last9-debug

name	last9-debug
description	Debug production issues using Last9 observability data

Debug Production Issues

When the user asks to debug a production issue, follow this workflow using Last9 MCP tools:

Investigation Flow

1. Start with the service summary — call get_service_summary to see which services have elevated error rates or latency.

2. Check for recent deployments — call get_change_events to see if a deploy or config change correlates with the issue timing.

3. Drill into the affected service — use get_service_performance_details for p50/p90/p95 latency, error rate, and throughput.

4. Check dependencies — call get_service_dependency_graph to see if the issue is in the service itself or a downstream dependency.

5. Look at exceptions — call get_exceptions filtered to the service. If exceptions are found, use get_service_traces with the service name and time range to find individual traces with full stack traces.

6. Search logs — use get_service_logs for the affected service and time range to find error-level log entries.

7. Check alerts — call get_alerts to see if any alerts have fired for this service.

Tips

• Always start broad (service summary) then narrow down
• Use did_you_mean if a service name lookup fails — it suggests correct entity names
• Correlate timestamps: compare exception first_seen/last_seen with change event times
• For database issues, use get_databases to check DB-level throughput and latency