Skip to content

AWS CloudWatch client library to send metrics conveniently and efficiently

License

Notifications You must be signed in to change notification settings

peterdemin/awsme

Repository files navigation

Amazon Web Services Cloud Watch Metrics Library

Updates

Configurable client library, that supports asynchronous and buffered sending of AWS cloud watch metrics.

Installation

By default awsme is installed without boto3 dependency to make it deploy faster in AWS Lambda environment (See #3 for details).

If you run awsme outside of Lambda, use following command to activate boto3 dependency:

pip install awsme[boto3]

Usage

from awsme import create_cloud_watch
cloud_watch = create_cloud_watch(
    namespace='Application',
    dimensions={'version': '1.0.0'},
)
cloud_watch.log('metric', dimensions={'key': 'dim'}, value=123)

Create Options

create_cloud_watch accepts following arguments:

  • asynchronous (optional bool): if True (default), send metrics from a separate thread.
  • buffered (optional bool): if True (default), metrics will be accumulated in a buffer and sent in batches.
  • dummy (optional bool): if True, ignore two previous options and create dummy recorder. False by default.
  • dimensions: (optional Dict[str, str]): dictionary of default dimensions, that will be attached to all metrics.
  • All other kwargs will be bypassed to boto3.client('cloudwatch', **kwargs)

Log Options

CloudWatch, returned by create_cloud_watch has two public methods. The primary method is log. It's arguments:

  • name (required str): name of the metric.
  • dimensions (optional Dict[str, str]): additional dimensions, that will be added to default dimension from factory.
  • value (optional float): metric's value, 1 by default.
  • unit (optional str): metric unit, e.g. Count, Seconds, Bytes, see AWS docs for a complete list of valid values.
  • storage_resolution (optional int): metric storage resolution in seconds, 60 by default.

Flushing

If create_cloud_watch was called with buffered=True (default) then you may want to forcefully flush the internal metrics buffer. In a standard application this will likely not be necessary, as the buffer will auto-flush atexit. However, if your application is running as an AWS Lambda function, the execution of the Lambda function will be "frozen" when the function completes, preventing exiting and thus flushing. In this use case, you will need to forcibly flush the buffer by calling flush. It's argument:

  • complete (optional bool): if True (default), perform a complete flush.

AWS configuration

Awsme uses boto3 library, that takes configuration from environment variables and configuration files. To check, that you have everything configured properly to send metrics, use included command-line tool:

$ awsme-test
...
Successfully sent metric "awsme-test" to "Test Namespace"

Credits

This package was created with Cookiecutter and the elgertam/cookiecutter-pipenv project template, based on audreyr/cookiecutter-pypackage.

About

AWS CloudWatch client library to send metrics conveniently and efficiently

Resources

License

Stars

Watchers

Forks

Packages

No packages published