Service metrics

Every service reports its own request series: requests, errors, error rate, and response latency at p50 and p95.

miosa service metrics lumen API
miosa service metrics lumen API --range 24h --json

The same series is at GET /api/v1/projects/:id/services/:sid/metrics/series?range= with 15m, 1h, 6h, 24h or 7d, scope projects:read.

Reading it honestly

The server buckets the series and reports step_seconds: 60 for 15m and 1h, 300 for 6h, 900 for 24h, and 3600 for 7d. The full timeline comes back even in a quiet window, with null fields where nothing happened; a point is empty when requests is null, and the CLI prints a dash rather than a zero.

available names what is actually measured. For a service that is requests and latency; cpu, memory and network are not collected for services, and the CLI says so under the table instead of printing empty columns.