How to create a custom UNIX provider for System Center 2012 Configuration Manager SP1

How to create a custom UNIX provider for System Center 2012 Configuration Manager SP1

  • Comments 1
  • Likes

~ Niel Orcutt | Senior SDE

GearsIn this article, I describe the process for setting up a development environment for creating and deploying a custom provider for AIX, HP-UX and Solaris (UNIX) clients for System Center 2012 Configuration Manager Service Pack 1.

Configuration Manager providers are developed using the open-source Open Management Infrastructure (OMI) development environment ( I will use the code from one of the samples provided with the OMI software to illustrate this process.

The tasks that need to be done to develop and deploy a new provider are:

I. Install and configure the OMI development environment

II. Create and compile the provider

III. Deploy the provider on the UNIX client computer

IV. Register the provider with the Configuration Manager server

Note that the same process can be used to develop custom providers for System Center 2012 Operations Manager R2 clients for UNIX computers.


Custom providers for Configuration Manager UNIX clients are developed using the open source OMI environment.

Development Machine Configuration. The UNIX computer used to develop OMI providers must have its native software development environment (compiler, linker, make program and runtime libraries) installed. In addition, it should have the GNU make package, the OpenSSL developer package and the gunzip utility program installed.

Getting the OMI Development Environment. The version of OMI needed for developing Configuration Manager providers is 1.0.6. This can be downloaded by going to the Open Group website at:

Selecting the download link "omi-1.0.6-tar.gz" at the bottom of the page.

Select a directory on the UNIX computer under which OMI will be installed. Go to the directory and use the following commands to extract the OMI sources to a subdirectory named omi-1.0.6 in that directory:

gunzip <path>/omi-1.0.6.tar

tar xf <path>/omi-1.0.6.tar


Patching OMI. If you are using a UNIX computer, a patch must be applied to the OMI 1.0.6 files to be able to use OMI. The patch changes some of the source files for OMI to compile and run under UNIX. It also sets the permissions on the UNIX files in OMI and changes the Windows line endings in many of the files to UNIX line endings. For development on Linux, the patches are not necessary.

To patch OMI, obtain the following file from


Uncompress the omi-1.0.6-patches.tar.gz file using gunzip. Change to the omi-1.0.6 directory that was created when the OMI archive was extracted and extract the resulting omi-1.0.6-patches.tar file there. After extraction, there should be a new subdirectory named omi-patches in the omi-1.0.6 directory.

From the omi-1.0.6 directory, run the following command to make the modifications necessary to OMI for use on UNIX:



Building OMI. Once OMI is installed and patched, from the omi-1.0.6 directory, execute the following command to configure the OMI build environment for your specific UNIX computer:

./configure --prefix="`pwd`/output"

Then execute the command make

On UNIX computers, this uses the UNIX make to run GNU make (gmake) to build each project. You may not invoke the GNU make (gmake) program directly here as the top-level make requires the UNIX make program.

This produces the OMI components needed to build providers. These components will be in the output/bin directory under the omi-1.0.6 directory.

Setting up the OMI Tools. The OMI tools omicli and omireg are shipped with the Configuration Manager client for UNIX and are installed in its directory. These tools need no further setup. The omigen tool is built by the make command described above. It must be set up from the developer build of OMI that was just built.

Set up of omigen can be done by setting the LD_LIBRARY_PATH environment variable in the shell that will be used to run omigen (or the SHLIB_PATH variable on HP-UX PA-RISC computers that use HP-UX 11.23). Execute these commands to set the variable:



Note that if the Configuration Manager client was installed to a directory other than /opt/microsoft (the default), you will have to change the /opt/microsoft in the command lines to use the correct path name.

If you are using the OMI tools in a System Center 2012 Operations Manager R2 environment the setup process is slightly different. This is described in the README.TXT file that is included in the OMI patch downloaded from the link above.


Creating the source directory. Create the source directory for the new provider. In this example, I create it under the omi-1.0.6 directory that OMI was installed in. I named it FrogProvider, because I will be using the Frog sample provided.

Defining the Provider MOF file. The structure of the information gathered and sent by the provider is defined in a text file in MOF format. The MOF file describes the fields in a class, somewhat similar to a class definition in the C++ programming language.

For this example, I will use the Frog Provider that OMI 1.0.6 provides. I created the following MOF file in the FrogProvider subdirectory:

#pragma namespace("\\\\.\\root\\cimv2")
[Description("Some information about frogs")]
class XYZ_Frog
[Key] string Name; // the name of a frog
uint32 Weight; // the weight of a frog
string Color; // the color of a frog

The #pragma line is ignored by OMI but is used by the Windows mofcomp program described later in this article.

In my project, I named this file schema.mof.

I created the following GNUmakefile in the FrogProvider directory:

TOP = ..
include $(TOP)/config.mak
module.cpp \
XYZ_Frog_Class_Provider.cpp \
stubs.cpp \
INCLUDES = $(TOP) $(TOP)/common
ifeq ($(OS),LINUX)
CXXFLAGS = -fno-exceptions -fno-enforce-eh-specs -fno-rtti
ifeq ($(OS),AIX)
CXXFLAGS = -blibpath:/usr/lib:/lib
LIBRARIES = micxx io base
include $(TOP)/mak/rules.mak
## gen -- rule to regenerate provider sources.
LD_LIBRARY_PATH=$(TOP)/output/lib export LD_LIBRARY_PATH ;
$(TOP)/output/bin/omigen -C $(TOP)/share/omischema/CIM_Schema.mof
-A --cpp schema.mof XYZ_Frog

NOTE: The last three lines above are a single line in the GNUmakefile.

The make variable TOP refers to the location of the OMI install, which in my case is the omi-1.0.6 directory. The “-C $(TOP)/share/omischema/CIM_Schema.mof” refers to the location of the CIM base classes used by omigen.

Generating the provider source files. From the FrogProvider directory, I typed:

gmake gen

This invoked the omigen program which read the schema.mof file I had created. Omigen created files called schema.c, stubs.cpp, module.cpp, module.h, X_Frog_Class_Provider.cpp, X_Frog_Class_Provider.h, and X_Frog.h in the FrogProvider directory.

The files created by omigen are all used as-is except for the X_Frog_Class_Provider.cpp file. This file must be edited to contain the code that gathers the information for the provider and posts it to OMI.

In the X_Frog_Class_Provider.cpp file, I replaced the single line (the Post of MI_RESULT_NOT_SUPPORTED) in the EnumerateInstances method with the lines in that same method from the file doc/omi/samples/frog/XYZ_Frog_Class_Provider.cpp in the directory where OMI was installed.

This adds two instances of Frog to the provider that are populated with data and posted to OMI.

Load, Unload, and EnumerateInstances are the only methods that are used by the Configuration Manger client. If you will be developing System Center 2012 Operations Manager R2 providers, the GetInstance method must also be defined to match a specific instance name and return only the instance(s) whose names match the give name. Operations Manager providers also require support for the keysOnly parameter in the EnumerateInstances method, which is not used by the Configuration Manager client.

Compiling the provider. After the XYZ_Frog_Class_Provider.cpp file was modified, I typed the following to create the provider file:


After gmake had completed, the provider runtime was placed in the output/lib directory under the omi-1.0.6 directory in a file named This file would have been named if the client were an HP-UX PA-RISC computer.


The Configuration Manager install directories and processes are only modifiable from the root account, so once the Frog class provider was built I logged in to the computer as root and changed to the omi-1.0.6 directory. I then copied the new provider to the Configuration Manager installation directory using this command:

cp ../output/lib/ /opt/microsoft/omi/lib

I then registered it with the Configuration Manager client using these commands:

/opt/microsoft/omi/bin/omiserver –s
/opt/microsoft/omi/bin/omireg –n root/cimv2 /opt/microsoft/omi/lib/
/opt/microsoft/omi/bin/omiserver &

These commands stop the omiserver daemon, register the Frog provider binary, and restart the daemon.

I then logged out from the root account and logged in to my account again. I tested the provider on the client using this command:

/opt/microsoft/omi/bin/omicli ei root/cimv2 XYZ_Frog

This produced the following output:

instance of XYZ_Frog
[Key] Name=Fred
instance of XYZ_Frog
[Key] Name=Sam


To configure the Configuration Manager site server to collect frog data as part of hardware inventory from UNIX clients, I copied the schema.mof file created for the provider to the Configuration Manager site server. On the server, I opened a command prompt window as Administrator, changed to the directory where I had copied the schema.mof file, and typed the following:

mofcomp schema.mof

I then opened the Configuration Manager console application and clicked on the Administration workspace near the bottom left side of the window. Next, in the navigation pane I clicked on Client Settings, right-clicked on the Default Client Setting object in the details pane and selected Properties. In the Default Settings screen, I clicked on Hardware Inventory and then clicked Set Classes. From the bottom of the Hardware Inventory Classes screen I selected Add and then in the Add Hardware Inventory Class screen I clicked Connect to connect to the root/cimv2 namespace on that computer. In the Connect to Windows Management Instrumentation screen, I clicked Connect. Next, in the Add Hardware Inventory Class screen, from the list of class names I selected the XYZ_Frog class and then clicked Edit. Then, in the Class qualifiers screen I entered a friendly name for the Display name, and clicked OK.

In the Hardware Inventory Classes screen, from the list of inventory classes, I located the Frog item and made sure that the box next to it was selected. I expanded the Frog item and made sure that all the boxes next to the property field names were checked. After this was done, I clicked on OK to close the window, and then clicked OK to close the Default Settings window and save the configuration changes.

The next time the Configuration Manger server received an inventory from my UNIX client, I was able to view the Frog information when I used the Resource Explorer to view details about my client computer.


There are two different things called namespaces in the CIM environment:

1. An identifier for the place where data about a class and its objects is stored. In Configuration Manager this is usually root/cimv2.

2. The part of a class name before the first underscore character. In the case of the XYZ_Frog class, this would be XYZ. The first character in a class name must be alphabetic. Some CIM tools require that all class names begin a namespace name followed by an underscore and that the character after the underscore also be alphabetic. You may need to follow this convention in naming your custom classes.

Note that this posting is provided "AS IS" with no warranties and confers no rights.

Niel Orcutt | Senior SDE | Microsoft

Get the latest System Center news on Facebook and Twitter:

clip_image001 clip_image002

System Center All Up:
System Center – Configuration Manager Support Team blog:
System Center – Data Protection Manager Team blog:
System Center – Orchestrator Support Team blog:
System Center – Operations Manager Team blog:
System Center – Service Manager Team blog:
System Center – Virtual Machine Manager Team blog:

Windows Intune:
WSUS Support Team blog:
The AD RMS blog:

App-V Team blog:
MED-V Team blog:
Server App-V Team blog:

The Forefront Endpoint Protection blog :
The Forefront Identity Manager blog :
The Forefront TMG blog:
The Forefront UAG blog:

Your comment has been posted.   Close
Thank you, your comment requires moderation so it may take a while to appear.   Close
Leave a Comment