ESP Workload Automation

 View Only
Expand all | Collapse all

Report on long running jobs in ESP

  • 1.  Report on long running jobs in ESP

    Posted Mar 01, 2019 10:18 AM

    Does anyone have any ideas on the best way to report on jobs in ESP that are running longer than normal? We would like to have some type of daily or weekly report that shows this information so we can build some trend charts.

     

    Thank you,

    Sharon



  • 2.  Re: Report on long running jobs in ESP

    Posted Mar 01, 2019 04:41 PM
    Hi Sharon, 
    I am hoping others weigh in on this..... This solution is not ideal. 
    The basics:
    Every job at the end of every stream and critical jobs in the stream need to have a late time. This includes standalone jobs. 
    When the job goes late it triggers the NOTIFY OVERDUE ( LATE ) at the top of the appl and notifies the responsible application team.  Depending on the LATECRIT for the job it creates a ticket, pages out, or just sends an E-Mail. CRIT and LATECRIT were something I set up for each job and part of the doc for the job. 
    Tracking what went late::
    The LATE application contains a IEBGENER job that appends the information about the late job (JOB QUAL APPL GEN current time......) to a GDG. This creates a basic list of what went late and when. This can be sent to the application teams to see if the late time needs to be moved or whether they need to address the issue. Monthly a new +1 of the GDG is created. This process is fairly easy to add. 
    Addionally
    It is is possible to run a ESP report job for the jobs in the late list to see when they actually completed. 
    The two outputs can be combined using REXX. This is more involved. 
    .  

    Don



  • 3.  Re: Report on long running jobs in ESP

    Posted Mar 03, 2019 02:12 PM

    Hello Don,

     

    Thanks for the response. Can you think of anything we can do by using history data to report on this...jobs running longer than normal? Maybe for some type of trending? Currently we do have some MAXRUNTIMEs set for a few jobs but not for many and we are looking something where we can report on any job that runs longer than the normal run time, now it may be ok that it did but need a way to report on this.

     

    Thanks,

    Sharon



  • 4.  Re: Report on long running jobs in ESP

    Posted Mar 04, 2019 08:24 AM

    Sharon,

     

    I wonder if you could use Report Server for this.  Once it's gathering history, you could have a batch job that queried the history for jobs that ran over a certain length of time, then output that report in a way that you could use the data in Excel, Tableau, etc.

     

    <JC>



  • 5.  Re: Report on long running jobs in ESP

    Posted Mar 04, 2019 11:10 AM

    Thanks, currently we are not using Report Server but are looking to do so hopefully soon.



  • 6.  Re: Report on long running jobs in ESP

    Broadcom Employee
    Posted Mar 04, 2019 08:38 AM

    Hi Sharon,

     

    Firstly please post in the new link in the future:

    CA Mainframe Workload Automation 

     

    The history report has following two fields:

    • EXECQT
      Specifies the elapsed time for job execution.
    • AVGRUNT
      Specifies the average elapsed execution time for the job.

    So you may DISPLAY both fields, and if EXECQT is greater than AVGRUNT, then the run is longer than normal. Note: if you will need to use REPORTCV (instead of REPORT) and export the report to a excel and then do the comparison.

     

    Hope this helps,

     

    Lucy



  • 7.  Re: Report on long running jobs in ESP

    Community Manager
    Posted Mar 04, 2019 08:52 AM

     I moved this to the still somewhat new CA Mainframe Workload Automation community. 



  • 8.  Re: Report on long running jobs in ESP

    Posted Mar 04, 2019 11:21 AM

    Thanks Lenn!



  • 9.  Re: Report on long running jobs in ESP

    Posted Mar 04, 2019 11:21 AM

    Hello Lucy,

     

    The only thing I would say with this one is AVGRUNT could get off I would think if you were to have a couple of long runs that were out of the norm.

     

    What are you thoughts around that?

     

    Thank you,

    Sharon



  • 10.  Re: Report on long running jobs in ESP

    Broadcom Employee
    Posted Mar 04, 2019 12:16 PM

    Hi Sharon,

     

    ESP doesn't have a way to exclude some runs of the same JOB. 

     

    Something you may consider to clear/ignore all the past runs:

    - Remove the related JOBSTATS record for the JOB by running UTIL;

    - Define and use JOBPROF for the JOB, so that the default profile by job name is not used.

    - When possible, add job qualifier, like JOBAAA.1, then the default profile JOBAAA.1 will be used instead of JOBAAA.

     

    Hope this helps,

     

    Lucy



  • 11.  Re: Report on long running jobs in ESP

    Posted Mar 04, 2019 02:56 PM

    Thanks Lucy, let me look at all of this.



  • 12.  Re: Report on long running jobs in ESP

    Posted Mar 05, 2019 12:26 PM

    Lucy, why does the AVGRUNT field contain a '-' at times?

     

    EXEC Avg
    QTIME Run
    1             -
    2             -

    0             -

    4             1M00

     

    Also can EXECQT and AVGRUNT be formatted differently?

     

    Thanks,

    Sharon



  • 13.  Re: Report on long running jobs in ESP

    Broadcom Employee
    Posted Mar 05, 2019 01:55 PM

    Hi Sharon,

     

    The format of the two fields EXECQT and AVGRUNT can't be changed.

     

    There are two common reasons that AVGRUNT shows the value of '-':

    - The value is rounded to zero minute. See below on how AVGRUNT is calculated:

    How is AVGRUNT calculated in the History Report fo - CA Knowledge 

    - The job is not submitted by ESP, so there is no JOBSTATS record for it. You can issue LJS JOB(jobname) I and see if it's the case.

     

    As you may not know, actually AVGRUNT is also available on ESP CSF, enter PR, you can see "Average RunTime".

     

    Hope this helps,

     

    Lucy



  • 14.  Re: Report on long running jobs in ESP

    Posted Mar 05, 2019 02:13 PM

    Thanks Lucy.