Struct iota_metrics::histogram::HistogramVec

pub struct HistogramVec { /* private fields */ }

Represents a collection of histograms for managing multiple labeled metrics. The HistogramVec struct allows for sending HistogramMessage instances via a channel to record data in a particular histogram, providing a way to track different metrics concurrently.

Implementations

impl HistogramVec

Reports the histogram to the given prometheus gauge. Unlike the histogram from prometheus crate, this histogram does not require to specify buckets It works by calculating ‘true’ histogram by aggregating and sorting values.

The values are reported into prometheus gauge with requested labels and additional dimension for the histogram percentile.

It worth pointing out that due to those more precise calculations, this Histogram usage is somewhat more limited comparing to original prometheus Histogram.

On the bright side, this histogram exports less data to Prometheus comparing to prometheus::Histogram, it exports each requested percentile into separate prometheus gauge, while original implementation creates gauge per bucket. It also exports _sum and _count aggregates same as original implementation.

It is ok to measure timings for things like network latencies and expensive crypto operations. However as a rule of thumb this histogram should not be used in places that can produce very high data point count.

As a last round of defence this histogram emits error log when too much data is flowing in and drops data points.

This implementation puts great deal of effort to make sure the metric does not cause any harm to the code itself:

• Reporting data point is a non-blocking send to a channel
• Data point collections tries to clear the channel as fast as possible
• Expensive histogram calculations are done in a separate blocking tokio thread pool to avoid effects on main scheduler
• If histogram data is produced too fast, the data is dropped and error! log is emitted

pub fn new_in_registry( name: &str, desc: &str, labels: &[&str], registry: &Registry, ) -> Self

pub fn new_in_registry_with_percentiles( name: &str, desc: &str, labels: &[&str], registry: &Registry, percentiles: Vec<usize>, ) -> Self

Allows to specify percentiles in 1/1000th, e.g. 90pct is specified as 900

pub fn with_label_values(&self, labels: &[&str]) -> Histogram

Creates a new Histogram with the specified label values. The function takes a slice of label strings, converts them into a HistogramLabelsInner structure, and returns a new Histogram instance that shares the same data channel as the original.

Trait Implementations

impl Clone for HistogramVec

fn clone(&self) -> HistogramVec

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.