Updated: April 15, 2013
Applies To: System Center Operations Manager 2007
|This topic has been updated in the latest version of the System Center Management Pack Authoring Guide on the TechNet Wiki.|
Agent tasks in System Center Operations Manager 2007 are run on the agent computer where the target object is managed. An agent task can be a script or an executable program run from a command line. If an executable program is used, the application must be installed on the agent computer.
Agent tasks are useful for performing actions on the agent computer or for retrieving information for the user. They provide the following capabilities:
Run a script or command locally on the agent computer without logging on to the computer interactively.
Run a script or command on multiple agents with a single action.
Run a script or command by using local user credentials with permissions not available to the user.
Unless otherwise specified, tasks run under the credentials of the Default Action Account on the agent computer. This account typically has sufficient privileges for accessing most application components, even if the user running the task does not have these user rights. If the task is required to perform an action requiring other credentials, such as accessing an external data source, then a secure reference can be created in the management pack for running the task. Or, the user can be required to specify credentials when they run the task.
Any output sent to the standard out stream (StdOut) from the script or command is provided to the user as Task Output in the Operations Console. Command line programs will typically output information to this stream. Scripts should output information by using commands such as WScript.Echo to provide this information.
Kinds of agent tasks
Whether an agent task is configured to run a command or a script, it must specify whether it uses a probe action or a write action. This specifies the kind of module that is used by the task as described in the Composition section of this guide. There is no difference in the configuration or the capabilities between two. Probe action modules are intended to gather information without making any change to the agent computer, whereas write action modules are intended to make some change or perform an action on the agent.