Cognos Analytics

Cognos Analytics

Connect, learn, and share with thousands of IBM Cognos Analytics users! 

 View Only
  • 1.  Where are the JVM Metrics located?

    Posted Thu October 24, 2019 04:26 PM
    cognos 11.0.9

    In the Cognos Administration and Security Guide there is "JVM Metrics" section that says there is "current heap size", "initial requested size", "maximum heap size", "up time" and "commited heap size". 

    When i go to BI Administration -> Status -> System -> Metrics - System, the only JVM metric that I see is "Up Time". How can i see the other JVM metrics? We went from CQM to DQM and are having a lot of issues. I really want to keep an eye on the JVM metrics but cannot figure out where they are located.

    ------------------------------
    brenda grossnickle
    ------------------------------

    #CognosAnalyticswithWatson


  • 2.  RE: Where are the JVM Metrics located?
    Best Answer

    Posted Thu October 24, 2019 04:45 PM
      |   view attached
    @brenda grossnickle If you select the query service you will see the JVM Mertics under the heading JVM. (Screenshot Attached) ​

    ------------------------------
    DENNY NAREZNY
    ------------------------------



  • 3.  RE: Where are the JVM Metrics located?

    Posted Fri October 25, 2019 11:58 AM
    JVM heap sizes are located in >> System> Settings >Last page.
    There are Initial and Max heap size that will need to defined. as well as, updating the garbage collection settings from generational to balanced.
    Keep in mind query engine needs to be reset or restart analytics after the changes.

    Few questions:
    Which OS analytics is installed on ?
    What are the spec's of the VM?
    Did you specify ulimit ?

    ------------------------------
    Biplob J Paul
    ------------------------------



  • 4.  RE: Where are the JVM Metrics located?

    Posted Mon November 04, 2019 11:42 AM
    So now that I know where the query service JVM metrics are located, how should i set the metrics to monitor their usage? And what do the 4 metrics mean? How can i set the metrics to monitor if the actual JVM heap size is close to or topping out at the  the maximum size limit of 8192? 

    We were getting intermittent Query Service Busy error when running reports. (We do not use cubes.) So in following the two URLs below, I increased the "Initial JVM Heap size for the query service" to 4096. And increased the "JVM Heap size limit for the query service" to 8192. Then restarted the Query Service.

    XQE-GEN-0028 errors

    Increase the JVM heap size

    https://www.ibm.com/support/pages/report-fails-your-request-could-not-be-completed-because-query-service-busy-cognos-analytics

    https://www.ibm.com/support/knowledgecenter/en/SS4Q96_6.1.1/com.ibm.help.scc.configure.doc/scc_configuring_JVM_heap_size_Cognos_Query_Service.html

     



    ------------------------------
    brenda grossnickle
    ------------------------------



  • 5.  RE: Where are the JVM Metrics located?

    Posted Tue November 05, 2019 07:45 PM

    Interesting question. Your heap size can grow up to 8GB. When the Query Service hits the max heap size, it should perform some garbage collection, discarding old records in order to make room for new requests. That garbage collection may take some time, during which the java process will be busy. 

    So what is the threshold you want to be alerted at and what will you do when you are alerted? Having the java heap full is not necessarily a problem, in fact it is a good use of resources. If you restart your query service, you are throwing out cached results that could be used to service requests without having to hit the database again. 

    The documents you linked are not directly related to the way you are using CA, but the principle still applies. If you increase the max heap size, you will have a reduction in the frequency of issues, and perhaps even avoid the XQE errors you are seeing. As far as setting a threshold on the system metric to be alerted about the heap size, it is up to you, but I'm not sure that you will be able to act on the alert. Having said that the alert can be a good way for you to understand when your max heap size may need to be increased (or conversely, decreased), according to user activity. 



    ------------------------------
    Luke Peterson
    Business Analytics Advocate
    IBM
    ------------------------------



  • 6.  RE: Where are the JVM Metrics located?

    Posted Sun October 27, 2019 06:27 PM

    You can also set thresholds on these system metrics to get alerts written to your audit log (audit database if you have it setup). It is also possible to extract the metric values at preset intervals and write to a database to get reports on the overall system health. 

    These topics are part of the System Management methodology. Although written for Cognos BI, the methodology is still applicable to Cognos Analytic metrics.

    https://www.ibm.com/developerworks/data/library/cognos/infrastructure/cognos_specific/page592.html



    ------------------------------
    Luke Peterson
    Business Analytics Advocate
    IBM
    ------------------------------