Metrics reference
Send metrics
Section titled “Send metrics”Use OTLP /v1/metrics, custom JSON /metrics or Prometheus remote-write /api/v1/write at https://otlp.siteqwality.com. Collector, Prometheus and StatsD guides have practical configs.
The custom v2 JSON shape contains resource and a metrics array. A metric has name, type, optional unit/description, temporality/monotonicity where relevant, and points. Gauges use numeric value; sums preserve cumulative/delta semantics; histograms use count, optional sum/min/max, increasing bounds and corresponding bucket counts. The adapter also accepts supported legacy counter/gauge shapes. Do not convert a distribution to percentile gauges if you need later reaggregation.
{ "resource": {"service.name": "checkout", "deployment.environment.name": "production"}, "metrics": [{"name": "queue.depth", "type": "gauge", "unit": "{message}", "points": [{"value": 42, "attributes": {"queue": "orders"}}]}]}Read routes
Section titled “Read routes”| API-relative path | Method | Purpose |
|---|---|---|
/metrics/ | GET | Names, types, units and series metadata. |
/metrics/ | GET | One metric’s metadata. |
/metrics/ | GET | Bounded label values. |
/metrics/ | GET | Series/cardinality analysis. |
/metrics/ | POST | Cursor-paged names by prefix/range. |
/metrics/ | POST | Series by metric/range, service/env and string tag maps. |
/metrics/ | POST | Samples for explicit returned series IDs. |
/metrics/ | POST | Typed multi-query/formula workspace. |
series accepts metric, range, optional tags, resource_tags, service, env, limit and cursor. Point requests contain metric, range, up to 100 decimal-string series_ids and optional interval. Do not round IDs. Use the query reference for complete typed query examples and formulas.
Raw points preserve kind, timestamp, start time, unit and flags. Histogram data preserves explicit or exponential layout, count and optional sum/min/max. no_recorded_value is a gap, receive_time means a fallback time, and no_sum prevents inventing a mean. Counter resets and predecessor carry affect rates. Missing carry or expired inputs can produce incomplete quality rather than zero.
The workspace aggregates time before space. Distribution merge is separate from scalar aggregation; merge distributions before calculating percentiles. It supports grouped queries, formulas, previous-period comparison, histogram heatmaps and real exemplar trace links. Exemplar target traces can be sampled away, restricted or expired.
Billing and retention
Section titled “Billing and retention”Metrics are billed by active series, not bytes or samples; see billing and retention for allowances and retention. Response source and precision identify the representation used. See also histograms and privacy.