Transporter - Datadog

[gkatsanos]

I added pino and tried to see my logs in Datadog unsuccessfully. Later I read the section regarding the limitations and design decisions of Pino with regards to logging in process vs out of process.

I was wondering how this can work with regards to passing the transporter on the cl level (keys?). I'm afraid the impact of this design decision is that integrating the logger involves devops (instead of having everything at the code level).

Could you provide some insights? Do I have to use pino-multi-stream ?

Thank you for your time :)

[mcollina]

What are you using to ship logs to DataDog?

[gkatsanos]

Hey Matteo, appreciate the quick feedback. Bear with me if I dont get the facts right but we have a dockerized app and from my understanding stdout/stderr is auto-magically piped to DD. ( https://docs.datadoghq.com/agent/docker/log/?tab=containerinstallation ) . We don't do anything manually. BTW the application is a Nuxt (node SSR+Client side) app which logs both on client and on server side.

locally I see the Pino logs just fine, but deployed and dont get logs from DD (which is to be expected as I did not add the pino-datadog library). I just need to understand what else is needed.

[jsumners]

DataDog natively supports Pino. Load dd-trace before you load pino and DD will instrument Pino. This will in turn ship your logs to DD via their sidecar.

[gkatsanos]

@jsumners

const tracer = require('dd-trace')

export default function Datadog() {
  tracer.init({
    hostname: process.env.DD_AGENT_HOST,
    service: 'xxx',
    env: process.env.STAGE,
    logInjection: true,
  })
}

later..

const logger = require('pino')()
$axios.onRequest((config) => {
      logger.info({
        // date: new Date().toISOString(),
        http: {
          params: config.params,
          baseURL: config.baseURL,
          referer: config.headers.common.referer,
          url: config.url,
          method: config.method
        },
      }, `[${config.method}] ${config.url}`)

in the server logs / terminal :

(if DD natively supports Pino, why's pino-datadog necessary? ) I thought pino doesn't send logs anywhere unless you explicitely set it up to do so.

[jsumners]

if DD natively supports Pino, why's pino-datadog necessary?

Such questions should be asked to the DD support team.

I thought pino doesn't send logs anywhere unless you explicitely set it up to do so.

As of v6, Pino never ships logs anywhere. That is completely the user's responsibility. What I said is that DD supports instrumenting the Pino logger. See https://docs.datadoghq.com/tracing/connect_logs_and_traces/nodejs/

[gkatsanos]

@jsumners can you elaborate on "instrumenting" ?

[jsumners]

It's outlined in the link I provided in my last reply. Additionally https://opentracing.io/docs/best-practices/instrumenting-frameworks/

[gkatsanos]

@jsumners ah yes , instrumenting is not what I'm asking here. Instrumenting as seen above in my screenshot does happen, but I'm one step before that: I'm not sure how to pipe the logs to Datadog in the first place. I hope @mcollina will have some pointers on how to proceed with regards to pino-datadog and pino-multi-stream ?

[Fdawgs]

@gkatsanos Did you manage to resolve this?

[gkatsanos]

We used simple console log and we did our own formatting.

[github-actions[bot]]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.