Struct MetricsManager 

pub struct MetricsManager { /* private fields */ }

Available on crate features metrics or otel-metrics only.

Metrics manager handling Prometheus and/or OTel exposition.

Implementations

impl MetricsManager

pub fn new(namespace: &str) -> Self

Create a new metrics manager with the given namespace.

pub fn with_config(config: MetricsConfig) -> Self

Create a metrics manager with custom configuration.

Installs the appropriate recorder(s) based on enabled features:

• metrics only: Prometheus recorder
• otel-metrics only: OTel recorder (OTLP push)
• Both: Fanout recorder (Prometheus scrape + OTel OTLP push)

pub fn counter(&self, name: &str, description: &str) -> Counter

Create a counter metric.

Automatically registers a MetricDescriptor in the manifest registry with empty labels and group = "custom".

pub fn counter_with_labels( &self, name: &str, description: &str, labels: &[&str], group: &str, ) -> Counter

Create a counter with label keys and group metadata for the manifest.

The labels parameter declares label key names for the manifest. The returned Counter is label-free – apply label values at recording time via metrics::counter!(key, "label" => value).

pub fn gauge(&self, name: &str, description: &str) -> Gauge

Create a gauge metric.

Automatically registers a MetricDescriptor in the manifest registry with empty labels and group = "custom".

pub fn gauge_with_labels( &self, name: &str, description: &str, labels: &[&str], group: &str, ) -> Gauge

Create a gauge with label keys and group metadata for the manifest.

pub fn histogram(&self, name: &str, description: &str) -> Histogram

Create a histogram metric with default buckets.

Automatically registers a MetricDescriptor in the manifest registry with empty labels and group = "custom".

pub fn histogram_with_labels( &self, name: &str, description: &str, labels: &[&str], group: &str, buckets: Option<&[f64]>, ) -> Histogram

Create a histogram with label keys, group, and optional buckets for the manifest.

pub fn histogram_with_buckets( &self, name: &str, description: &str, buckets: &[f64], ) -> Histogram

Create a histogram metric with custom buckets.

Note: The buckets parameter is captured in the manifest registry but currently ignored by the metrics crate at runtime (buckets are set globally at recorder installation time).

pub fn render(&self) -> String

Available on crate feature metrics only.

Get the Prometheus metrics output.

Returns the rendered Prometheus text format. Only available when the metrics feature is enabled.

pub fn render_handle(&self) -> Option<RenderHandle>

Available on crate feature metrics only.

Get a cloneable render handle for use in route handlers.

Returns a closure that renders the current Prometheus metrics text. The closure is Send + Sync + Clone, making it safe to move into axum route handlers or share across tasks via Arc.

Returns None if no Prometheus recorder is installed.

Example

let mut mgr = MetricsManager::new("myapp");
let render = mgr.render_handle().expect("recorder installed");

// Use in axum route
let route = axum::Router::new().route("/metrics", axum::routing::get(move || {
    let r = render.clone();
    async move { r() }
}));

mgr.start_server_with_routes("0.0.0.0:9090", route).await?;

pub fn set_readiness_check( &mut self, f: impl Fn() -> bool + Send + Sync + 'static, )

Set a readiness check callback.

When set, /readyz and /health/ready call this function and return 503 Service Unavailable if it returns false. Without a callback, these endpoints always return 200.

pub fn mark_started(&self)

Mark the service as started (startup probe passes).

Call this once init is complete. K8s startupProbe hits /startupz which returns 503 until this is called, then 200 thereafter. Separate from readiness – startup has a longer timeout for slow starters.

pub fn set_scaling_pressure(&mut self, sp: Arc<ScalingPressure>)

Available on crate features metrics and scaling only.

Attach a ScalingPressure instance.

When set and using start_server_with_routes, a /scaling/pressure endpoint is automatically added that returns the current pressure value.

pub fn set_memory_guard(&mut self, mg: Arc<MemoryGuard>)

Available on crate features memory and metrics only.

Attach a MemoryGuard instance.

When set and using start_server_with_routes, a /memory/pressure endpoint is automatically added that returns the current memory status.

pub fn update(&self)

Update process and container metrics.

pub async fn start_server(&mut self, addr: &str) -> Result<(), MetricsError>

Available on crate feature metrics only.

Start the metrics HTTP server.

Serves /metrics (Prometheus), /healthz, /health/live, /readyz, /health/ready endpoints.

Only available when the metrics feature is enabled (for scraping).

Errors

Returns an error if the server fails to start.

pub async fn start_server_with_routes( &mut self, addr: &str, extra_routes: Router, ) -> Result<(), MetricsError>

Available on crate features http-server and metrics only.

Start the metrics HTTP server with additional custom routes.

Serves the same built-in endpoints as start_server: /metrics, /healthz, /health/live, /readyz, /health/ready.

Additionally:

• If set_scaling_pressure was called, adds /scaling/pressure returning the current pressure value.
• If set_memory_guard was called, adds /memory/pressure returning memory status JSON.
• Any routes in extra_routes are merged (service-specific endpoints).

Requires both metrics and http-server features.

Errors

Returns an error if the server fails to start.

pub async fn stop_server(&mut self) -> Result<(), MetricsError>

Stop the metrics server.

Errors

Returns an error if the server is not running.

pub fn shutdown_otel(&mut self)

Available on crate feature otel-metrics only.

Gracefully shut down the OTel provider (flushes pending exports).

Call this before application exit to ensure all metrics are exported.

pub fn set_build_info(&self, version: &str, commit: &str)

Set application version and git commit for the manifest.

Uses interior mutability (writes through the registry’s Arc<RwLock>), so only &self is needed. Called automatically by dfe_groups::AppMetrics::new() if the metrics-dfe feature is enabled.

pub fn set_use_cases(&self, metric_name: &str, use_cases: &[&str])

Set operational use cases for a metric (by full prefixed name).

No-op if the metric is not found in the registry.

pub fn set_dashboard_hint(&self, metric_name: &str, hint: &str)

Set the suggested Grafana panel type for a metric (by full prefixed name).

No-op if the metric is not found in the registry.

pub fn registry(&self) -> MetricRegistry

Get a cloneable handle to the metric registry.

Use this to pass into route handlers. The handle is Clone + Send + Sync. Call before starting the server, consistent with the render_handle() pattern.

pub fn namespace(&self) -> &str

Get the namespace prefix (e.g. dfe_loader).

Used by dfe_groups metric structs to build labelled metric keys.