Skip to Content
avatar image
Former Member

interim perform issue

We are facing interim perform issue.

The end user will report system will "hang" 2-3 mins and come back to normal. It just happen at once or a few times per day. When Basis got this message and it come back to normal already.

I want to know is there an auto tool which can tell us what is happening when user fell system "hang", such as work processores status, program running, CPU, Mem, HDD, database snapshot?

Or can we check in history such as ST03, STAT?

Or any suggest to let Basis know system is in trouble so they can capture system status on time?

Solution Manager is already implemented.

TIA

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

1 Answer

  • Best Answer
    avatar image
    Former Member
    Mar 23, 2009 at 08:04 AM

    Hi

    You can configure CCMS alerts in your system. These alerts can be configured on the work processes, the time of execution, CPU and memory usage etc. You can also generate mails whenever an an alert threshold is attained.

    This hang may also be a consequence of expensive SQL execution from the database that may cause the system to behave slow. This can be minimized however, by maintaining the variant settings to fetch the appropriate data.

    Active background jobs can also be checked at the time of slow performance. There may be jobs consuming maximum resources of the system.

    For performance monitoring:

    Workload Monitor (ST03N ) is the best transaction to check for performance issues.

    However, there are certain methods to look for things in there depending on the details available with you.

    Firstly if you are monitoring the data in runtime, you will have to check data for a specific instance. For problems occuring in past days, a TOTAL data can be viewed which would give data combined for all the application servers.

    Next , depending on the details available with you ,i.e. specific transactions or user etc. who are facing performance problem, a selection between transaction profile or user profile can be made.

    Then further, the response times can be checked. You need to check what constitutes maximum in the average response time. For e.g. a high DB time suggests that the query is taking long at the DB level.etc.

    Apart from this, real time performance problem can also be detected by transaction STAD/STAT. This would also give you ample data to be indicative on the exact issue.

    I hope this helps.

    Regards

    Chen

    Add comment
    10|10000 characters needed characters exceeded

    • Former Member

      Additionally, you can tell to user to call you when he hangs, then look at:

      - SM50 -> shows you the workprocesses

      - depending on your OS: top (for unix), task manager (windows) -> to see which process is running

      - depending on your OS: disc activity, for example sar on unix

      - if the SM50 shows the user sitting on a db table, then ST04 -> Wait Event Analysis -> Session Monitor

      Check, if several users are affected (at the same time or not). Look out for something unusual, for example paging activity.

      Best regards, Michael