locked
Can i monitor the CPU utilization of a single server using windows 2008 management pack? RRS feed

  • Question

  • Hi,

    We don't actually monitor the CPU utilization, but one of our critical server(Windows 2008) had been running at high CPU and we had to reboot the server to bring it online. So, i'm asked if there is possibility to track which service is taking more CPU and also an alert when the CPU reaches 90%.

    I created an override for this server alone on the windows server 2008 OS pack with 90% cpu, 2 samples and 60sec interval.

    (Authoring->MP objects->Monitors-> Windows server 2008 operating system->Total cpu %)

    when i tried to create a subscription, it gives me the option to select the whole MP which means i'm getting alerts from all the windows 2008 servers.

    Can i monitor the CPU utilization of a single server using windows 2008 management pack?

     

    Please suggest.

     

    Thanks,

    Siva

    Tuesday, September 4, 2012 12:22 PM

Answers

  • Hi,

    Thank you.

    I tried the below and it seemed to fulfill my requirement.

    Information on how it has been implemented,

      • Created a group named “Servername” and added my trouble server to that group.
      • Overrides have been set with CPU=90%, Interval=120sec, Samples=2
      • Subscription has been set like, “Raised by any instance in a specific group” and created by “Total CPU Utilization percentage” rule/monitor of a “Critical” severity.

    I used stress tool and tested the alerting and i'm able to receive the alerts for that server alone.

    Thanks,

    Siva

    • Marked as answer by Graham Davies Monday, September 10, 2012 1:00 PM
    Monday, September 10, 2012 12:48 PM

All replies

  • hi Siva

    You are talking about 2 different things here.

    1) Monitoring - it sounds like you are still monitoring every server. If you want to turn this off, you can set an override against the specific monitor for enable = false for all objects of windows 2008 server (do likewise for the windows 2003 server monitor). This will stop monitoring of cpu on all the servers.

    2) Notifications - if you want to monitor the servers for CPU but only send out notifications against a specific server then on the subscription you get the chance to on the criteria page to specify "raised by an instance with a specific name".

    For saving the override:

    http://support.microsoft.com/kb/943239

    Notifications:

    http://blogs.technet.com/b/kevinholman/archive/2012/04/28/opsmgr-2012-configure-notifications.aspx

    http://technet.microsoft.com/en-us/library/dd440890

    Cheers

    Graham


    Regards Graham New System Center 2012 Blog! - http://www.systemcentersolutions.co.uk
    View OpsMgr tips and tricks at http://systemcentersolutions.wordpress.com/

    Tuesday, September 4, 2012 12:48 PM
  • Hi Graham,

    Thanks for your quick reply.

    Your are correct, the monitoring is enabled but we there is no subscription in place for the alerting.

    I am going to try the option you have given in the 2nd point for a specific server and will let you know the results soon. (not long, will use stress tool and test it)


    Thanks,
    Siva

    Tuesday, September 4, 2012 1:54 PM
  • Hi Graham,

    I selected the critiria advised by you and created a subscription. I spiked the CPU of my server to 100%, i can see the alert in SCOM console, but never received an email on that. Attched the critiria selected for my server, please review and suggest.

    Thanks,
    Siva

    Wednesday, September 5, 2012 10:36 AM
  • Hi

    Sadly, I can't see the screenshot. 

    If you have got an alert then the monitor itself is fine - it alerts when both CPU threshold and Disk Queue threshold are exceeded but that must be the case if you got an alert. 

    Can you make sure that the server is healthy before testing. If the alert was generated before the test started or the subscription was configured then you wouldn't get a notification off this alert. 

    Cheers

    Graham


    Regards Graham New System Center 2012 Blog! - http://www.systemcentersolutions.co.uk
    View OpsMgr tips and tricks at http://systemcentersolutions.wordpress.com/

    Wednesday, September 5, 2012 10:50 AM
  • Hi Graham,

    The server is healthy and i have the subscription in place before starting the test. Let me try and attach the screenshot again.

    Pls try the below url....

    http://social.technet.microsoft.com/Forums/getfile/161815

    Thanks,

    Siva

    Wednesday, September 5, 2012 11:27 AM
  • Hi Graham,


    I have my overrides for a perticular server(instance) targetted at "Windows server 2008 operating System" class. While creating the subscription, i selected "Raised by an instance with specific name" and in the entity box i selected "Windows server 2008 Operating System" and filtered it by the server name, nothing was displayed. When i left the filter section blank,i can see lot of Windows 2008 R2 ENT, Windows 2008 R2 STD instances with no server names. My question is how do i know and select which instance belongs to my trouble server where i set the overrides before?

    I tried creating a group for that single server alone and tried selecting the option "raised by any instance in specific group", but the problem is i have heartbeat, disk space and BES monitoring in place and they alert me on issues and this server is part of the above.

    I don't want to get alerted on everything, i just need a separate subscription for this server alone on CPU utilization. The reason we are planning to enable Paging in place for this notification once it is implemented.

    Please advise...


    Thanks,
    Siva

    Thursday, September 6, 2012 12:33 PM
  • Create a new monitor (don't use the one from the 2008 MP) that monitors CPU utilization, set it to the thresholds you want and have it DISABLED by default. Save the monitor to an unsealed MP (other than Default Management Pack). Enable the monitor for the specific system by overriding it.

    Create a subscription that sends out notifications based on this new monitor.


    "Fear disturbs your concentration"

    Thursday, September 6, 2012 3:39 PM
  • Hi,

    Thank you.

    I tried the below and it seemed to fulfill my requirement.

    Information on how it has been implemented,

      • Created a group named “Servername” and added my trouble server to that group.
      • Overrides have been set with CPU=90%, Interval=120sec, Samples=2
      • Subscription has been set like, “Raised by any instance in a specific group” and created by “Total CPU Utilization percentage” rule/monitor of a “Critical” severity.

    I used stress tool and tested the alerting and i'm able to receive the alerts for that server alone.

    Thanks,

    Siva

    • Marked as answer by Graham Davies Monday, September 10, 2012 1:00 PM
    Monday, September 10, 2012 12:48 PM
  • Hi Siva

    I'll mark this as closed then - thanks for the update and if you need any further assistance, please feel free to reopen the thread.

    Cheers

    Graham


    Regards Graham New System Center 2012 Blog! - http://www.systemcentersolutions.co.uk
    View OpsMgr tips and tricks at http://systemcentersolutions.wordpress.com/

    Monday, September 10, 2012 1:01 PM