<script> ... <criticality>partial</criticality> ... </script>
Some resources has ability to execute scripts. MidPoint binds execution of scripts to specific operations. Therefore a script can be automatically executed before of after the account is created, modified or deleted. This is ofter used to supplement the accounts with aspects that cannot be managed directly by the connector. Typical use cases are creating a home directory after the account is created, creating mailbox and so on.
In addition to this the provisioning script may be also used before and after reconciliation. In this case the script will be invoked for each reconciliation regardless whether anything changed. Such script can be used to internally reconcile attributes that are not exposed directly to midPoint and and only handled by provisioning scripts. Please note that if midPoint detects a reconciliation change then a regular operation before/after script will also be invoked regardless if there is a reconciliation script or not.
Languages of the provisioning scripts depend on the specific resource (target system). For example Unix server may support bash scripting, but Windows server will only support rough command execution or PowerShell. There is no universally supported language for resource-side script execution.
Please see connector documentation for a list of supported script languages for each connector.
Provisioning scripts and midPoint script expressions
Provisioning scripts are a different kind of animal than midPoint script expressions. Provisioning scripts are usually executed in the target system and they are mostly outside midPoint control. Therefore also the provisioning script languages are only those languages that are supported by target system.
Scripts can be executed on resource or on connector.
Execution on resource is execution directly on the target system. This is what is usually needed to create home directory, manage mailboxes and so on.
Execution on the connector is a rarely-used option. This executes the script on the "client side" which usually means midPoint host or a connector server. This option is not very practical.
MidPoint can provide arguments to the scripts. This is almost always needed for the script to do something useful. For example a script must know the identifier of a created account to create home directory for that account.
Argument specification may contain midPoint script expressions that determine argument values. There is one important thing to keep in mind: provisioning scripts are usually executed on the resource, but the midPoint expressions that determine argument values are executed in midPoint. These expressions are the essential glue that can pass midPoint values to the provisioning scripts.
As provisioning script languages are connector-specific, also argument passing and formatting is connector specific. There may even be several ways how to pass arguments to the script (e.g. as command-line parameters or script variables). See the connector documentation for details. For example, for Active Directory Connector (LDAP) read also Powershell Support in AD/LDAP Connector for information how the arguments can be passed.
Since 3.6.1This functionality is available since version 3.6.1.
Error in a provisioning script is equivalent to the error in the main provisioning operation by default.
Therefore if there is an error in create script then midPoint cannot assume that an account was created.
This behavior is configurable by using the
Criticality has following meaning:
fatal: Error in the operation will cause fatal error. The processing will be interrupted, fatal error will be indicated.
partial: The error will be indicated, composite result of the operation will be presented as partial error, but the processing will not be interrupted. The operation will try to recover and continue.
The following is an example of simple powershell script configuration for Active Directory:
after the account is created (operation: add), the script will be run on the resource as a Windows PowerShell script (language: powershell)
argument (specific to the script) "username" will be set to midPoint user name
<scripts> <script> <operation>add</operation> <order>after</order> <host>resource</host> <language>powershell</language> <argument> <path>$focus/name</path> <name>username</name> </argument> <code>Add-Content c:\conn.txt $username</code> </script> </scripts>
Of course, complex scripting expressions may be used as well:
<scripts> <script> <operation>add</operation> <order>after</order> <host>resource</host> <language>powershell</language> <argument> <script> <code> username = focus.getName().toString() log.info("username: " +username) return username </code> </script> <name>username</name> </argument> <code>Add-Content c:\conn.txt $username</code> </script> </scripts>
Ad-hoc Provisioning Scripts
Since 3.7.1This functionality is available since version 3.7.1.
Vast majority of provisioning scripts are coupled with provisioning operations. E.g. the scripts are executed before or after modification of an account. However, there are cases when a script has to be executed in an ad-hoc manner. For example there may be need to execute a provisioning script from expression because the script is used to determine a value used by the expression. For that purpose there is executeAdHocProvisioningScript function in MidPoint Script Library:
<expression> <script> <code> 'homeDir=' + midpoint.executeAdHocProvisioningScript(resource, 'shell', 'echo $HOME') </code> </script> </expression>
Output of the script depends on the ability of the connector to execute script and return appropriate output. Return type of the output also depends on the connector. The output of the usual provisioning scripts is not used by midPoint and it is discarded. Therefore some experimentation and/or connector improvement may be needed to make the ad-hoc script execution work properly.
Note: both resource and resource OID can be used as the first parameter to the executeAdHocProvisioningScript function.
Object kind and script variables
MidPoint 3.0 (and later) is designed with generic synchronization in mind. Therefore the provisioning scripts are executed not just for accounts but also for other kinds of objects. And there are two major consequences of this:
Unconstrained provisioning scripts are still available. There may be valid reasons for an "universal" script that applies to all object kinds.