Menu Close

Monitoring Microsoft Endpoint Configuration Manager (MECM)

Quick Download:  https://github.com/thekevinholman/MECM

image

The old System Center Configuration Manager 2012 Management Pack is no longer available.  Customers often used this to monitor current branch, and it still works well for the most part.

Now that it was pulled, there is a gap for customers who don’t have that old MP, or want something a little more relevant to MECM deployment monitoring.

I have taken the old SCCM 2012 MP’s, and combined them into a single MP.  It largely works the same way, but I made some enhancements based on things I saw in the field.

The biggest changes are around controlling noise for transient errors, and changing the alert source to the Unit monitors, as alerting from rollup monitors was a terrible design as a method to limit noise, because you also limit quality detailed alerts.  Fixed some workflows that flat out didn’t work, and aligned the MP closer to best practices.

  • MECM Client Class discovers sitecode and version properties
  • Fixed errors on MECM client discovery that logged events
  • Added MatchCount property to most monitors to support multiple consecutive samples to reduce noise
  • Removed overrides that were muting alerts from unit monitors
  • Added overrides to disabled alerts from the aggregate rollup monitors and dependency rollup monitors
  • Added ScaleBy to Process based CPU Rules and Monitors so they actually work.
  • Changed Max sample separation on Perf Collection rules from 10 to 4 so one data point will be collected every hour on any perf collection rules that are enabled.
  • Deleted all Manual Reset Monitors and replaced with Rules.
  • Deleted all Classes, Rules, and Monitors for Deprecated and Unsupported roles
  • Renamed all Element ID’s to include the MP ID, and follow structured rules:
    • Datasource ends with .DS and with Displaystring Datasource
    • ProbeAction ends with .PA and with Displaystring Probe Action
    • WriteAction ends with .WA and with Displaystring Write Action
    • MonitorType ends with .MT and with Displaystring MonitorType
    • Discovery ends with .Discovery and with Displaystring Discovery
    • Rule ends with .Rule and with Displaystring Rule
    • Monitor Ends with .Monitor and with Displaystring Monitor
    • Aggregate Monitor ends with .AggregateRollup.Monitor
    • Dependency Monitor ends with .DependencyRollup.Monitor
    • Groups end with .Group and with Displaystring Group
    • Task ends with .Task
  • Removed all Override Displaystrings which were pointless
  • Reduced frequency of MECM Client Discovery from 60 seconds to once a day.
  • Created an override to disable the SMSExec service monitor for members of the Site Database Computers Group, since this was causing false alarms as Database servers are not typically Site server roles with SMSExec service present.

image

image

 

You can download this management pack here:

https://github.com/thekevinholman/MECM

 

If you have any feedback or monitoring requests, I will be happy to consider them for future updates.

12 Comments

    • Kevin Holman

      This is a new MP with new ID’s, so it will not upgrade the old one. The SCCM 2012 MP’s should be removed, but you can run them side vby side for a short time if you need to. It will be confusing because some objects in the MP will have duplicate names. Obviously, customers will have a lot of overrides referencing the old MP’s so removing can be a pain. I’d still recommend it, because the old overrides likely won’t apply directly so I’d use it as a chance to start fresh.

  1. Ollie Woodall

    Great news! Do we understand why MS ceased development on SCCM MPs?

    Would there be any problem having this new MP imported, side-by-side with the SCCM 2012 MP? While we worked through the overrides required in the new MP (we have a LOT of overrides for the original 2012 MP).Or should you remove SCCM 2012 MP first?

    • Kevin Holman

      You can run the new one side by side. However, I’d keep it short, because some object names like groups will be duplicated.

        • Kevin Holman

          Meaning I’d make it a project to get the old one removed as soon as you can. I would not run both of them for months and months, because it could cause noise and confusion.

  2. Jay

    Hi Kevin,

    I just installed this MP and one of the groups we support that have no SCCM related objects in their server group/email subscription received email alerts from multiple servers that aren’t theirs for the critical alert MECM Windows Deployment Service Not Running. Is this a known issue?

    • Kevin Holman

      No. It isn’t. At least, not yet.

      MECM WDS Availability Monitor (MECM.PXEServicePoint.WDS.Service.Monitor) targets the class: ConfigMgr PXE service point (MECM.PXEServicePoint).

      If you look at discovered inventory for this class, do you have discovered objects that you do not expect in there?
      If you look at the group definition for the group scoping the people receiving alerts – how is it defined?

  3. David Zemdegs

    Thanks heaps for that. I uninstalled the old one and installed this new one. I get a monitor alert about the notification port not being open. According to the doco I needed to open TCP port 10123 which I did but the monitor is still actively alerting. What I couldnt find was how to check which port the alert monitor is looking for?

    • Kevin Holman

      The Monitor for client notification port (like many monitors in this MP) gets this data directly from MECM. MECM publishes component state and severity in the registry, the monitors read this registry value straight from what MECM thinks about the health of itself:

      MECM.BGBServerFirewallBlock.StatusMessage.Monitor

      SOFTWARE\Microsoft\SMS\Operations Management\Components\SMS_Notification_Server\D07ACE61-FB84-4461-9F52-ABBA07C2EE3A\State

      SOFTWARE\Microsoft\SMS\Operations Management\Components\SMS_Notification_Server\D07ACE61-FB84-4461-9F52-ABBA07C2EE3A\Severity

      If Severity = 2 or 3 the monitor is marked unhealthy. If Severity = 0 or 1 the monitor is healthy. So if your monitor is unhealthy – then MECM thinks there is a problem with this component. You should have status messages in MECM reflecting this?

      • David Zemdegs

        the bgbserver.log has a few errors around specific clients timing out but nothing that suggests an entire port is not open. Looks like an erroneous alert. Override time.

Leave a Reply

Your email address will not be published.