Python log collection
Overview
Use your favorite Python logger to log into a file on your host. Then monitor the file with the Datadog Agent to send your logs to Datadog.
Python logs are quite complex to handle, mainly because of tracebacks. They are split into multiple lines which make them difficult to associate with the original log event.
To address this use case it is strongly recommended to use a JSON formatter when logging in order to:
- Ensure each stack_trace is properly wrapped into the correct log
- Ensure all the attributes of a log event are properly extracted (severity, logger name, thread name, etc.)
Here are setup examples for the following logging libraries:
Connect your service across logs and traces
If APM is enabled for this application, connect your logs and traces by automatically adding trace IDs, span IDs, env
, service
, and version
to your logs by following the APM Python instructions.
Note: If the APM tracer injects service
into your logs, it overrides the value set in the agent configuration.
Once this is done, the log should have the following format:
2019-01-07 15:20:15,972 DEBUG [flask.app] [app.py:100] [dd.trace_id=5688176451479556031 dd.span_id=4663104081780224235] - this is an example
Then configure the Datadog Agent to collect Python logs from the file.
Log into a file
Usage example with JSON-log-formatter:
import logging
import json_log_formatter
formatter = json_log_formatter.JSONFormatter()
json_handler = logging.FileHandler(filename='/var/log/my-log.json')
json_handler.setFormatter(formatter)
logger = logging.getLogger('my_json')
logger.addHandler(json_handler)
logger.setLevel(logging.INFO)
logger.info('Sign up', extra={'referral_code': '52d6ce'})
The log file contains the following log record (inline):
{
"message": "Sign up",
"time": "2015-09-01T06:06:26.524448",
"referral_code": "52d6ce"
}
Usage example with Python-json-logger:
import logging
from pythonjsonlogger import jsonlogger
logger = logging.getLogger()
logHandler = logging.FileHandler(filename='/var/log/my-log.json')
formatter = jsonlogger.JsonFormatter()
logHandler.setFormatter(formatter)
logger.addHandler(logHandler)
logger.setLevel(logging.INFO)
logger.info('Sign up', extra={'referral_code': '52d6ce'})
The log file contains the following log record (inline):
{
"message": "Sign up",
"referral_code": "52d6ce"
}
For information on configuring the handler, see the Python-json-logger documentation.
Create a file conf.yaml
in the Agent’s conf.d/python.d/
directory with the following content:
init_config:
instances:
##Log section
logs:
- type: file
path: "<PATH_TO_PYTHON_LOG>.log"
service: "<YOUR_APPLICATION>"
source: python
sourcecategory: sourcecode
# For multiline logs, if they start by the date with the format yyyy-mm-dd uncomment the following processing rule
#log_processing_rules:
# - type: multi_line
# name: new_log_start_with_date
# pattern: \d{4}\-(0?[1-9]|1[012])\-(0?[1-9]|[12][0-9]|3[01])
Restart the Agent to apply the configuration changes.
Further Reading
Additional helpful documentation, links, and articles: