KOMOJU Datadog

This is our in-house Datadog integration for Rust. It is similar to Datadog's own integrations for other languages, providing support for the following:

• Logs and tracing via tracing, with automatic correlation
• StatsD metrics
• Axum integration, automatic tracing for each request
• Simple header injection for distributed tracing across HTTP requests
• AWS ECS container correlation, container metrics
• SQLx integration, correct tracing for SQL queries

Usage

This library is configured through the Config struct, which automatically picks up DD_ environment variables. See the API documentation for a complete list of options.

use komoju_datadog::{Config, tracing::Tracer, statsd::StatsD};

fn main() {
    let o11y_config = Config::builder().service("my-service").build().expect("invalid config");
    StatsD::init_global(&o11y_config);
    Tracer::new(&o11y_config);

    // ...
}

Tracing

Once the tracer is initialized, tracing spans can be used to create spans on Datadog, either manually or via the instrument macro.

use tracing::{info, info_span, instrument};

#[instrument(err)]
fn answer(question: &str) -> Result<u64, &'static str> {
    info!(question, "Answering the question");

    let answer = {
        // This span lives for the duration of this block.
        let _span = info_span!("deep_thought").entered();
        std::thread::sleep(std::time::Duration::from_secs(u64::MAX));
        42
    };

    if answer != 42 {
        return Err("Got the wrong answer");
    }

    Ok(answer)
}

Certain span tags have special semantic meaning in Datadog:

• operation will be mapped to the operation name in Datadog, defaulting to the span name otherwise
• resource will be mapped to the resource name in Datadog

Errors

Errors can be rendered in Datadog by including the following tags:

• error.type
• error.msg
• error.stack (optional)

Metrics

Metrics can be sent to Datadog using the StatsD struct. A global instance is available via StatsD::global() after initialization.

use komoju_datadog::{Config, statsd::StatsD};

StatsD::init_global(&Config::builder().build().expect("invalid config"));

let _ = StatsD::global().incr("questions.answered", &["answer:42"]);

Features

This crate has several optional features which can be enabled:

feature	use case	requirements
ahash	slightly better performance for one extra dependency	-
aws_ecs	Running on AWS ECS/Fargate	Requires ECS_CONTAINER_METADATA_URI_V4 env var
axum	Using Axum web framework	-
gcp_gke	Running on GCP GKE	Requires POD_UID env var via Downward API
sqlx	Using SQLx for database queries	-

See the API documentation for more details and usage examples.

GKE/Kubernetes Setup

To enable container correlation on GKE, configure the Downward API in your pod spec:

env:
  - name: POD_UID
    valueFrom:
      fieldRef:
        fieldPath: metadata.uid