Overview and Stats

Observe the health of the system through:

• high-level runtime state such as fires, success rate, latency, and circuit-breaker count for the selected window
• active services and their current liveness, version, SDK metadata, and manifest_hash
• run throughput broken into ok, failed, and halted outcomes
• DLQ pressure, both as aggregate depth and as attention-worthy failures

In practice, the overview pages combine /api/v2/stats/* with /api/v2/services and /api/v2/fleet/releases. Start with the stats totals to answer "is the system healthy right now?", then jump to Runs, DLQ, or Fleet depending on what moved.

Two details matter today:

• Attention items currently flag failing flows from recent run stats and open circuit breakers. They are a triage list, not a full diagnosis.
• The recent_dlq_1h and recent_dlq_24h fields in service summaries are placeholders right now. Use stats and the DLQ endpoints for real counts.