ezgh obs metrics query <promql> [--time t | --from t [--to t] [--step d]] [flags]Evaluate a PromQL expression over the metrics of every dataset (environment) in the project that you may query. Without range flags it’s an instant query, at --time (default now). With --from, --to or --step it’s a range query: from --from (default 1h ago) to --to (default now), every --step (default: a step that gives about 250 points).
Metric names are Prometheus-style (http.server.request.duration in seconds is http_server_request_duration_seconds), and every series has the labels environment, application, job and instance. To read one environment, say so in the query: up{environment="production"}. Times are RFC 3339, a date, or a duration ago (2h, 7d); steps and timeouts are durations (15s, 5m, 1h) or seconds. -o json prints the Prometheus API’s own answer.
Part of ezgh obs metrics.
Examples
ezgh obs metrics query 'sum by (job) (rate(http_server_request_duration_seconds_count[5m]))'
ezgh obs metrics query 'histogram_quantile(0.99, sum by (le) (rate(http_server_request_duration_seconds_bucket{environment="production"}[5m])))' --from 6h --step 5mFlags
| Flag | Type | Default | Description |
|---|---|---|---|
--from |
string | a range query’s start (default 1h ago) | |
--step |
string | a range query’s resolution, like 15s or 5m (default: about 250 points) | |
--time |
string | an instant query’s time (default now) | |
--timeout |
string | evaluation timeout, at most the server’s (30s) | |
--to |
string | a range query’s end (default now) |
This command also takes the global flags.