Access the Jaeger query API at /api/traces?service=<name>&limit=20&lookback=1h to retrieve recent traces for a service
Use /api/trace/{traceID} to fetch all spans for a specific trace ID and inspect parent-child relationships via references
Configure the Jaeger backend storage plugin by setting SPAN_STORAGE_TYPE environment variable to elasticsearch, cassandra, or badger for local dev
For Elasticsearch storage, tune the es.num-shards, es.num-replicas, and index-date-layout settings to balance shard count with retention window
Enable adaptive sampling by deploying the jaeger-agent with the remote sampling endpoint pointing to jaeger-collector to dynamically adjust per-operation rates
Known gotchas
Jaeger's Elasticsearch index names are date-partitioned; querying across long time windows requires multi-index search, which requires the correct index prefix configuration
The in-memory storage backend (badger) is not suitable for production; it loses all data on process restart
Jaeger adaptive sampling state is stored in-memory in the collector; collector restarts reset learned sampling rates and may cause temporary over-sampling
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp