s3keyring 0.2.1

S3 bucket

The S3 keyring backend requires you to have read/write access to a S3 bucket. If you want to use bucket mysecretbucket to store your keyring, you will need to attach the following IAM policy to all the IAM user accounts or roles that will have read and write access to the keyring:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "s3:ListBucket"
            ],
            "Resource": "arn:aws:s3:::mysecretbucket",
            "Condition": {}
        },
        {
            "Effect": "Allow",
            "Action": [
                "s3:DeleteObject",
                "s3:GetObject",
                "s3:PutObject"
            ],
            "Resource": "arn:aws:s3:::mysecretbucket/*"
        }
    ]
}

You can easily create a policy that grants read-only access to the keyring by removing the s3:PutObject and s3:DeleteObject actions from the policy above.

Encryption key

You need to create a KMS encryption key. Write down the ID of the KMS key that you create. You will need to communicate this KMS Key ID to the users of the keyring.

IMPORTANT: You will need to grant read access to the KMS key to every IAM user or role that needs to access the keyring.

One-time configuration

If you haven’t done so already, you will need to configure your local installation of the AWS CLI by running:

aws configure

Then configure the S3 Keyring:

s3keyring configure

Your keyring administrator will provide you with the KMS Key ID, Bucket and Namespace configuration options. Option AWS profile allows you to specify the local AWS CLI profile you want to use to sign all requests sent to AWS when accessing the keyring. Most users will want to use the default profile.

IMPORTANT: when deploying the s3keyring in EC2 instances that are granted access to the keyring by means of an IAM role you should not specify a custom AWS profile when configuring s3keyring in the instances.

You can configure the s3keyring module without user input by setting the following environment variables: KEYRING_BUCKET, KEYRING_NAMESPACE, KEYRING_KMS_KEY_ID, KEYRING_AWS_PROFILE. If these environment variables are properly set then you can configure the s3keyring module with:

s3keyring configure --no-ask

Configuration profiles

You can use s3keyring to store (read) secrets in (from) more than one backend S3 keyring. A typical use case is creating different keyrings for different user groups that have different levels of trust. For instance your keyring administrator may have setup a S3 keyring that only IAM users with admin privileges can access. Using the bucket, KMS Key ID and namespace provided by your keyring admin you can configure a separate s3keyring profile to access that admins-only keyring:

s3keyring --profile administrators configure

Your keyring admin may have also setup a separate S3 keyring to store secrets that need to be accessed by EC2 instances that act as website workers in a project you are working on. To access that keyring you would configure a second s3keyring profile:

s3keyring --profile website-workers configure

Then, to store and retrieve secrets in the administrators keyring:

s3keyring --profile administrators set SERVICE ACCOUNT PASSWORD
s3keyring --profile administrators get SERVICE ACCOUNT

And you could do the same for the website-workers keyring using option --profile website-workers.

Configuration file

By default s3keyring reads configuration options from ~/.s3keyring.ini. You can also store the configuration in a .s3keyring.ini file stored in your current working directory by using:

s3keyring configure --local

s3keyring will always read the configuration first from a .s3keyring.ini file under your current work directory. If it is not found then it will read it from ~/.s3keyring.ini.

Recommended workflow

This is how I use s3keyring in my Python projects.

Let’s assume that my project root directory looks something like this:

setup.py
my_module/
          __init__.py

In my project root directory I run:

s3keyring configure --local

I keep the generated .s3keyring.ini file as part of my project source code (i.e. under version control). Then in my project code I use the keyring like this:

from s3keyring.s3 import S3Keyring

keyring = S3Keyring(config_file="/path/to/s3keyring.ini")
keyring.set_password('service', 'username', '123456')
assert keyring.get_password('service', 'username') == '123456'