Metrics And Dashboards

CloudGrid separates metric exploration from saved dashboard composition.

Route	Job
/metrics	Discover metric names, inspect descriptors, query series, test aggregations, group by attributes, and inspect exemplars.
/dashboards	Build saved visual workspaces from typed metric, log, trace, and live trace widgets.

The split is intentional. Metrics are technical telemetry evidence. Dashboards are reusable project configuration.

Metric Explorer

The metric explorer uses:

• Query.metricNames for metric discovery;
• Query.metricSeries for ordinary series queries;
• Query.richMetricSeries for rich dashboard metric computations.

Storage-read owns aggregation compatibility:

Metric kind	Supported aggregations
gauge	avg, min, max, count
sum	sum, rate, count
histogram	avg, count, sum, p50, p90, p95, p99
exponential_histogram	avg, count, sum, p50, p90, p95, p99
summary	avg, count, p50, p90, p95, p99 when quantiles exist

Unsupported combinations return a canonical validation error. The frontend does not calculate missing aggregations locally.

Dashboard Workspace

Dashboards are control-plane records scoped to one project. Dashboard widgets are typed. They do not store executable code, raw SQL, raw SurrealQL, arbitrary JSON, bearer tokens, provider secrets, or external embeds.

Widget execution paths:

Widget kind	Execution path
metric_timeseries	Query.metricSeries
metric_stat	Query.metricSeries
metric_table	Query.metricSeries
metric_rich	Query.richMetricSeries
log_table	Query.logs
trace_table	Query.traces
live_trace_table	Subscription.liveTraces

Dashboard pins are user preferences returned by dashboard contracts and pin mutations. They are not browser-local shortcuts.

Computation Boundary

diagram

The frontend may draft layout and display choices. It must not compute metric rates, percentiles, formulas, trace counts, log counts, or dashboard widget telemetry from broad raw records.

Next Step

Use Metrics guide for exploration and Dashboards guide for saved workspaces.