Skip to content
Docs Pydantic Docs

Send standard library logs to Logfire

Attach Logfire to Python’s built-in logging module and your existing log output shows up in Logfire as structured logs (individual timestamped records of something that happened), next to the traces (the full journey of one request through your app) from the rest of your code. Every message written through the standard logging module (including logs from third-party libraries that use it) becomes a Logfire log, with no change to how you call logger.info(...) and friends.

What you’ll capture

  • Each standard-library log message
  • Its severity level (debug, info, warning, error, and so on)
  • Its timestamp
  • The logger name and any structured fields attached to the record

Before you start

You’ll need a Logfire project. Open Add data in your project (top navigation) and follow the setup for your language: it signs your machine in with logfire auth (a browser sign-in, no token to copy) and, for production or other languages, creates a write token (the credential your app uses to send data). New to Logfire? Start with Getting Started.

Installation

Install logfire:

Terminal
pip install logfire

No extra library needed: this works with Python’s built-in logging module.

Usage

Call logfire.configure() to connect to your project, then attach LogfireLoggingHandler to your logging configuration so every log record is also sent to Logfire. You can wire it up with either basicConfig() or dictConfig():

main.py
from logging import basicConfig, getLogger

import logfire

logfire.configure()
basicConfig(handlers=[logfire.LogfireLoggingHandler()])

logger = getLogger(__name__)

logger.error('Hello %s!', 'Fred')
# 10:05:06.855 Hello Fred!

The LogfireLoggingHandler will emit log records to the Logfire instance, unless instrumentation is suppressed, in which case a fallback handler will be used (defaults to StreamHandler, writing to sys.stderr).

Verify it worked

Run your program, then open the Live view in the Logfire web app. Within a few seconds you’ll see your message as a record.

Advanced

Quiet a noisy logger

Logging can be too verbose, especially from third-party libraries. You can raise a logger’s level to suppress logs that are less important. Let’s see an example with the apscheduler logger:

main.py
import logging

logger = logging.getLogger('apscheduler')
logger.setLevel(logging.WARNING)

In this example, we set the log level of the apscheduler logger to WARNING, which means that only logs with a level of WARNING or higher will be emitted.

Disabling urllib3 debug logs

As instrumentation is suppressed when sending log data to Logfire, unexpected DEBUG logs can appear in the console (through the use of the fallback handler), emitted when performing the API request to send logs.

To disable such logs, a filter can be used:

main.py
import logging
from logging import DEBUG, basicConfig

import logfire

logfire.configure()
logfire_handler = logfire.LogfireLoggingHandler()

urllib3_filter = logging.Filter('urllib3')
# Disable urllib3 debug logs on the fallback handler
# (by default, writing to `sys.stderr`):
logfire_handler.fallback.addFilter(lambda record: not urllib3_filter.filter(record))

basicConfig(handlers=[logfire_handler], level=DEBUG)

Troubleshooting

Not seeing your logs in Logfire? Check that logfire.configure() ran first, your write token is set, and the LogfireLoggingHandler is attached to your logging configuration (via basicConfig(), dictConfig(), or added to the relevant logger).

Reference