AutoSys Workload Automation

 View Only
  • 1.  SAP Plugin Issues - Java Heap Memory Issues, Jobs complete in SAP but not AE/WCC

    Posted Jan 04, 2017 01:25 PM

    This question is for any SAP plugin users.  We have been experiencing intermittent outages which have been linked to threading issues (processing/exceptions/timeouts).  CA Support recommended that we apply RO93359 which is supposed to resolve these timeout/thread process issues.  However we still have issues after a period of time.  The only way to recover is to bounce the system agent, but this is only a temporary fix.

     

    We have a current support ticket open, but I wanted to know if anyone in the community had issues like this after they applied RO93359 or recently updated or patched SAP 6.0

     

    Our Scheduler runs Win 2012 R2; AE 11.3.6 SP4

    WCC Win 2012 R2; 11.4 SP3 

    System Agent Win 2012 R2; R11.3 SP4

    SAP Plugin latest version with RO93359 applied

    SAP JCO 3.0.14

     

    Thanks in advance for any feedback.

     

    Chris Kent

    Dart Container Corp



  • 2.  Re: SAP Plugin Issues - Java Heap Memory Issues, Jobs complete in SAP but not AE/WCC

    Posted Jan 05, 2017 11:03 AM

    We've similar issue and CA support had recommended that we apply T6BI004. We're yet to apply the fix but looks like it is now been superseded by RO93359. Please let me know if RO93359 fixes your issue.



  • 3.  Re: SAP Plugin Issues - Java Heap Memory Issues, Jobs complete in SAP but not AE/WCC
    Best Answer

    Posted Jan 07, 2017 05:11 AM

    We have been running stable with R11.3 SP5 SysAgent with T6BI004 ( Build 41) from August 2016 with JCo 3.0.15 . 

     

    Agent , Plug-in debug could be enabled and try to run on RSLG0000 report with SAP_TODAY variant in your LE / additional production agent to isolate the issue 

    If the delay is occurring on SAP end , you might need to review the SAP connectivity . Are we using Load balancing option ? if yes , then does the SAP message server , gateway response issue are occurring. Look at this approach also.

     

    Apparently , Most of  AE admin's would have less/ moderate understanding on SAP System landscape & even CA support L1, L2 Teams. We always tend to do trouble shoot at source end rather destination . Since, they SAP teams are key to every enterprise and they don't wish to go aggressive the SAP teams on understanding issues rather they come to  AE Admins. 

     

    Looks like , Current JCo available with SAP is 3.0.1



  • 4.  Re: SAP Plugin Issues - Java Heap Memory Issues, Jobs complete in SAP but not AE/WCC

    Posted Jan 09, 2017 08:07 AM

    I run RO93359 on win2012r2 and win2008 without problems.

    Have you tested modifying the values for:
    sap.default.connection.wait.sec
    sap.request.timeout.min
    sap.default.polling.rate.sec

    If BW is SAP note 2123865 installed? The number of available processes for backgroundprocessing is also important.

    Hopefully this might help.



  • 5.  Re: SAP Plugin Issues - Java Heap Memory Issues, Jobs complete in SAP but not AE/WCC

    Posted Jan 09, 2017 09:09 AM

    Lars/Chandra

     

    Thanks for the feedback it is appreciated.  We have not run the SAP_TODAY, we can definitely try to do that and pass that on to our technician at CA Support.

     

    Lars - We were informed that the wait.sec did not need to be defined because it was rolled up in RO93359.  CA did not allude to timeout or polling.rate.  I did see those in the troubleshooting guide and did attempt that but got mixed results.  I will be working with CA and once we do come up with a firm stable resolution, I will publish what we did.\

     

    Thanks again for the responses.  It is extremely appreciated.

     

    Thanks,

    Chris



  • 6.  Re: SAP Plugin Issues - Java Heap Memory Issues, Jobs complete in SAP but not AE/WCC

    Posted Jan 13, 2017 04:15 PM

    All,

     

    CA Support asked us to run the jmap command (Java Development Kit) to get the heap dump during an instance of this issue.  It was found that spool files (rather large) were being created and emailed to a SAP DL (as configured through the JIL - sap_step_parms).  Because of the amount of data, the heap exceeded its threshold and basically quits working.  Eliminating the spool (new_spool=n) and deleting the recipient ended up allowing us to process as we normally expected.

     

    CA is consulting with SAP regarding the plugin and how it can improve processing and thresholds on sending spool through the plugin/job definition.

     

    Again, thank you for your feedback and I hope this helps anyone that sees a similar issue.

     

    Chris Kent

    Dart Container