One of the big shifts coming in Operations Manager 2012 R2 is a re-written version of the Operations Manager agent for Linux/Unix. The big change for this agent is to provide a faster and smaller footprint for the agent which will also be better for mobile devices. Information on this change is available at:

I wanted to see what was changing as a result of this new agent, so I configured two environments as follows:

  • Operations Manager 2012 SP1 monitoring a Unix system
  • Operations Manager 2012 R2 (Preview Release) monitoring a copy of the same Unix system with a different IP address [Please note, this blog post was written on the Preview Release version of R2, the production release may be different]

This blog post will not focus on the what OMI is or how it is different from OpenPegasus. This blog will focus on any changes in the Operations Manager administrator user experience and impacts of the agent change shown through tasks in the management pack and visible changes on the Unix system.

Operations Manager User Experience – What is the same?

When using Operations Manager 2012 and Operations Manager 2012 R2 for Unix monitoring the user experience is close to identical. The following shows two Solaris systems monitored by Operations Manager (one by OpsMgr 2012 SP1, one by OpsMgr 2012 R2). In this configuration the views are identical (this makes sense of course since the management pack includes the views and the management pack is the same).

Logical Disk State view:

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

This view was identical.

image

Network Adapter State view:

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

This view was identical with the exception of the different IP address for the Solaris system (as expected).

image

Operating System Performance view:

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

This view was identical and displays the same 16 performance counters.

image

Management Pack templates:

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

Management pack templates are identical

image

Management Pack tasks:
Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

Tasks available are identical but perform slightly differently (shown later in this blog post).

image

Reporting Pane:

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

The reports appear to be identical between the different agents (this makes sense of course since the management pack includes the reports and the management pack is the same)

image

Authoring Pane: Creating new monitors

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

It does not appear that there are any changes which will impact the built-in monitor types for Unix/Linux systems.

image

Authoring Pane: Creating new rules

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

image

It does not appear that there are any changes which will impact the built-in rule types for Unix/Linux systems.

Operations Manager User Experience – What’s different?

Agent versions:

The agent version number has changed (which it should)

Operations Manager 2012 SP1: (version # is 1.4.0-906)

image

Operations Manager 2012 R2: (version # is 1.5.0-114)

image

There is a slight difference in the Solaris Computers Diagram indicating changes in the diagram view and health model. Additionally, some of the tasks behave differently for the different agents.

Diagram view:

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

The omiserver section of the health model was added to provide information on the new agent.

image

image

Operations Manager 2012 SP1:

Health model:

The health model in Operations Manager 2012 SP1 does not include a health state for the Application/Service Availability rolllup (which makes sense since we see the change reflected in the diagram view as well)

image

image

Operations Manager 2012 R2:

This includes a health state for the omiserver using the Generic UNIX/Linux Service Health as shown below.

image

image

Differences in Unix/Linux tasks in Operations Manager

Memory information task:
Operations Manager 2012 SP1:
image

Operations Manager 2012 R2:

As expected, the values returned are slightly different due to the different agents. Note the AvailableMemory value at 70 on agent monitored by Operations Manager 2012 SP1, and 96 on the agent monitored by Operations Manager 2012 R2. Also note the UsedMemory value at 441 on the agent monitored by Operations Manager 2012 SP1, and 415 on the agent monitored by Operations Manager 2012 R2.

image

VMStat task:

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

As expected, the amount of free space available in memory is higher on the system using the Operations Manager 2012 R2 agent (28016 versus 27064).
image

Top 20 CPU Processes task:

Operations Manager 2012 SP1:

image

Operations Manager 2012 R2:

image

What differences do we see on the Unix system itself?

What about on the Unix system itself? What differences can be seen on the Unix system?

Operations Manager 2012 SP1:

(using ps-ael | grep scx)

image

Operations Manager 2012 R2:

(using ps-ael | grep omi)

image

Note the difference in size (SZ = size in kb units). The total for the scx agent shown above is 10707 and the total for the omi agent is 5375.

Are there any changes to Agent installation or uninstallation?

The Unix agent is still installed and uninstalled through the same process as was documented in Operations Manager 2012. Installation worked the same from the Operations Manager console. The agent could be removed from the Unix system by uninstalling the agent from the Operations Manager console or by logging into the system and using the same steps that worked for Operations Manager 2012:

pkgrm MSFTscx

Remove the /etc/opt/microsoft/scx/ssl directory to remove the certificate.

Summary:

Microsoft’s claim that the footprint for the new agent appears to be spot-on in Operations Manager 2012 R2 (as of the preview release)

The changes in the Operations Manager agent from OpenPegasus to OMI have had no significant changes which impact the user experience in Operations Manager. This is both good and bad news. The good news is the change will be basically transparent to the Operations Manager administrator.

The bad news is if you were looking for additional functionality for the Unix/Linux agent such as:

  • Advanced logfile monitoring for Unix/Linux systems
  • The ability to install the Unix/Linux agent in a different directory
  • Baseline type monitoring for Unix/Linux systems
  • (Insert your own personal area where you think that Unix/Linux monitoring could be enhanced here via feedback to this blog post)

These don’t appear to have changed at least as of the Operations Manager 2012 R2 preview release.

I’m really impressed with what Microsoft has done by making an investment not only to monitor Unix/Linux systems but to go back and re-create the monitoring to make it more efficient. I think that shows a real commitment to making monitoring better on Unix/Linux systems in Operations Manager.