Python function runtime logging

Cloud Functions automatically captures a Python application's standard output streams and sends them to the centralized logging system available in Yandex Cloud. In addition to the application run history, the system logs request execution events.

Additional messages are logged using standard language constructs:

1. print: Outputs a message to the standard output stream, stdout.
2. logging: Outputs a message in the set format to the selected output stream.

System messages

When processing each function call, the system also logs START, END, and REPORT system messages:

START RequestID: <request_ID> Version: <function_version_ID>
END RequestID: <request_ID>
REPORT RequestID: <request_ID>
    Duration: 236.606 ms
    Billed Duration: 300 ms
    Memory Size: 128 MB
    Max Memory Used: 22 MB
    Queuing Duration: 0.027 ms
    Function Init Duration: 225.298 ms

All lines contain the ID of the request (RequestID), which is generated automatically when the function is invoked.

The REPORT line provides a report on the function run. It contains additional information about the resources consumed:

• Duration: Time spent invoking the function. It includes the Queuing Duration and Function Init Duration parameters.
• Billed Duration: Time billed based on the pricing policy.
• Memory Size: Amount of memory specified at version creation, MB.
• Max Memory Used: Memory used when the request starts running.
• Queuing Duration: Time spent by the request in the internal queue. If this time increases, it may indicate a lack of function instances. The maximum number of instances is determined by the quotas.
• Function Init Duration: Time spent initializing the runtime and loading the function code.

Custom messages

Before running a Cloud Functions function, configures the following for the root logger:

• A handler for logging data to the stdout output stream.
• A formatter that adds to the message its creation timestamp, the request ID, and the logging level.

By default, the root logger level is Warning, all logs with a lower level are ignored. You can change the logging level using the setLevel method for:

• The entire app.

  logging.getLogger().setLevel(logging.DEBUG)
  

• Any logger except the root one.

  logging.getLogger('requests.packages.urllib3').setLevel(logging.DEBUG)
  logging.getLogger('myapp').setLevel(logging.DEBUG)
  

You can change the log format as follows:

root_handler = logging.getLogger().handlers[0]
root_handler.setFormatter(logging.Formatter(
	'[%(levelname)s]\t%(name)s\t%(request_id)s\t%(message)s\n'
))

The example shows the logger name output instead of the message creation timestamp.

Learn more about how to configure logging in the Python documentation.