If you are keen to make SqlServerDsc better, why not consider contributing your work to the project? Every little change helps us make a better resource for everyone to use, and we would love to have contributions from the community.
Please check out common DSC Community contributing guidelines.
In each resource folder there is a README.md that is the main resource documentation, parameter descriptions are documented in the resource schema MOF file. And examples are added to the resource examples folder.
On build the resources README, schema.mof and example files will be parsed and build a wiki page and conceptual help for each resource.
The parameter descriptions in the schema MOF file can contain markdown code. That markdown code will only be used in the Wiki page and will be automatically removed in the conceptual help.
The following table is a guideline on when to use markdown code in parameter
description. There can be other usages that are not described here. Backslash
must always be escaped (using \
, e.g \\
).
Type | Markdown syntax | Example |
---|---|---|
Parameter reference | **ParameterName** (bold) |
ParameterName |
Parameter value reference | `'String1'` , `$true` , `50` (inline code-block) |
'String1' , $true , 50 |
Name reference (resource, modules, products, or features, etc.) | _Microsoft SQL Server Database Engine_ (Italic) |
Microsoft SQL Server Database Engine |
Path reference | `C:\\Program Files\\SSRS` |
C:\\Program Files\\SSRS |
Filename reference | `log.txt` |
log.txt |
If using Visual Studio Code to edit Markdown files it can be a good idea to install the markdownlint extension. It will help to do style checking. The file .markdownlint.json is prepared with a default set of rules which will automatically be used by the extension.
There is a VS Code workspace settings file within this project with formatting settings matching the style guideline. That will make it possible inside VS Code to press SHIFT+ALT+F, or press F1 and choose 'Format document' in the list. The PowerShell code will then be formatted according to the Style Guideline (although maybe not complete, but would help a long way).
Any resource should be able to target at least all SQL Server versions that are currently supported by Microsoft (also those in extended support). Unless the functionality that the resource targets does not exist in a certain SQL Server version. There can also be other limitations that restrict the resource from targeting all supported versions.
Those SQL Server products that are still supported can be listed at the Microsoft life cycle site.
There are several Script Analyzer rules to help with the development and review process. Rules come from the modules ScriptAnalyzer, DscResource.AnalyzerRules, Indented.ScriptAnalyzerRules, and SqlServerDsc.AnalyzerRules.
Some rules (but not all) are allowed to be overridden with a justification.
This is an example how to override a rule from the module SqlServerDsc.AnalyzerRules.
[System.Diagnostics.CodeAnalysis.SuppressMessageAttribute('SqlServerDsc.AnalyzerRules\Measure-CommandsNeededToLoadSMO', '', Justification='The command Connect-Sql is called when Get-TargetResource is called')]
param ()
This is an example how to override a rule from the module ScriptAnalyzer.
[System.Diagnostics.CodeAnalysis.SuppressMessageAttribute('PSAvoidGlobalVars', '', Justification='Because $global:DSCMachineStatus is used to trigger a Restart, either by force or when there are pending changes')]
[System.Diagnostics.CodeAnalysis.SuppressMessageAttribute('PSUseDeclaredVarsMoreThanAssignments', '', Justification='Because $global:DSCMachineStatus is only set, never used (by design of Desired State Configuration)')]
param ()
This is an example how to override a rule from the module Indented.ScriptAnalyzerRules.
[System.Diagnostics.CodeAnalysis.SuppressMessageAttribute('AvoidThrowOutsideOfTry', '', Justification='Because the code throws based on an prior expression')]
param ()
Credential that needs to be passed to a DSC resource might include Manged Service Account (MSA), Group Managed Service Account (gMSA), and built-in accounts (e.g. 'NT AUTHORITY\NetworkService').
For a resource to support these types of accounts (credentials) the DSC resource need to ignore the password part of the credential object when it is passed to the DSC resource. We should not add separate parameters for passing such account names.
This was discussed in issue #738 and issue #1230.
The DSC resources contained in SqlServerDsc use the following naming convention:
<Module Identifier>[<Component>][<Action>]<Scope>{<Feature>|<Property>}
The following list describes the components that make up a resource name and lists possible names used for each of the components. The component names are not limited to the names in this list.
- Module Identifier
- Sql
- Component
- <none> - Database Engine (No component abbreviation)
- AS - Analysis Services
- IS - Integration Services
- RS - Reporting Services
- Action (not required)
- Setup
- WaitFor
- Scope - Where the action, feature, or property is being applied.
- AG (AvailabilityGroup)
- Database
- ServiceAccount
- Windows
- Feature
- AlwaysOn - This is for the overall AlwaysOn feature
- Endpoint
- Firewall
- Network
- Script
- Property (not required)
- Alias
- Configuration
- Database
- DatabaseMembership
- DefaultLocation
- Listener
- Login
- MaxDop
- Memory
- Owner
- Permission
- RecoveryModel
- Replica
- Replication
- Role
- SecureConnectionLevel
- Service
- State
The SqlEndpointPermission
resource name is built using the defined
naming structure using the following components.
- Module Identifier: Sql
- Component: <blank>
- Action: <none>
- Scope: Server
- Feature: Endpoint
- Property: Permission
A mof-based resource is a resource tha has the functions Get-, Set-, and Test-TargetResource in a PowerShell module script file (.psm1) and a schema.mof that describes the properties of the resource.
Please note that for the examples folder we don't use the 'DSC_' prefix on the resource folders. This is to make those folders more user friendly, to resemble the name the user would use in the configuration file.
source/DSCResources/DSC_SqlConfiguration/DSC_SqlConfiguration.psm1
source/DSCResources/DSC_SqlConfiguration/DSC_SqlConfiguration.schema.mof
source/DSCResources/DSC_SqlConfiguration/en-US/DSC_SqlConfiguration.strings.psd1
source/DSCResources/DSC_SqlConfiguration/README.md
source/Examples/Resources/SqlConfiguration/1-AddConfigurationOption.ps1
source/Examples/Resources/SqlConfiguration/2-RemoveConfigurationOption.ps1
tests/Unit/DSC_SqlConfiguration.Tests.ps1
tests/Integration/DSC_SqlConfiguration.config.ps1
tests/Integration/DSC_SqlConfiguration.Integration.Tests.ps1
The class name should be prefixed with 'DSC_Sql', e.g. DSC_SqlConfiguration.
Please note that the FriendlyName
in the schema mof file should not contain the
prefix DSC\_
.
[ClassVersion("1.0.0.0"), FriendlyName("SqlConfiguration")]
class DSC_SqlConfiguration : OMI_BaseResource
{
# Properties removed for readability.
};
Any composite (with a Configuration) or class-based resources should be prefixed with 'Sql'
In each resource folder there should be, at least, a localization folder for english language 'en-US'.
Read more about this in the localization style guideline.
Helper functions or wrapper functions that are used only by the resource can preferably be placed in the resource module file. If the functions are of a type that could be used by more than one resource, then the functions can also be placed in the common module SqlServerDsc.Common module file.
If a helper function can be used by more than one DSC module it is preferably that the helper function is added to the PowerShell module DscResource.Common. Once the helper function is in a full release (not preview) then it can be automatically be used by DSC resources in this module. This is because the DscResource.Common module is incorporating during the build phase.
For a review of a Pull Request (PR) to start, all tests must pass without error. If you need help to figure why some test don't pass, just write a comment in the Pull Request (PR), or submit an issue, and somebody will come along and assist.
If want to know how to run this module's tests you can look at the Testing Guidelines
There are stub classes for the SMO classes which can be used and improved on when creating tests where SMO classes are used in the code being tested.
There are stub modules for the modules SQLPS and SqlServer. The stub modules should be used when writing unit test so that mocks have something to hook onto without the need to have the real modules installed.
There is documentation how to generate the stub modules for the unit tests in 'tests/Unit/Stubs`. The stub modules only need to be updated when there is changes to the source module.
Integration tests should be written for resources so they can be validated by the CI.
There are also configuration made by existing integration tests that can be reused to write integration tests for other resources. This is documented in Integration tests for SqlServerDsc.
Since integration tests must run in order because they are dependent on each
other to some degree. Most resource are dependent on that integration tests
for the DSC resource SqlSetup have installed the instance to connect to.
To make sure a integration tests is run in the correct order the integration
tests are grouped in the file azure-pipelines.yml
in the integration tests
jobs.
There are three, separate, integration tests jobs that each, independently, test SQL Server 2016, SQL Server 2017 and SQL Server 2019.
When sending in a Pull Request (PR) all example files will be tested so they can be compiled to a .mof file. If the tests find any errors the build will fail.
A terminating error is an error that prevents the resource to continue further.
If a DSC resource shall throw an terminating error the commands of the module
DscResource.Common shall be used primarily; New-InvalidArgumentException
,
New-InvalidDataExcpetion
,
New-InvalidOperationException
,
New-InvalidResultException
,
or New-NotImplementedException
.
If neither of those commands works in the scenarion then throw
shall be used.
Commands are publicly exported commands from the module, and the source for
commands are located in the folder ./source/Public
.
A non-terminating error should only be used when a command shall be able to handle (ignoring) an error and continue processing and still give the user an expected outcome.
With a non-terminating error the user is able to decide whether the command
should throw or continue processing on error. The user can pass the
parameter and value -ErrorAction 'SilentlyContinue'
to the command to
ignore the error and allowing the command to continue, for example the
command could then return $null
. But if the user passes the parameter
and value -ErrorAction 'Stop'
the same error will throw a terminating
error telling the user the expected outcome could not be achieved.
The below example checks to see if a database exist, if it doesn't a
non-terminating error are called. The user is able to either ignore the
error or have it throw depending on what value the user specifies
in parameter ErrorAction
(or $ErrorActionPreference
).
if (-not $databaseExist)
{
$errorMessage = $script:localizedData.MissingDatabase -f $DatabaseName
Write-Error -Message $errorMessage -Category 'InvalidOperation' -ErrorId 'GS0001' -TargetObject $DatabaseName
}
A terminating error is an error that the user are not able to ignore by passing a parameter to the command (like for non-terminating errors).
If a command shall throw an terminating error then the statement throw
shall
not be used, neither shall the command Write-Error
with the parameter
-ErrorAction Stop
. Always use the method $PSCmdlet.ThrowTerminatingError()
to throw a terminating error. The exception is when a [ValidateScript()]
has to throw an error, then throw
must be used.
Important
Below output assumes $ErrorView
is set to 'NormalView'
in the
PowerShell session.
When using throw
it will fail on the line with the throw statement
making it look like it is that statement inside the function that failed,
which is not correct since it is either a previous command or evaluation
that failed resulting in the line with the throw
being called. This is
an example when using throw
:
Exception:
Line |
2 | throw 'My error'
| ~~~~~~~~~~~~~~~~
| My error
When instead using $PSCmdlet.ThrowTerminatingError()
:
$PSCmdlet.ThrowTerminatingError(
[System.Management.Automation.ErrorRecord]::new(
'MyError',
'GS0001',
[System.Management.Automation.ErrorCategory]::InvalidOperation,
'MyObjectOrValue'
)
)
The result from $PSCmdlet.ThrowTerminatingError()
shows that the command
failed (in this example Get-Something
) and returns a clear category and
error code.
Get-Something : My Error
At line:1 char:1
+ Get-Something
+ ~~~~~~~~~~~~~
+ CategoryInfo : InvalidOperation: (MyObjectOrValue:String) [Get-Something], Exception
+ FullyQualifiedErrorId : GS0001,Get-Something