome.
Confirm you are on the right instance before following a scenario. The examples come from fault-injection scripts on a lab cluster; you do not need to reproduce them in production.
- Long-running SQL: find a statement that has been running for a while and what it waits on.
- Lock waits: find who waits, follow the verify line to the blocker, and confirm the waits are gone.
- Read a real result: a WARN in text and JSON, line by line.
Find a statement that has been running for a while and see what it is waiting on.
Find who waits, follow the verify line to the blocker, decide, and confirm the waits are gone.