Flushing the Opentelemetry logger provider in `main()` results in a flushing timeout error

[wuuer]

When using actix-web-rt, flushing the Opentelemetry logger provider with GRPC tonic in main() results in a flushing timeout error. However, flushing within a router function completes without issue.

Expected Behavior

Flushing Opentelemetry logger provider succeeds.

Current Behavior

Flushing Opentelemetry logger results in a timeout error.

Possible Solution

use tokio runtime

replace actix-web::main with tokio::main

Steps to Reproduce (for bugs)

cargo.toml

[package]
name = "actix-web-sample"
version = "0.1.0"
edition = "2024"
 
[dependencies]
actix-web = "4.11.0"
futures-executor = "0.3.31"
tokio = { version = "1.47.1", features = ["macros","rt-multi-thread"] }
opentelemetry_sdk = "0.30.0"
opentelemetry = "0.30.0"
opentelemetry-stdout = "0.30.0"
opentelemetry-otlp = { version = "0.30.0", features = ["grpc-tonic"] }
opentelemetry-appender-tracing = "0.30.1"
anyhow = "1.0.99"
tracing = { version = "0.1.41", features = ["log"] }
tracing-appender = "0.2.3"
tracing-subscriber = { version = "0.3.20", features = ["env-filter","time"] }
tracing-opentelemetry = "0.31.0"

main.rs

use actix_web::{
    App, HttpRequest, HttpServer,
    web::{self, Data},
};
use anyhow::{Ok, Result, anyhow};
use opentelemetry_appender_tracing::layer::OpenTelemetryTracingBridge;
use opentelemetry_otlp::WithExportConfig;
use opentelemetry_sdk::logs::SdkLoggerProvider;
use tracing_subscriber::{Layer, layer::SubscriberExt, util::SubscriberInitExt};
 
async fn index(logger_provider: Data<SdkLoggerProvider>, req: HttpRequest) -> &'static str {
    tracing::info!("REQ: {:?}", req);
    // flushing logs
    if let Err(err) = logger_provider.force_flush() {
        return "flushing error";
    }
    "Hello world!\r\n"
}
 
#[actix_web::main]
async fn main() -> Result<()> {
    let otlp_endpoint = "http://localhost:4317";
 
    let otlp_log_exporter = opentelemetry_otlp::LogExporter::builder()
        .with_tonic()
        .with_endpoint(otlp_endpoint)
        .build()?;
 
    let logger_provider = SdkLoggerProvider::builder()
        .with_batch_exporter(otlp_log_exporter)
        .build();
    tracing_subscriber::registry()
        .with(
            OpenTelemetryTracingBridge::new(&logger_provider)
                .with_filter(tracing::level_filters::LevelFilter::INFO),
        )
        .init();
 
    if let Err(e) = database_setup() {
        tracing::error!("database setup error");
        logger_provider.force_flush()?;
        return Err(e);
    }
 
    HttpServer::new(move || {
        App::new()
            .app_data(web::Data::new(logger_provider.clone()))
            .service(web::resource("/").route(web::get().to(index)))
    })
    .bind(("127.0.0.1", 8080))?
    .run()
    .await?;
 
    Ok(())
}
 
fn database_setup() -> Result<()> {
    // simluate error
    Err(anyhow!("database connection error"))
}

Context

The application can not log the error msg before exit at main()

The batch_log_processor in opentelemetry sdk uses futures_executor::block_on() to flush the log, which appears to cause this issue. But why does flushing within a router function complete without issue?

Your Environment

• Rust Version (I.e, output of rustc -V): 1.90.0
• Actix Web Version: 4.11.0