Boto3 Extensions

Overview

This module adds more resource files to the Boto3 library and includes some functionality enhancements.

Quick Start

First, install the library:

$ pip install boto3_extensions

Follow the Boto3 docs on setting up your region and credentials (https://github.com/boto/boto3).

Then, from a Python interpreter:

import boto3
import boto3_extensions

r = boto3.resource('cloudtrail', region_name='us-east-1')
for trail in r.trails.all():
    print(trail.trail_arn)

Resource Files

The following resource files are added to the Boto3 library.

• acm
• autoscaling
• cloudfront
• cloudtrail
• cur
• datapipeline
• directconnect
• elasticache
• elb
• elbv2
• emr
• glue
• health
• kinesis
• lambda
• rds
• redshift
• route53
• support

The following resource files are updated in the Boto3 library.

• dynamodb
• ec2
• iam
• s3

RefreshableAssumeRoleProvider

If your code needs to AssumeRole into another role before performing actions against the AWS API (be it in the same or another AWS account), you run the risk that the credentials you are using expire during their use. You can either add code to your application to constantly check the credential expiry time or using this extension offload the credential refresh to boto3 itself. By using the ConnectionManager in boto3_extensions not only will it automattically assumeRole when the credentials get below 15 mins left, but it will also cache the credentials. This means that if your application is calling boto3 to get credentials for another role more than once the ConnectionManager will cache the first call and then hand out the same session for the subsequent calls.

role_arn = 'arn:aws:iam::1234567890:role/test-role'
role_session_name = 'test'
connections = boto3_extensions.ConnectionManager(region_name='us-east-1')
session = connections.get_session(role_arn=role_arn, role_session_name=role_session_name)

r = session.resource('cloudtrail', region_name='us-east-1')
for trail in r.trails.all():
    print(trail.trail_arn)

ARN Patch

It would be nice to have a consistent way to get the ARN of resources. The ARN patch feature of boto3_extensions allows you to get the arn from resources via an arn attribute.

import boto3
import boto3_extensions
from imp import reload
boto3_extensions.arn_patch_boto3()
reload(boto3)

r = boto3.resource('rds', region_name='us-east-1')
for db in r.db_instances.all():
    print(db.arn)

Session Caching

There is basic Session caching builtin that simply stores the Session for a each assumed Role into a dict, but if you require another caching mechanism you can monkey patch the boto3_extensions.Cache class, as long as you supply the Cache.get() and Cache.set() methods.

import boto3
import boto3_extensions
from boto3_extensions import ConnectionManager

class MyCache:
    def __init__(self):
        self._cache = {}

    def get(self, role_arn, role_session_name):
        print("inside MyCache.get()")
        return self._cache.get((role_arn, role_session_name), None)

    def set(self, role_arn, role_session_name, session):
        print("inside MyCache.set()")
        self._cache[(role_arn, role_session_name)] = session

boto3_extensions.Cache = MyCache
connections = ConnectionManager()

connections.get_session(role_arn="arn:aws:iam::012345678912:role/test_role", role_session_name="testing")
inside MyCache.get()
inside MyCache.set()

connections.get_session(role_arn="arn:aws:iam::012345678912:role/test_role", role_session_name="testing")
inside MyCache.get()

In the above we monkey patch the Cache class and call get_session() twice. The output shows that the first time we have a cache get which results in a cache miss and thena cache set after the credential is retrieved from STS. We then run get_session() a second time and as the Session is now cached we only see a cache get resulting in a cache hit.

Getting Help

Please raise issue ticket inside our Bitbucket repo: https://bitbucket.org/atlassian/boto3_extensions/issues

License

Copyright (c) 2023 Atlassian US., Inc. Apache 2.0 licensed, see LICENSE file.

Developing

To develop this library, you will need to install the following:

• Python 3.7
• Poetry (https://poetry.eustace.io/docs/)

# Install development environment and setup pre-commit hooks
$ make setup