Struct PrometheusMetrics 

pub struct PrometheusMetrics { /* private fields */ }

Prometheus metrics backend for solti runners.

Implements MetricsBackend and exposes runner-level metrics in Prometheus format. Runners call the trait methods during task lifecycle.

Metrics

Metric	Type	Labels	Description
solti_runner_tasks_started_total	Counter	runner	Task spawn events
solti_runner_tasks_completed_total	Counter	runner, outcome	Task completion events
solti_runner_task_duration_seconds	Histogram	runner, outcome	Per-attempt execution duration
solti_runner_errors_total	Counter	runner, error	Runner setup/teardown errors

Labels

All label sets have low, bounded cardinality:

Label	Values	Cardinality
runner	subprocess, wasm, container	Low
outcome	success, failure, canceled, timeout	Low
error	cgroup_prepare_failed, backend_config_failed, spawn_failed, module_load_failed (from RunnerErrorKind)	Low

Duration histogram buckets

Buckets (seconds): 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30, 60, 300, 1800, 3600.

Also

• PrometheusSubscriber is a supervision-level metrics from the event stream.
• Registry is a shared registry for unified /metrics endpoint.

Implementations

impl PrometheusMetrics

pub fn new(registry: Arc<Registry>) -> Result<Self, Error>

Create a new metrics backend, registering all counters and histograms into the given Registry.

Primary constructor — mirrors the shape used by other backends in this crate (PrometheusSubscriber::new, PrometheusApiMetrics::new, PrometheusDiscoverMetrics::new).

pub fn new_isolated() -> Result<Self, Error>

Create a new metrics backend with an isolated registry.

Convenience for tests / standalone use. Most agents share a single registry across collectors via Self::new.

pub fn new_with_registry(registry: Arc<Registry>) -> Result<Self, Error>

👎Deprecated since 0.0.2:

use PrometheusMetrics::new(registry) — same signature, consistent with the other backends

Deprecated alias of Self::new.

pub fn gather(&self) -> Vec<MetricFamily>

Gather all metrics for exposition.

pub fn registry(&self) -> &Arc<Registry>

Get reference to underlying prometheus registry.

Trait Implementations

impl Debug for PrometheusMetrics

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl MetricsBackend for PrometheusMetrics

fn record_task_started(&self, runner_type: RunnerType)

Increments solti_runner_tasks_started_total{runner=<runner_type>}.

fn record_task_completed( &self, runner_type: RunnerType, outcome: TaskOutcome, duration_ms: u64, )

Records a task completion event.

Updates two metrics:

• solti_runner_tasks_completed_total{runner, outcome} - incremented by 1.
• solti_runner_task_duration_seconds{runner, outcome} - observes the duration converted from milliseconds to seconds.

The outcome label is derived from TaskOutcome::as_label: success | failure | canceled | timeout.

fn record_runner_error( &self, runner_type: RunnerType, error_kind: RunnerErrorKind, )

Increments solti_runner_errors_total{runner=<runner_type>, error=<error_kind>}.

Called for runner setup/teardown errors (e.g. spawn failures), not for task-level failures which go through record_task_completed.