Cloud Custodian is a rules engine for managing public cloud accounts and resources. It allows users to define policies to enable a well managed cloud infrastructure, that's both secure and cost optimized. It consolidates many of the adhoc scripts organizations have into a lightweight and flexible tool, with unified metrics and reporting.
Custodian can be used to manage AWS, Azure, and GCP environments by ensuring real time compliance to security policies (like encryption and access requirements), tag policies, and cost management via garbage collection of unused resources and off-hours resource management.
Custodian policies are written in simple YAML configuration files that enable users to specify policies on a resource type (EC2, ASG, Redshift, CosmosDB, PubSub Topic) and are constructed from a vocabulary of filters and actions.
It integrates with the cloud native serverless capabilities of each provider to provide for real time enforcement of policies with builtin provisioning. Or it can be run as a simple cron job on a server to execute against large existing fleets.
"Engineering the Next Generation of Cloud Governance" by @drewfirment
- Comprehensive support for public cloud services and resources with a rich library of actions and filters to build policies with.
- Supports arbitrary filtering on resources with nested boolean conditions.
- Dry run any policy to see what it would do.
- Automatically provisions serverless functions and event sources ( AWS CloudWatchEvents, AWS Config Rules, Azure EventGrid, GCP AuditLog & Pub/Sub, etc)
- Cloud provider native metrics outputs on resources that matched a policy
- Structured outputs into cloud native object storage of which resources matched a policy.
- Intelligent cache usage to minimize api calls.
- Supports multi-account/subscription/project usage.
- Battle-tested - in production on some very large cloud environments.
$ python3 -m venv custodian
$ source custodian/bin/activate
(custodian) $ pip install c7nThe first step to using Cloud Custodian is writing a YAML file containing the policies that you want to run. Each policy specifies the resource type that the policy will run on, a set of filters which control resources will be affected by this policy, actions which the policy with take on the matched resources, and a mode which controls which how the policy will execute.
The best getting started guides are the cloud provider specific tutorials.
As a quick walk through, below are some sample policies for AWS resources.
- will enforce that no S3 buckets have cross-account access enabled.
- will terminate any newly launched EC2 instance that do not have an encrypted EBS volume.
- will tag any EC2 instance that does not have the follow tags "Environment", "AppId", and either "OwnerContact" or "DeptID" to be stopped in four days.
policies:
- name: s3-cross-account
description: |
Checks S3 for buckets with cross-account access and
removes the cross-account access.
resource: aws.s3
region: us-east-1
filters:
- type: cross-account
actions:
- type: remove-statements
statement_ids: matched
- name: ec2-require-non-public-and-encrypted-volumes
resource: aws.ec2
description: |
Provision a lambda and cloud watch event target
that looks at all new instances and terminates those with
unencrypted volumes.
mode:
type: cloudtrail
role: CloudCustodian-QuickStart
events:
- RunInstances
filters:
- type: ebs
key: Encrypted
value: false
actions:
- terminate
- name: tag-compliance
resource: aws.ec2
description: |
Schedule a resource that does not meet tag compliance policies to be stopped in four days. Note a separate policy using the`marked-for-op` filter is required to actually stop the instances after four days.
filters:
- State.Name: running
- "tag:Environment": absent
- "tag:AppId": absent
- or:
- "tag:OwnerContact": absent
- "tag:DeptID": absent
actions:
- type: mark-for-op
op: stop
days: 4You can validate, test, and run Cloud Custodian with the example policy with these commands:
# Validate the configuration (note this happens by default on run)
$ custodian validate policy.yml
# Dryrun on the policies (no actions executed) to see what resources
# match each policy.
$ custodian run --dryrun -s out policy.yml
# Run the policy
$ custodian run -s out policy.ymlYou can run Cloud Custodian via Docker as well:
# Download the image
$ docker pull cloudcustodian/c7n
$ mkdir output
# Run the policy
#
# This will run the policy using only the environment variables for authentication
$ docker run -it \
-v $(pwd)/output:/home/custodian/output \
-v $(pwd)/policy.yml:/home/custodian/policy.yml \
--env-file <(env | grep "^AWS\|^AZURE\|^GOOGLE") \
cloudcustodian/c7n run -v -s /home/custodian/output /home/custodian/policy.yml
# Run the policy (using AWS's generated credentials from STS)
#
# NOTE: We mount the ``.aws/credentials`` and ``.aws/config`` directories to
# the docker container to support authentication to AWS using the same credentials
# credentials that are available to the local user if authenticating with STS.
# This exposes your container to additional credentials than may be necessary,
# i.e. additional credentials may be available inside of the container than is
# minimally necessary.
$ docker run -it \
-v $(pwd)/output:/home/custodian/output \
-v $(pwd)/policy.yml:/home/custodian/policy.yml \
-v $(cd ~ && pwd)/.aws/credentials/home/custodian/:.aws/credentials \
-v $(cd ~ && pwd)/.aws/config:/home/custodian/.aws/config \
--env-file <(env | grep "^AWS") \
cloudcustodian/c7n run -v -s /home/custodian/output /home/custodian/policy.ymlCustodian supports other useful subcommands and options, including outputs to S3, CloudWatch metrics, STS role assumption. Policies go together like Lego bricks with actions and filters.
Consult the documentation for additional information, or reach out on gitter.
For specific instructions for AWS, Azure, and GCP, visit the relevant getting started page.
The Custodian project also develops and maintains a suite of additional tools here https://github.com/cloud-custodian/cloud-custodian/tree/master/tools:
-
Org: Multi-account policy execution.
-
PolicyStream: Git history as stream of logical policy changes.
-
Salactus: Scale out s3 scanning.
-
Mailer: A reference implementation of sending messages to users to notify them.
-
Trail Creator Retroactive tagging of resources creators from CloudTrail
-
TrailDB: Cloudtrail indexing and time series generation for dashboarding.
-
LogExporter: Cloud watch log exporting to s3
-
Index: Indexing of custodian metrics and outputs for dashboarding
-
Sentry: Cloudwatch Log parsing for python tracebacks to integrate with https://sentry.io/welcome/
See https://cloudcustodian.io/docs/contribute.html
This project adheres to the Open Code of Conduct. By participating, you are expected to honor this code.