Skidder

Skidder will drag your logs to where they need to go. A small, uniform and extensible logging library, implemented across major technologies.

Installation

pip install git+https://github.com/novemberfiveco/skidder-python@3.0.0

or using poetry

poetry add git+https://github.com/novemberfiveco/skidder-python@3.0.0

Usage

Setup

Configure logging using the configure_logging() function once at the beginning of your program. You can pass a value for the component field as required by N5's standards.

from skidder import configure_logging

configure_logging(component='my-backend-service')

# note that we configure logging outside the AWS Lambda handler function
def handler(event, context):
    ...
 

Logging events

Now that the logger is configured, just use the get_logger() function from structlog to get the correctly configured logger anywhere inside your project.

from structlog import get_logger

log = get_logger()

log.debug("This is a debug log")
log.info("This is an info log")
log.warn("This is a warn log")
log.error("This is an error log")
log.critical("This is a critical log")

When running this program locally, you receive pretty formatted log output:

2021-07-12T18:04:48.248064Z [info     ] This is an info log                             component=foo environment=None source=application type=event

However, in an actual environment, you should set the ENV environment variable, which will enable JSON logging:

{"component": "foo", "environment": "dev", "level": "info", "message": "This is an info log", "source": "application", "timestamp": "2021-07-12T18:07:08.674527Z", "type": "event"}

Logging exceptions

You can pass an exception parameter to include a stacktrace in the log event:

try:
    raise Exception
except Exception as e:
    log.error("API call to external service X failed due to timeout", exception=e)

Including additional fields

You can also pass extra parameters to populate the data field with additional structured information:

log.info("Synchronizing product catalog to store", store_id='store-123')

Setting a request ID on your log events

Whenever possible, you should set the requestId field on log events to link them to a specific request for easy investigation.

You can manually set or unset:

from structlog import bind_request_id, unbind_request_id

log.info("This log event is not part of a request")

def handler(event, context):
    bind_request_id("my-request-123")
    log.info("This log event is part of my-request-123")
    unbind_request_id()

For convenience, there is also a decorator that automatically (un)binds a request ID whenever your function is called. By default, a new UUID is generated.

from structlog import request_context_logging

log.info("This log event is not part of a request")

@request_context_logging
def handler(event, context):
    log.info("This log event is part of a request")

You can also pass a function that provides the request ID (for example you might want to get it from an HTTP request header to continue a request spanning multiple distributed services):

from structlog import request_context_logging

log.info("This log event is not part of a request")

def request_id_from_header(event, context):
    return event["headers"]["X-Trace-ID"]

@request_context_logging(request_id_provider=request_id_from_header)
def handler(event, context):
    log.info("This log event is part of a request")

Adding a LUMIGO_LOG prefix to error logs

When you want to have your error logs get noticed by Lumigo, they need a LUMIGO_LOG prefix. This can be enabled by enabling the corresponding flag:

from skidder import configure_logging

configure_logging(component='my-backend-service', enable_lumigo_prefix=True)
 

Changelog

• 3.1.1
  • removed Lumigo prefix from warning logs
• 3.1.0
  • added Lumigo prefix to error logs
• 3.0.0
  • rename repo to Skidder
• 2.0.2
  • set minimum python version to 3.7
• 2.0.1
  • Added check for IS_LOCAL env variable to check local development env
• 2.0.0
  • Pin structlog dependency (because using private function that is not part of public structlog API and can change without notice)
  • Set minimum logging level to DEBUG when in dev environment
  • Log uncaught exceptions
  • Intercept & process STDLIB logging
  • Populate type field dynamically
  • Populate source field dynamically
  • Refactor into functions
  • Allow binding request ID field
  • Allow binding component field
  • Disable JSON logging & enable pretty logging when executing locally
  • Nest extra fields under the 'data' key
• 1.0.0
  • basic setup according to N5 standard.