app-logger 0.1.2

A drop in logger configurable through environment variables

app-logger

This is a drop in logger that uses python-json-logger to log json to stdout and stderr appropriately. DEBUG and INFO go to stdout, WARNING, ERROR, CRITICAL go to stderr.

Configuration

All configuration is through environment variables, all have defaults so none are required.

LOGGER_NAME

What should the logger be called? Default will use the root logger. Using the default or root will log not only log statements from app_logger but also any statements from libraries you're using. Format: Just a regular string Default: root Example: mysweetapp

LOG_LEVELS

Used to set log levels. Can be used to set any log level you wish so you can selectively control logging. May be upper or lower case. Format: A comma separated string of key=value pairs Default: root=INFO Example: root=DEBUG,asyncio=INFO

LOG_FORMAT

Used to control the log format. Format: A log format string of log record attributes Default: `%(levelname)%(name)%(asctime)%(module)%(funcName)%(lineno)%(message)``

Usage

The logger initializes itself and makes itself available as a variable called app_logger. It's a regular python logger and can be used as such. Example app_logger.info("I always hated python logging but now it's easy and just works") Resulting log {"levelname": "INFO", "name": "root", "asctime": "2020-04-11 11:24:17,299", "module": "main", "funcName": "main", "lineno": 14, "message": "I always hated python logging but now it's easy and just works"}

Adding context

You can use the extra= feature of python-json-logger to add context to your messages. This makes it really easy to have logging that's easy to parse, search, learn, and alert on if you're using log aggregation. Example Given you have something that's a dictionary, you can include it in log statements without verbose string formatting. app_logger.error('Error handling message', extra=message) Resulting log {"levelname": "ERROR", "name": "root", "asctime": "2020-04-11 11:27:36,584", "module": "main", "funcName": "main", "lineno": 16, "message": "Error handling message", "guid": "7cc81eba-3bbc-4555-8fab-2a7556072f5d", "subject": "test"} You can see how easy it would be to find logs like this in kibana, and it's already in json format so it's easily indexable/searchable, including the context given via extra=