Pino and dd-trace not working together

[furstenheim-goodnotes]

Expected behaviour Pino should integrate nicely with datadog Actual behaviour I'm seeing something similar to https://github.com/DataDog/dd-trace-js/issues/2189. Logs of pino do not include the datadog fields

Steps to reproduce

I'm using "dd-trace": "4.11.0" and "pino": "8.14.2"

// index.ts
// Import tracer first thing so that pino can be hooked
import tracer from './metrics/Tracer.js';

// import { pino } from "pino";

(async function main () {
    await tracer.trace('Some Trace', async () => {
        console.log(tracer.scope().active())
        const logger = pino({
        } )
        logger.info('Starting service...')    })
})()

// Tracer.ts
import tracer from 'dd-trace'
// BTW, types are wrong here, it claims that tracer.init does not exist. I've also tried tracer.tracer.init(..) but same result as this bug
// @ts-ignore
tracer.init({
    logInjection: true,
})

// also tried
/*
tracer.tracer.init({
    logInjection: true,
})
*/

// Note that docs https://docs.datadoghq.com/tracing/trace_collection/dd_libraries/nodejs/?tab=containers suggest exporting just tracer but that seems to break
export default tracer.tracer

This logs:

// DatadogSpan { ... 
// {"level":30,"time":1691057718718,"pid":59863,"hostname":"mm","msg":"Starting service..."}

From the logs I can see that the scope of the tracer is defined. I'm even creating the logger inside the scope just for the testing.

Environment

• Operation system: Linux
• Node.js version: Node 18
• Tracer version: "dd-trace": "4.11.0"
• Agent version:
• Relevant library versions: "pino": "8.14.2"

[furstenheim-goodnotes]

BTW, with winston it does work. So I guess that all the hooks to datadog are in place

[zomgbre]

I'm seeing the exact same thing.

[ChrisBellew]

I think I know what is broken.

This works

import pino from "pino";

This does not work for log injection even though the pino logger itself still works.

import { pino } from "pino";

Don't ask me how many days it took to figure that out :)

[tlhunter]

@khanayan123 do you think this could be an ESM bug based on the comment that @ChrisBellew left?