Skip to main content

Useful AWS CLI Extras

Project description


This is a general all-purpose tool for managing things in AWS that Terraform is not responsible for -- you can think of it as an extension to the aws CLI.

At the moment it only does three things; blue/green deploys for plugging into Gitlab, AMI cleanups, and RDS copies to other accounts.


pip3 install akinaka

Requirements and Presumptions

Format of ASG names: "whatever-you-like*-blue/green*" — the part in bold is necessary, i.e. you must have two ASGs, one ending with "-blue" and one ending with "-green".

The following permissions are necessary for the IAM role / user that will be running Akinaka:


NOTE: Going forward, IAM policies will be listed separately for their respective subcommands (as is already the case for Transfer). For now however, the following single catch-all policy can be used, attach it to the IAM profile that Akinaka will be assuming:

    "Version": "2012-10-17",
    "Statement": [
            "Sid": "2018121701",
            "Effect": "Allow",
            "Action": [
            "Resource": "*"
            "Sid": "2018121702",
            "Effect": "Allow",
            "Action": [
            "Resource": [

A Note on Role Assumption

Akinaka uses IAM roles to gain access into multiple accounts. Most commands require you to specify a list of roles you wish to perform a task for, and that role must have the sts:AssumeRole permission. This is not only good security, it's helpful for ensuring you're doing things to the accounts you think you're doing things for ;)


Done with the update parent command, and then the asg and targetgroup subcommands (update targetgroup is only needed for blue/green deploys).


# For standalone ASGs (not blue/green)
akinaka update \
  --region eu-west-1 \
  --role-arn arn:aws:iam::123456789100:role/management_assumable \
asg \
  --asg workers \
  --ami ami-000000

# For blue/green ASGs
akinaka update \
  --region eu-west-1 \
  --role-arn arn:aws:iam::123456789100:role/management_assumable \
asg \
  --lb lb-asg-ext \
  --ami ami-000000

# For blue/green ASGs with multiple Target Groups behind the same ALB
akinaka update \
  --region eu-west-1 \
  --role-arn arn:aws:iam::123456789100:role/management_assumable \
asg \
  --target-group application-1a \
  --ami ami-000000

For blue/green deploys, the next step is to check the health of your new ASG. For the purposes of Gitlab CI/CD pipelines, this will be printed out as the only output, so that it can be used in the next job.

Once the new ASG is confirmed to be working as expected:

akinaka update --region eu-west-1 --role-arn arn:aws:iam::123456789100:role/management_assumable asg --new blue

The value of --role-arn is used to assume a role in the target account with enough permissions to perform the actions of modifying ASGs and Target Groups. As such, akinaka is able to do cross-account deploys. It will deliberately error if you do not supply an IAM Role ARN, in order to ensure you are deploying to the account you think you are.


Currently AMI, EBS, and RDS snapshot cleanups are supported.

Common option:

--role-arns is a space separated list of IAM ARNs that can be assumed by the token you are using to run this command. The AMIs for the running instances found in these accounts will not be deleted. Not to be confused with --role-arn, accepted for the update parent command, for deploys.


Cleans up AMIs and their snapshots based on a specified retention period, and deduced AMI usage (will not delete AMIs that are currently in use). You can optionally specify an AMI name pattern, and it will keep the latest version of all the AMIs it finds for it.


akinaka cleanup \
    --region eu-west-1 \
    --role-arns "arn:aws:iam::198765432100:role/management_assumable arn:aws:iam::123456789100:role/management_assumable" \
    ami \
        --exceptional-amis cib-base-image-*
        --retention 7

The above will delete all AMIs and their snapshots, except for those which:

  1. Are younger than 7 days AND
  2. Are not in use by AWS accounts "123456789100" or "198765432100" AND
  3. WHERE the AMI name matches the pattern "cib-base-image-*", there is more than one match AND it is the oldest one

--exceptional-amis is a space seperated list of exact names or patterns for which to keep the latest version of an AMI for. For example, the pattern "cib-base-image-*" will match with normal globbing, and if there is more than one match, only the latest one will not be deleted (else there is no effect).

--retention is the retention period you want to exclude from deletion. For example; --retention 7 will keep all AMIs found within 7 days, if they are not in the --exceptional-amis list.

EBS Volumes

Delete all EBS volumes that are not attached to an instance (stopped or not):

akinaka cleanup \
    --region eu-west-1 \
    --role-arns "arn:aws:iam::198765432100:role/management_assumable arn:aws:iam::123456789100:role/management_assumable" \

RDS Snapshots

This will delete all snapshots tagged "akinaka-made":

akinaka cleanup \
    --not-dry-run \
    --region eu-west-1 \
    --role-arns "arn:aws:iam::876521782800:role/OlinDataAssumedAdministrator" \
    rds \
        --tags "akinaka-made"


Perform often necessary but complex tasks with RDS.


Copy encrypted RDS instances between accounts:

akinaka copy --region eu-west-1 \
    rds \
        --source-role-arn arn:aws:iam::198765432100:role/management_assumable \
        --target-role-arn arn:aws:iam::123456789100:role/management_assumable \
        --snapshot-style running_instance \
        --source-instance-name DB_FROM_ACCOUNT_198765432100 \
        --target-instance-name DB_FROM_ACCOUNT_123456789100 \
        --target-security-group SECURITY_GROUP_OF_TARGET_RDS \
        --target-db-subnet SUBNET_OF_TARGET_RDS \

--region is optional because it will default to the environment variable AWS_DEFAULT_REGION.

Disaster Recovery

Akinaka has limited functionality for backing up and restoring data for use in disaster recovery.


Transfer data from S3, RDS, and RDS Aurora into a backup account:

akinaka dr \
  --region eu-west-1 \
  --source-role-arn arn:aws:iam::[LIVE_ACCOUNT_ID]:role/[ROLE_NAME] \
  --destination-role-arn arn:aws:iam::[BACKUP_ACCOUNT_ID]:role/[ROLE_NAME] \
  transfer \
    --service s3

Omitting --service will include all supported services.

You can optionally specify the name of the instance to transfer with --names in a comma separated list, e.g. --names 'database-1, database-2. This can be for either RDS instances, or S3 buckets, but not both at the same time. Future versions may remove --service and replace it with a subcommand instead, i.e. akinaka dr transfer rds, so that those service can have --names to themselves.

A further limitation is that only a single region can be handled at a time for S3 buckets. If you wish to backup all S3 buckets in an account, and they are in different regions, you will have to specify them per run, using the appropriate region each time. Future versions will work the bucket regions out automatically, and remove this limitation.

Akinaka must be run from either an account or instance profile which can use sts:assume to assume both the source-role-arn and destination-role-arn. This is true even if you are running on the account that destination-role-arn is on. You will therefore need this policy attached to the user/role that's doing the assuming:

    "Version": "2012-10-17",
    "Statement": [
            "Sid": "akinakaassume",
            "Effect": "Allow",
            "Action": "sts:AssumeRole",
            "Resource": [

Note: A period of 4 hours (469822 seconds) is hardcoded into the sts:assume call made in the RDS snapshot class, since snapshot creation can take a very long time. This must therefore be the minimum value for the role's max-session-duration.

The following policy is needed for usage of this subcommand, attach it to the role you'll be assuming:

    "Version": "2012-10-17",
    "Statement": [
            "Sid": "KMSEncrypt",
            "Effect": "Allow",
            "Action": [
            "Resource": [
            "Sid": "KMSCreate",
            "Effect": "Allow",
            "Action": [
            "Resource": "*"

The following further policies need to be attached to the assume roles to backup each service:

RDS / RDS Aurora

    "Version": "2012-10-17",
    "Statement": [
            "Sid": "RDSBackup",
            "Effect": "Allow",
            "Action": [
            "Resource": "*"


    "Version": "2012-10-17",
    "Statement": [
            "Sid": "S3RW",
            "Effect": "Allow",
            "Action": [
            "Resource": "*"


Limited functionality for interactive with EKS and ECR. At the moment it's just getting a docker login via an assumed role to another assumed role:

akinaka container --region eu-west-1 --role-arn arn:aws:iam::0123456789:role/registry-rw get-ecr-login --registry 0123456789

The above will assume the role arn:aws:iam::0123456789:role/registry-rw in the account with the registry, and spit out a docker login line for you to use — exactly like aws ecr get-login, but working for assumed roles.


Get a view of your daily AWS estimated bill for the x number of days. Defaults to today's estimated bill.

akinaka reporting --region us-east-1 \
  --role-arn arn:aws:iam::1234567890:role/billing_assumerole \
  bill-estimates --from-days-ago 1

Example output:

Today's estimated bill
| Date       | Total     |
| 2019-03-14 | USD 13.93 |

You can specify any integer value to the --days-ago flag. It's optional. Default value set for today (current day).

You can specify any region to the --region flag.


Modules can be added easily by simply dropping them in and adding an entry into akinaka to include them, and some click code in their __init__ (or elsewhere that's loaded, but this is the cleanest way).

For example, given a module called akinaka_moo, and a single command and file called moo, add these two lines in the appropriate places of akinaka:

from akinaka_update.commands import moo as moo_commands

and the following in the module's
@click.option("--make-awesome", help="The way in which to make moo awesome")
def moo(make_awesome):
    import .moo

Adding commands that need subcommands isn't too different, but you might want to take a look at the already present examples of update and cleanup.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

akinaka-1.0.1.tar.gz (54.3 kB view hashes)

Uploaded source

Built Distribution

akinaka-1.0.1-py3-none-any.whl (63.3 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page