Querying Data

BlameTrail provides three ways to query your observability data: the service overview, the metrics explorer, and the incident observability tab.

Service overview

Navigate to Observability > Overview and select a service. The overview shows:

• Provider health — Status of each connected provider
• Key metrics — Sparkline charts for mapped metrics (request rate, error rate, latency, etc.)
• Recent deploys — Deploy events overlaid on the timeline for correlation

The overview auto-refreshes every 60 seconds.

Metrics explorer

Navigate to Observability > Metrics to query metrics with full control:

• Service selector — Choose which service's providers to query
• Time range — Select from presets (1h, 6h, 24h, 7d) or the maximum allowed by your plan
• Environment — Filter by environment if your mappings use environment-based filters
• Query presets — One-click presets for common queries (Request Rate, Error Rate, P95 Latency, CPU Usage, Memory, etc.)
• Custom query — Enter any PromQL or Datadog query directly

Results are displayed as interactive time-series charts with deploy events and active incidents overlaid for correlation.

Incident observability tab

On any incident detail page, click the Observability tab to see:

• Metrics around the incident time window (15 minutes before the incident started to 15 minutes after resolution, or now for open incidents)
• Logs from the same time window, merged and sorted across all mapped providers
• Deploy markers overlaid on charts for quick correlation

This view is automatically populated from the service's mapped providers. No manual querying is needed.

Log search

Navigate to Observability > Logs to search logs across providers:

• Filter by severity level (error, warn, info, debug)
• Full-text search within log messages
• Cursor-based pagination for browsing large result sets
• Forward and backward time navigation

Caching

Query results are cached in Redis for 30 seconds (60 seconds for incident context). Cursor-based paginated queries bypass the cache to ensure fresh results when scrolling through logs.