Struct rama::telemetry::opentelemetry::sdk::trace::BatchSpanProcessor

pub struct BatchSpanProcessor<R>
where
    R: RuntimeChannel,
{ /* private fields */ }

A SpanProcessor that asynchronously buffers finished spans and reports them at a preconfigured interval.

Batch span processors need to run a background task to collect and send spans. Different runtimes need different ways to handle the background task.

Note: Configuring an opentelemetry Runtime that’s not compatible with the underlying runtime can cause deadlocks (see tokio section).

Use with Tokio

Tokio currently offers two different schedulers. One is current_thread_scheduler, the other is multiple_thread_scheduler. Both of them default to use batch span processors to install span exporters.

Tokio’s current_thread_scheduler can cause the program to hang forever if blocking work is scheduled with other tasks in the same runtime. To avoid this, be sure to enable the rt-tokio-current-thread feature in this crate if you are using that runtime (e.g. users of actix-web), and blocking tasks will then be scheduled on a different thread.

Examples

This processor can be configured with an executor of your choice to batch and upload spans asynchronously when they end. If you have added a library like tokio or async-std, you can pass in their respective spawn and interval functions to have batching performed in those contexts.

use opentelemetry::global;
use opentelemetry_sdk::{runtime, testing::trace::NoopSpanExporter, trace};
use opentelemetry_sdk::trace::BatchConfigBuilder;
use std::time::Duration;

#[tokio::main]
async fn main() {
    // Configure your preferred exporter
    let exporter = NoopSpanExporter::new();

    // Create a batch span processor using an exporter and a runtime
    let batch = trace::BatchSpanProcessor::builder(exporter, runtime::Tokio)
        .with_batch_config(BatchConfigBuilder::default().with_max_queue_size(4096).build())
        .build();

    // Then use the `with_batch_exporter` method to have the provider export spans in batches.
    let provider = trace::TracerProvider::builder()
        .with_span_processor(batch)
        .build();

    let _ = global::set_tracer_provider(provider);
}

Implementations

impl<R> BatchSpanProcessor<R>

where R: RuntimeChannel,

pub fn builder<E>(exporter: E, runtime: R) -> BatchSpanProcessorBuilder<E, R>

where E: SpanExporter,

Create a new batch processor builder

Trait Implementations

impl<R> Debug for BatchSpanProcessor<R>

where R: RuntimeChannel,

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

Formats the value using the given formatter.

impl<R> SpanProcessor for BatchSpanProcessor<R>

where R: RuntimeChannel,

fn on_start(&self, _span: &mut Span, _cx: &Context)

on_start is called when a Span is started. This method is called synchronously on the thread that started the span, therefore it should not block or throw exceptions.

fn on_end(&self, span: SpanData)

on_end is called after a Span is ended (i.e., the end timestamp is already set). This method is called synchronously within the Span::end API, therefore it should not block or throw an exception.

fn force_flush(&self) -> Result<(), TraceError>

Force the spans lying in the cache to be exported.

fn shutdown(&self) -> Result<(), TraceError>

Shuts down the processor. Called when SDK is shut down. This is an opportunity for processors to do any cleanup required.

fn set_resource(&mut self, resource: &Resource)

Set the resource for the log processor.