Dude. Got R2? Then go get this update!
Seriously.
Operations Manager 2007 R2 Management Pack version 6.1.7533.0 from the MP CATALOG.
This is basically a set of the core MP’s in OpsMgr R2. You should expect to see these get updated on a fairly regular basis now. These are the MP’s that are built in to OpsMgr, that monitor the health of the Management group, from the server role, and agent role perspective. There should be a SP1 version of this update available soon to follow. These are a series of updates to OpsMgr, based on Community, Customer, and internal feedback…. and it’s good stuff.
Everyone running R2 should run these through your standard testing cycle and get these implemented. This is a simple MP import-update.
The MOMTEAM Blog has some more details here: http://blogs.technet.com/momteam/archive/2009/10/07/opsmgr-2007-r2-mp-version-6-1-7553-0-is-released.aspx
This includes a long list of updates (available in the guide in the download)… but I will hit a few high points.
- You might have seen some of my blog posts on agents restarting all the time, in SP1 and R2, and the significant impact that can have:
http://blogs.technet.com/kevinholman/archive/2009/06/22/health-service-and-monitoringhost-thresholds-in-r2-how-this-has-changed-and-what-you-should-know.aspx
Well, this is mostly resolved in this update…. as the default threshold for HealthService and MonitoringHost has been changed from 100MB to 300MB. This will stop the majority of your agents from hitting this limit, and restarting. Once you import this MP update – you should review any overrids you made on these Monitors… and make sure you dont have any conflicts. The only other override I like to recommend is to generate an alert on these monitors – so that you will know when there is a problem that will cause the agent to get restarted – specifically on HealthService or MonitoringHost, and PrivateBytes or HandleCount:
You might find that there are still some agents that need a much higher threshold…. but those should be fairly rare, and limited to very large servers with a very high instance count of objects (large SQL clusters, large Exchange servers, etc) and also possibly for agents that are a watcher node or proxy for a large number of devices (network, VMWare, etc…)
- In THIS blog post I discussed adding a custom perf counter collection rule and creating a view to see how many consoles or SDK connections you have. This is now built-in:
- Lots of Knowledge and descriptions added to help troubleshoot alerts. For instance – on the common “Script or Executable Failed to run”
Summary
The Health Service was attempting to run an executable and was unable to create the process. This may affect some monitoring or discovery.
Causes
This can be caused by:
- The executable could not be found.
- The computer does not have enough resources (for example; memory) to run the executable.
- The antivirus software on the computer blocking Visual Basic scripts or Java scripts. The following link is a link to the KB article regarding this issue.
- Antivirus software blocking script execution
Resolutions
The alert description and context has information indicating which rule or monitor failed. The following link will display all events indicating a failure to run the executable:
- View Batch Response Events
- After reviewing the error in the context, check:
- That the path to the executable exists on the computer.
- The antivirus software is not blocking scripts from running.
- That the computer is not over utilized.
- Check Task Manager to see if there is enough free memory.
- Check Task Manager to see if there are any processes consuming all the CPU.
- Check the error information in the event or alert context for the script path and name. There could be a problem with the script not handling an error correctly and exiting. If the script exits without outputting the data that is expected (e.g. property bag data), this error is raised.
- Another error can be caused by the misconfiguration of the workflow executing this script. The configuration (script params, policy, timeout) could be wrong causing no output or timeout.
There is a big list of other enhancements or fixes in the MP guide. Don't take this lightly – this is good stuff…. get it updated.
