multiple instances of cscripts using high system resources RRS feed

  • Question




    I have 2 questions with regards to System resource usage by monitoring scripts used by SCOM.


    1. SCE 2007 uses 'cscripts.exe' to collect data and send back to Management server. For each rule defined in 'Authoring' we have a cscript running on the Server to collect that data. Each rule has its own collection interval, but if we went with defaults (inbuilt MP related to OS and Database only for eg) most of these collection intervals would be the same. Since these collection intervals are same, it means that around 10-12 cscripts for OS and 8-10 csripts (these are numbers for Default rules automatically enabled with the OS and SQL DB MP, without considering any other custom rules) for Database information collection would get fired roughly around the same second. In Servers having optimal resources, this scenario is fine. Since each cscript would take a few milliseconds to collect data, and even if you were watching the process console, you wouldnt be able to see them all together.


    However say in a specific case of a high user query load on a SQL Server which takes up around 95-98 % CPU (which is fine, because that is the primary purpose of that SQL Server), when 12+10 = 22 cscripts fire simultaneously, inadvertently they get 'stuck'. This is because not enough CPU resource is not available for their execution to complete. They all sit, waiting for CPU resource to be available, each cscript we see occupies from 50 to 90 MB of memory. Say an average of 75 MB of memory for 22 cscripts, that becomes 75 x 22 = 1.6 GB of RAM just for a Monitoring collection agent.


    More often than not the Server runs so low on memory that the Server stops responding, the Database on server refuses to accept new connections, and even queries running on the SQL Server could timeout.


    All this caused by a simple monitoring agent running on the Server. In some cases, I have seen around 10-12 cscript processes continously running even if enough CPU resources was available.

    My first question is: Why are each of these cscripts consuming around 60 MB each. Is there any way we can bring the memory consumption of cscripts down.


    2. My second question is:  Is there a way the 'Timeout interval' for each of the csripts be reduced (the default values for some seem to be 120 seconds and for others 300).

    We are looking to actually reduce this timeout to a lower value like say 30 seconds so that all the cscripts would be automatically killed after 30 seconds, instead of them just 'sitting' there waitiing. How and where do I go in the Authoring-> Rules console to change this timeout value?


    Thursday, September 25, 2008 2:20 PM


  • Hi Siddharth,

    This forum is for SCE related issues. For questions about System Center Operations Manager 2007, please use the Microsoft public newsgroups for SCOM (microsoft.public.opsmgr.setup) which would be the most relevant newsgroup for your question.


    Eric Zhang

    Monday, September 29, 2008 9:17 AM