DX Unified Infrastructure Management

 View Only
  • 1.  nas

    Posted Jun 20, 2019 06:39 AM
    I have a question regarding nas actually . How to send an alarm for the people about the events that are not coming continuously in CA UIM  or whenever the data is missing in CA UIM , events are missing in CA UIM I need to create a alarm and send them . Please share me the process or kb  to do this so that I can send the alarms whenever data is not coming continuously or events are not coming continuously . 


  • 2.  RE: nas

    Broadcom Employee
    Posted Jun 20, 2019 09:02 AM

    HI,

    Unfortunately this too generic to really provide you an answer.
    out of the box, UIM does not have any built-in monitoring for missing events or data because every customer setup is unique.

    I do know clients who have written sql queries to monitor incoming QOS or information they expect into the database and then use the sqlserver probe to check these custom queries and send alarms based on this. Support would not have an example of this to provide but maybe someone else in the community might.

    if you can provide more details on exactly what you would like to monitor with use case examples we might be able to provide more direction.



    ------------------------------
    [Designation]
    Principal Support Engineer
    Broadcom
    ------------------------------



  • 3.  RE: nas

    Posted Jun 20, 2019 11:39 AM
    I think that perhaps the word you want to search the forums for is "heartbeat". 

    It winds up being quite difficult to tell when something is wrong based on a lack of evidence that it's right. 

    Gene hit on the crux of it though - you need to identify something that reliably happens at a predictable period and then test for that.

    And keep in mind that logmon can run a command or script too so if you already have a manual process that does the test then you can use logmon to run that manual test automatically.


  • 4.  RE: nas
    Best Answer

    Posted Jun 21, 2019 02:58 AM
    What I normally use (as example):
    - take a QoS that is generated by each robot, example: QOS_CPU_USAGE
    - verify that you executed the QoS view sql script (so your qos can be accessed by V_..)
    At that moment you can execute a SQL query where you can check for robots that didn't publish this QoS the last xx hours: (example query)
    ---
    select a.robot, a.origin, MAX(b.sampletime) as last_sampletime
    from CM_NIMBUS_ROBOT a with(nolock)
    inner join V_QOS_CPU_USAGE b with(nolock)
    on a.robot = b.robot
    group by a.origin, a.robot
    having MAX(b.sampletime) <= DATEADD(hh, -1, getdate())
    order by last_sampletime
    ---
    - if this query returns what you want, you can add this query in sql_response and generate an alarm for each robot in the output.  Or you can create a custom Jasper report and schedule this and mail yourself the output,...