logs-ingestion 0.2.1

Package for logging to Azure Logs Ingestion API using decorators and regular logging statements.

Introduction

This logs_ingestion package provides several methods for logging data into Azure Monitor using the Logs Ingestion API.

The following attributes are part of the logging with the coding attribute names between brackets:

• TimeGenerated (time_generated): the datetime the logging entry was created (required)
• Message (message): the message of the log entry (optional)
• Duration (duration): the duration of the function in case the decorator is used (optional)
• Status (status): the status of the system (optional)
• RunId (run_id): the run ID of the flow that is being processed (required)
• Tag (tag): the tag for grouping log entries (required)

The attribute TimeGenerated is automatically set for you. The RunId and Tag are part of the configuration of the logger you need to create and only need to be set once per logger.

Usage

You'll need a logger to perform the actual logging:

logger: Logger = get_logger(__name__, run_id="42", tag="logger1")

With the logger instantiation you'll also set the RunId and Tag to be used in all logging entries as generated through this logger.

The first method for logging information is by using a decorator in your Python code:

@time_and_log(logger=logger, message="bla", status="timed")
def my_function():
    pass

Whenever the my_function() is called a log entry is created with when the function end that automatically includes the duration of the function call. This is a convenient way for monitoring the performance of functions and the possible drift in processing times. With the message and status arguments you can add additional details to the message logged.

The second method is by calling the usual logging lines, for example:

from logs_ingestion.logs_record import LogsRecord
logger.warning(message='testing azure logging', record=LogsRecord(
               status="OK",
               duration=1.23))

The arguments are:

• message, speaks for itself
• record, the record(s) to be logged

The record argument must be either a LogsRecord or a list of LogsRecords. By using a list, you can simply log a whole batch of log records in one command. The rund_id and tag from the logger are pushed down to the individual log messages.