* Remove deprecated yumrepo. (#3435) * Remove deprecations for cli `--format` and metadata.rb (#3452) * Remove deprecated database_helpers stderr/stdout methods. Update deprecation text for processes/apache. * Remove deprecations for `--format` and metadata.rb Remove deprecated `format` code. Remove deprecated code test and change json-config format test to use reporter. Remove deprecated metadata.rb code Remove deprecation notice for old supports syntax. Deprecate metadata.rb from source_reader Remove rubocop disables as they are no longer required for this code block. Remove deprecated legacy metadata.rb mock profiles. Remove deprecated metadata.rb profile tests. Remove deprecated yumrepo test. * Allow inspec-3.0 branch to be tested. * Allow appveyor to test inspec-3.0 branch * Change runner tests to use reporter rather than format. Remove deprecated `supports: linux` tests. * Remove skip from inherited profiles from showing up in reporting (breaking change) (#3332) * Skip loading dependency profiles if they are unsupported on the current platform. Skip loading dependencies if they are unsupported on the current platform. Wrap our log and next in a conditional checking if the platform is supported. Change a `if !` into a `unless` Check if the backend is a Train Mock Connection and if so say that the profile does support the platform. While iterating through tests being loaded skip when the platform is unsupported. We now log a WARN when a profile is skipped due to unsupported platform, so lets check that. Modified existing test to log that there are 0 skipped tests, instead of 2. Add functional test that loads profile-support-skip with a json reporter to check that our controls are not loaded and that stderr contains our warning. * Rather than iterating through each test return before recursion if the platform is unsupported. * Resolve tests using a supported platform different from testing platform Add a control to `test/unit/mock/profiles/complete-profile` that would work on any OS with a Internet connection. This allows the profile to execute on any OS with success. `filesystem_spec.rb` was a control that would only work on Linux and some BSD's. We want profile tests to consistently work across development and testing platforms, and not get 'skipped' in some cases. Travis-CI tests on Linux, Inspec Dev team uses Linux and MacOS, Appveyor tests on Windows Also Updated `file_provider_test.rb` for `complete-profile` content changes. If you `MockLoader.load_profile` on a unsupported platform you might not hit the usual skip. Lets handle situations where the tests array in Profile#load_checks_params could be nil. * Use safe navigation rather than checking if tests is nil. Update tests to point to unsupported_inspec and account for WARN changes. Make unsupported_inspec profile support os-family 'unsupported_inspec' * Fix skip bug when using include/require controls. (#3487) * Fix skip bug when using include/require controls. * fix test and feedback. * Remove need for UUID detection for Automate report (#3507) * Add json metadata for skipped profiles (#3495) * Add skip metadata to json reports * Unify skip messages. * Update with status field. * Add testing. * Fix tests. * lint * Add skip exit codes for profile skips. * Update website for 3.0 launch Add `plugins` to sidebar. Change 2.0 -> 3.0 in slim files. Update 3.0 features list. * Fix comments * Update float to numeric. * Change Float to numeric. * updated feature list and impact doc * Change "What's new in InSpec 3.0" -> "Announcing InSpec 3.0" * Bump VERSION to 3.0.0 (#3511) * Remove 3.0 testing checks. * Fix azure link.
12 KiB
title |
---|
InSpec DSL |
InSpec DSL
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 |InSpec resource| framework is compatible with |chef compliance|.
The 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 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 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 controlimpact
,title
, anddesc
define metadata that fully describes the importance of the control, its purpose, with a succinct and complete descriptiondesc
when given only one argument it sets the default description. As of InSpec 2.3.4, when given 2 arguments (see:'rationale'
) it will use the first argument as a header when rendering in Automateimpact
is a string, or numeric that measures the importance of the compliance results. Valid strings for impact arenone
,low
,medium
,high
, andcritical
. The values are based off CVSS 3.0. A numeric value must be between0.0
and1.0
. The value ranges are:0.0 to <0.01
these are controls with no impact, they only provide information0.01 to <0.4
these are controls with low impact0.4 to <0.7
these are controls with medium impact0.7 to <0.9
these are controls with high impact0.9 to 1.0
these are critical controls
tag
is optional meta-information with with key or key-value pairsref
is a reference to an external documentdescribe
is a block that contains at least one test. Acontrol
block must contain at least onedescribe
block, but may contain as many as requiredsshd_config
is an InSpec resource. For the full list of InSpec resources, see InSpec resource documentationits('Port')
is the matcher;{ should eq '22' }
is the test. Adescribe
block must contain at least one matcher, but may contain as many as required
Advanced concepts
With InSpec it is possible to check if at least one of a collection of checks is true. For example: If a setting is configured in two different locations, you may want to test if either configuration A or configuration B have been set. This is accomplished via describe.one
. It defines a block of tests with at least one valid check.
describe.one do
describe ConfigurationA do
its('setting_1') { should eq true }
end
describe ConfigurationB do
its('setting_2') { should eq true }
end
end
Sensitive resources
In some scenarios, you may be writing checks involving resources with sensitive content (e.g. a file resource). In the case of failures, it may be desired to suppress output. This can be done 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
Exclude specific test
This shows how to allow skipping certain tests if conditions are not met, by using only_if
.
In this example the test will not be performed if 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 (like checking existence of the files etc.) can help to test different test paths using InSpec. This way you can skip certain tests which would 100% fail due to the way servers are prepared, but you know that the same test suites are reused later in different circumstances by different teams.
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 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 InSpec control that users Ruby variables to instantiate an 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 /home/gordon/git/inspec/docs').stdout
=> "ctl_inspec.rst\ndsl_inspec.rst\ndsl_resource.rst\n"
inspec> command('ls').stdout.split("\n")
=> ["ctl_inspec.rst", "dsl_inspec.rst", "dsl_resource.rst"]
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