Skip to content

Latest commit

 

History

History
215 lines (173 loc) · 7.02 KB

README.md

File metadata and controls

215 lines (173 loc) · 7.02 KB
Raw

Akashi / 証

Akashi is a Go tool that can be used to parse terraform plan outputs and validate the changes.

Still a WIP project. Supports terraform v0.12 currently.

  • Created and destroyed resources
  • JSON Output
  • stdout output
  • Map / Array rules
  • Updated resources
  • Index matching
  • Module matching
  • Multiple rule matching
  • Other validations(regex, int in range, etc)
  • Combining multiple rulesets
  • Customizable output

Why

By parsing the terraform plan result, you can validate changes shown in the plan output before actually applying. Also, it can be used to track unexpected changes during routine operations that may have occured from misconfiguration or infrastructure drift.

For example, say you had a google_container_node_pool with autoscaling configured, and due to some incident you had to modify the autoscaling configuration outside of terraform. Some time later, you update the version of your google_container_node_pool as a routine operation. Due to the declarative nature of terraform, the previous configuration of autoscaling would be applied and overwrite the modified configuration set during the incident. This might be unintentional and would not be caught from parsing the terraform configuration, but can be caught by validating the terraform plan output.

Installation

go get -u github.com/drlau/akashi

Usage

Usage:
  akashi <compare | diff> <path to ruleset> [flags]

By default, akashi will read a terraform plan output from stdin, so you should pipe the result of terraform plan:

terraform plan | akashi <compare | diff> <path to ruleset>

If you produced a plan output with terraform plan -out=<file>, akashi can parse it by specifying --json. Note that this file is encoded and needs to be decoded by terraform before akashi can parse it properly:

terraform state show -json <file> | akashi <compare | diff> <path to ruleset> --json

If the terraform plan output or the decoded json is in a file, you can read directly from the file by specifying the path with -f.

Ruleset schema

NOTE: Ruleset schema is in the early stages and is subject to change in later versions.

Rulesets are written in YAML and have the following schema:

# Rules to apply to created resources.
createdResources:
  # Set to true if you want all created resources to match a rule.
  # Default is false.
  strict: true

  # Set to true if you want to validate the ruleset specifies names for
  # created resources.
  # Default is false.
  requireName: true

  # Default compare options to apply to all resources.
  # If a resource specifies the same option, the resource's value will be used.
  default:
    # Set to true if you want every enforcedValue to be present in a resource's plan.
    # If set to true, if an enforcedValue is missing, it will count as a failure.
    # Default is false.
    enforceAll: true

    # Set to true if you want to ignore extra arguments in a resource's plan
    # that don't match any enforcedValue or ignoredArg.
    # If set to false, any extra arguments will count as a failure.
    # Default is false.
    ignoreExtraArgs: true

    # Set to true if you want to ignore computed arguments in a resource's plan.
    # Default is false.
    ignoreComputed: true

    # Set to true if you want every enforcedValue and ignoredArg to be present in a resource's plan.
    # If set to true, if any enforcedValue or ignoredArg is missing, it will count as a failure.
    # Default is false.
    requireAll: true

    # Set to true if you want to automatically fail before comparison
    # if a matching resource is found.
    # Default is false.
    autoFail: true

  # List of rules.
  resources:
    -
      # Resource name to match on.
      # At least one of "name" or "type" must be set.
      # Default is empty.
      name: resource-name

      # Resource type to match on.
      # At least one of "name" or "type" must be set.
      # Default is empty.
      type: resource-type

      # The same compare options from "default" can be specified per resource.
      # The resource level option will take priority over the option specified in "default"
      # If omitted, the option specified in "default" is used.
      # If not specified in "default", the value is false.
      enforceAll: true
      ignoreExtraArgs: true
      ignoreComputed: true
      requireAll: true
      autoFail: true

      # List of arguments to ignore.
      # Default is empty.
      ignored:
        - ignored-arg-1
        - ignored-arg-2

      # List of arguments to enforce.
      # Default is empty.
      enforced:
        stringEnforced:
          value: string
        intEnforced:
          value: 1
        boolEnforced:
          value: true
        mapEnforced:
          value:
            mapKey: mapValue
        arrayEnforced:
          value:
          - array1
          - array2
        stringMatchAny:
          matchAny:
          - validValue1
          - validValue2

# Rules to apply to destroyed resources.
# Has the exact same schema as createdResources.
destroyedResources:

# Rules to apply to updated resources.
updatedResources:
  # Set to true if you want all updated resources to match a rule.
  # Default is false.
  strict: true

  # Set to true if you want to validate the ruleset specifies names for
  # updated resources.
  # Default is false.
  requireName: true

  # Default compare options to apply to all resources.
  # If a resource specifies the same option, the resource's value will be used.
  # All options for created and destroyed resources work here, but also has a few additional options that can be enabled
  default:
    # Set to true if you want to ignore all unchanged attributes
    # Default is false.
    ignoreNoOp: true

  # List of rules.
  resources:
  -
    # Similar to created and destroyed resources, matching can be done on a name and/or type of resource
    name: resource-name
    type: resource-type

    # Every compare option can also be specified at resource level, overriding the top level default
    ignoreNoOp: true

    # Rules to enforce on the attributes before the planned changes
    # Consists of ignored and enforced, with the same behaviour as created and destroyed resources
    before:
      # List of arguments to ignore.
      # Default is empty.
      ignored:
        - ignored-arg-1
        - ignored-arg-2

      # List of arguments to enforce.
      # Default is empty.
      enforced:
        stringEnforced:
          value: string

    # Rules to enforce on the attributes after the planned changes
    # Same schema as before.
    after:

Example

Say you provision google_compute_instance and you want to validate that all new instances are created in zone us-central1-a, and you don't care about any other argument. To validate that, you would create the following ruleset:

createdResources:
  default:
    enforceAll: true
    ignoreExtraArgs: true
  resources:
    - type: google_compute_instance
      enforced:
        zone:
          value: us-central1-a