Make pino Log Messages Appear with the Correct Status in Datadog

Posted on

If you’re using pino with Datadog’s Log Management product, you might notice that all of your logs are reported at the info level, regardless of what log level you’re using in your code.

screenshot of the datadog log UI showing all info-level messages

Having all logs categorized as info-level messages makes it challenging to find what you’re looking for.

The Problem

By default, pino uses a numerical serialization to represent the level of the message.

For example, this code produces the following output:

import pino from "pino";

const logger = pino();
logger.info("info message");
logger.warn("warn message");
{"level":30,"msg":"info message","time":1603492511716,"pid":66621,"hostname":"71b5c8be.local"}
{"level":40,"msg":"warn message","time":1603492511716,"pid":66621,"hostname":"71b5c8be.local"}

Datadog’s default Status Mapper does not understand how to convert these numbers over to their string representation (like info, warn.)

The Fix

Luckily, there’s a pretty easy fix for this problem. According to the Status Remapper docs, we need to report the log level as a string representation of the level vs. the numbered defaults used by pino.

You’ll need to configure this in two places:

  1. Pino Configuration
  2. Datadog Pipeline

Pino Configuration

When creating your logger, use the formatters configuration to override the default level serialization. This will produce string output instead of the numerical representation.

Here’s the same code from before, with a level formatter.

import pino from "pino";

const logger = pino({
  formatters: {
    level(level) {
      return { level };
    },
  },
});

logger.info("info message");
logger.warn("warn message");
{"level":"info","msg":"info message","time":1603492741115,"pid":66835,"hostname":"71b5c8be.local"}
{"level":"warn","msg":"warn message","time":1603492741115,"pid":66835,"hostname":"71b5c8be.local"}

Datadog will be able to understand these new string representations of the log level after we configure the pipeline.

Datadog Pipeline

Next, we’ll set up a Log Pipeline with a Status Remapper Processor in your Datadog configuration.

  1. View your Log Pipelines in Datadog
  2. Create a new pipeline for the service that uses Pino. You can skip this step if there’s already a pipeline created for your service. screenshot of the datadog UI creating a new log pipeline
  3. Add a Status Remapper processor on the level attribute. screenshot of the datadog UI creating a new processor

Now, new messages should appear with the proper status identified in the Datadog UI. Happy log spelunking!

screenshot of the datadog log UI showing the proper log-levels being reported
Find an issue?
Open a pull request against my blog on GitHub.
Ben Limmer