Chef InSpec DSL

Chef InSpec is a run-time framework and rule language used to specify compliance, security, and policy requirements. It includes a collection of resources that help you write auditing controls quickly and easily. The syntax used by both open source and |chef compliance| auditing is the same. The open source |Chef InSpec resource| framework is compatible with |chef compliance|.

The Chef InSpec DSL is a Ruby DSL for writing audit controls, which includes audit resources that you can invoke.

The following sections describe the syntax and show some simple examples of using the Chef InSpec resources.

Syntax

The following resource tests |ssh| server configuration. For example, a simple control may described as:

describe sshd_config do
  its('Port') { should cmp 22 }
end

In various use cases like implementing IT compliance across different departments, it becomes handy to extend the control with metadata. Each control may define an additional impact, title or desc. An example looks like:

control 'sshd-8' do
  impact 0.6
  title 'Server: Configure the service port'
  desc 'Always specify which port the SSH server should listen.'
  desc 'rationale', 'This ensures that there are no unexpected settings' # Requires Chef InSpec >=2.3.4
  tag 'ssh','sshd','openssh-server'
  tag cce: 'CCE-27072-8'
  ref 'NSA-RH6-STIG - Section 3.5.2.1', url: 'https://www.nsa.gov/ia/_files/os/redhat/rhel5-guide-i731.pdf'

  describe sshd_config do
    its('Port') { should cmp 22 }
  end
end

where

  • 'sshd-8' is the name of the control
  • impact, title, and desc define metadata that fully describes the importance of the control, its purpose, with a succinct and complete description
  • desc when given only one argument it sets the default description. As of Chef InSpec 2.3.4, when given 2 arguments (see: 'rationale') it will use the first argument as a header when rendering in Automate
  • impact is a string, or numeric that measures the importance of the compliance results. Valid strings for impact are none, low, medium, high, and critical. The values are based off CVSS 3.0. A numeric value must be between 0.0 and 1.0. The value ranges are:
    • 0.0 to <0.01 these are controls with no impact, they only provide information
    • 0.01 to <0.4 these are controls with low impact
    • 0.4 to <0.7 these are controls with medium impact
    • 0.7 to <0.9 these are controls with high impact
    • 0.9 to 1.0 these are critical controls
  • tag is optional meta-information with with key or key-value pairs
  • ref is a reference to an external document
  • describe is a block that contains at least one test. A control block must contain at least one describe block, but may contain as many as required
  • sshd_config is a Chef InSpec resource. For the full list of Chef InSpec resources, see Chef InSpec resource documentation
  • its('Port') is the matcher; { should eq '22' } is the test. A describe block must contain at least one matcher, but may contain as many as required

Advanced concepts

Checking if at least one condition passes with describe.one

With Chef InSpec, you can check if at least one of a collection of checks is true. For example, if you configure a setting in two different locations, then you may want to test if either configuration A or configuration B is set. Do this task with the describe.one block. describe.one defines a set of describe blocks where only one block needs to pass.

describe.one do
  describe ConfigurationA do
    its('setting_1') { should eq true }
  end

  describe ConfigurationB do
    its('setting_2') { should eq true }
  end
end

describe.one usage notes

  • A describe.one block passes if one of its nested describe blocks has all assertions passing. A describe.one block needs an entire describe block to pass and not just a single assertion.
  • Chef InSpec will always evaluate all the tests contained within describe.one. It does not short-circuit upon evaluating a passing describe block.
  • Nesting a describe.one block inside another describe.one block is not supported.

Sensitive resources

In some scenarios, you may be writing checks that involve resources with sensitive content, such as a file resource. In case of failures, you may desire to suppress output. Do this task by adding the :sensitive flag to the resource definition:

describe file('/tmp/mysecretfile'), :sensitive do
  its('content') { should match /secret_info/ }
end

Examples

The following examples show simple compliance tests using a single control block.

Test System Event Log

The following test shows how to audit machines running Windows 2012 R2 that password complexity is enabled:

control 'windows-account-102' do
  impact 'critical'
  title 'Windows Password Complexity is Enabled'
  desc 'Password must meet complexity requirement'
  describe security_policy do
    its('PasswordComplexity') { should cmp 1 }
  end
end

Test if PostgreSQL passwords are empty

The following test shows how to audit machines running PostgreSQL to ensure that passwords are not empty.

control 'postgres-7' do
  impact 1.0
  title "Don't allow empty passwords"
  describe postgres_session('user', 'pass').query("SELECT * FROM pg_shadow WHERE passwd IS NULL;") do
    its('output') { should cmp '' }
  end
end

Test if MySQL passwords are in ENV

The following test shows how to audit machines running MySQL to ensure that passwords are not stored in ENV:

control 'mysql-3' do
  impact 1.0
  title 'Do not store your MySQL password in your ENV'
  desc '
    Storing credentials in your ENV may easily expose
    them to an attacker. Prevent this at all costs.
  '
  describe command('env') do
    its('stdout') { should_not match /^MYSQL_PWD=/ }
  end
end

Test if /etc/ssh is a Directory

The following test shows how to audit machines to ensure that /etc/ssh is a directory:

control 'basic-1' do
  impact 1.0
  title '/etc/ssh should be a directory'
  desc '
    In order for OpenSSH to function correctly, its
    configuration path must be a folder.
  '
  describe file('/etc/ssh') do
    it { should be_directory }
  end
end

Test if Apache running

The following test shows how to audit machines to ensure that Apache is enabled and running:

control 'apache-1' do
  impact 'medium'
  title 'Apache2 should be configured and running'
  describe service(apache.service) do
    it { should be_enabled }
    it { should be_running }
  end
end

Test if insecure packages are installed

The following test shows how to audit machines for insecure packages:

control 'cis-os-services-5.1.3' do
  impact 0.7
  title '5.1.3 Ensure rsh client is not installed'
  describe package('rsh') do
    it { should_not be_installed }
  end
  describe package('rsh-redone-client') do
    it { should_not be_installed }
  end
end

Test Windows Registry Keys

The following test shows how to audit machines to ensure Safe DLL Search Mode is enabled:

control 'windows-base-101' do
  impact 1.0
  title 'Safe DLL Search Mode is Enabled'
  desc '
    @link: https://msdn.microsoft.com/en-us/library/ms682586(v=vs.85).aspx
  '
  describe registry_key('HKLM\\System\\CurrentControlSet\\Control\\Session Manager') do
    it { should exist }
    it { should_not have_property_value('SafeDllSearchMode', :type_dword, '0') }
  end
end

Use only_if to exclude a specific control

This example shows how to allow skipping certain controls if conditions are not met by using only_if. In this example, the control will not be performed if the redis-cli command does not exist. A optional message can say why it was skipped.

control 'nutcracker-connect-redis-001' do
  impact 'critical'
  title 'Check if nutcracker can pass commands to redis'
  desc 'execute redis-cli set key command, to check connectivity of the service'

  only_if('redis is not installed.') do
    command('redis-cli').exist?
  end

  describe command('redis-cli SET test_inspec "HELLO"') do
    its('stdout') { should match /OK/ }
  end
end

Mixing this with other conditionals, such as checking existence of the files, can help to test different test paths using Chef InSpec. With this way, you can skip certain controls, which would 100% fail due to the way servers are prepared, but you know that the same control suites are reused later in different circumstances by different teams.

Some notes about only_if:

  • only_if applies to the entire control. If the results of the only_if block evaluate to false, any Chef InSpec resources mentioned as part of a describe block will not be run. Additionally, the contents of the describe blocks will not be run. However, bare Ruby expressions and bare Chef InSpec resources (not assocated with a describe block) preceding the only_if statement will run

To illustrate:

control "whatruns" do
  command("do_something") # This will ALWAYS run
  describe command("do_another_thing") do # This will not run
    command("do_yet_another_thing") # This will not run
  end
  only_if { false }
  command("do_something_else") # This will not run
end
  • Only one only_if is permitted per control block. If multiple only_if blocks are present, only the last only_if block will be honored
  • If used outside a control block, only_if skips all controls in the current file
  • To implement complex logic, use Ruby ‘or’ (||) and ‘and’ (&&) inside your only_if block:
  only_if('ready for launch') do
    rocket_is_ready && weather_is_clear
  end

Additional metadata for controls

The following example illustrates various ways to add tags and references to control

control 'ssh-1' do
  impact 1.0

  title 'Allow only SSH Protocol 2'
  desc '
    Only SSH protocol version 2 connections should be permitted.
    The default setting in /etc/ssh/sshd_config is correct, and can be
    verified by ensuring that the following line appears: Protocol 2
  '

  tag 'production','development'
  tag 'ssh','sshd','openssh-server'

  tag cce: 'CCE-27072-8'
  tag disa: 'RHEL-06-000227'

  tag remediation: 'stig_rhel6/recipes/sshd-config.rb'
  tag remediation: 'https://supermarket.chef.io/cookbooks/ssh-hardening'

  ref 'NSA-RH6-STIG - Section 3.5.2.1', url: 'https://www.nsa.gov/ia/_files/os/redhat/rhel5-guide-i731.pdf'
  ref 'http://people.redhat.com/swells/scap-security-guide/RHEL/6/output/ssg-centos6-guide-C2S.html'

  describe ssh_config do
    its('Protocol') { should cmp 2 }
  end
end

Using Ruby in InSpec

The Chef InSpec DSL is a Ruby based language. This allows you to be flexible with Ruby code in controls:

json_obj = json('/file.json')
json_obj['keys'].each do |value|
  ..
end

Ruby allows a lot of freedoms, but should be limited in controls so that they remain portable and easy to understand. Please see our profile style guide.

Core and custom resources are written as regular Ruby classes which inherit from Inspec.resource.

Interactive Debugging with Pry

Here’s a sample Chef InSpec control that uses Ruby variables to instantiate a Chef InSpec resource once and use the content in multiple tests.

control 'check-perl' do
  impact 0.3
  title 'Check perl compiled options and permissions'
  perl_out = command('perl -V')
  #require 'pry'; binding.pry;
  describe perl_out do
    its('exit_status') { should eq 0 }
    its('stdout') { should match /USE_64_BIT_ALL/ }
    its('stdout') { should match /useposix=true/ }
    its('stdout') { should match /-fstack-protector/ }
  end

  # extract an array of include directories
  perl_inc = perl_out.stdout.partition('@INC:').last.strip.split("\n")
  # ensure include directories are only writable by 'owner'
  perl_inc.each do |path|
    describe directory(path.strip) do
      it { should_not be_writable.by 'group' }
      it { should_not be_writable.by 'other' }
    end
  end
end

An advanced but very useful Ruby tip. In the previous example, I commented out the require 'pry'; binding.pry; line. If you remove the # prefix and run the control, the execution will stop at that line and give you a pry shell. Use that to troubleshoot, print variables, see methods available, etc. For the above example:

[1] pry> perl_out.exit_status
=> 0
[2] pry> perl_out.stderr
=> ""
[3] pry> ls perl_out
Inspec::Plugins::Resource#methods: inspect
Inspec::Resources::Cmd#methods: command  exist?  exit_status  result  stderr  stdout  to_s
Inspec::Resource::Registry::Command#methods: inspec
instance variables: @__backend_runner__  @__resource_name__  @command  @result
[4] pry> perl_out.stdout.partition('@INC:').last.strip.split("\n")
=> ["/Library/Perl/5.18/darwin-thread-multi-2level",
 "    /Library/Perl/5.18",
...REDACTED...
[5] pry> exit    # or abort

You can use pry inside both the controls DSL and resources. Similarly, for dev and test, you can use inspec shell which is based on pry, for example:

$ inspec shell
Welcome to the interactive InSpec Shell
To find out how to use it, type: help

inspec> command('ls ~/projects/github/inspec/docs').stdout
=> "README.md\nconfig.md\ndev\ndsl_inspec.md\ndsl_resource.md\nglossary.md\nhabitat.md\ninputs.md\ninspec_and_friends.md\nmatchers.md\nmigration.md\nplatforms.md\nplugin_kitchen_inspec.md\nplugins.md\nprofiles.md\nreporters.md\nresources\nshared\nshell.md\nstyle.md\nwaivers.md\n"
inspec> command('ls ~/projects/github/inspec/docs').stdout.split("\n").first
=> "README.md"

inspec> help command
Name: command

Description:
Use the command InSpec audit resource to test an arbitrary command that is run on the system.

Example:
describe command('ls -al /') do
  it { should exist }
  its('stdout') { should match /bin/ }
  its('stderr') { should eq '' }
  its('exit_status') { should eq 0 }
end