BW system Work process742 release kernel not releasing EM automatically

[Former Member]

Hi All,

I need your help to provide me some SAP Note or any known solution for my BW system .

The system is having 740 release having 742/200 Kernel patch which is new in market place but not releasing the EM automatically .

Only to do that we need to restart the work process manually which is crating huge problem.

The system is having Zero memory administration implemented with 742/200 kernel.

Let me know if any one has any SAP Note or any solution to the problem.

Thanks

Biswajit

[isaias_freitas]

Hello,

EM (Extended Memory) is not released... SAP keeps it allocated and controls (internally) what EM blocks are free or in use.

However, if memory is being hold by the work processes and it is freed when you restart them, please ensure:

1. That you have the latest DBSL patch applied; and
2. That you have the latest database client installed.

As a workaround, you could set the parameter "rdisp/wp_auto_restart".

Regards,

Isaías

[Former Member]

Hi isaias,

Hope you are doing fine and thanks for your quick reply .

I always appreciate your help. But here let me know the related you draw between DBSL Patch and SAP Work process EM leak?

1. Our client is 11.X.X.4. , so this is the highest patch installed.

Can you provide me some SAP Note regarding the EM leak in SAP kernel or DBSL 110 which is applied in the 11g oracle with 742/200 Kernel .

I will check the parameter in the mean time .

Please give me some evidence from which I can draw a related of EM leak or any Bug you indicate ?if.

Thanks

Biswajit

[Former Member]

Hi Isaias,

I have checked the parameter rdisp/wp_auto_restart = 28800,10 set in my system .

What you recommend to set this parameter , as 14400 is the value recommended by SAP .

Do you have any other guideline to this ?

Current DBSL is available in Marketplace 210? Do you recommend to proceed it with this Patch level?

Thanks

Biswajit

[isaias_freitas]

Hello Biswajit,

This would not be an EM leak...

It would be a memory leak, but not from the EM.

You can see the corrections / SAP notes delivered at each patch level by clicking at the "Info" link to the right of the kernel/DBSL file, at the SAP Marketplace Software Download Corner.

For the DBSL, I do not see a note addressing a memory leak.

However, I recently worked on an incident that was sent to Oracle team, and their development provided a fix to the DBSL.

I do not see a note about a memory leak at the kernel either. But notice I did not suggest to update the kernel, only the DB client and DBSL .

Regards,

Isaías

[isaias_freitas]

Hello,

Yes, apply the latest DBSL (210).

About the parameter, that would depend on your situation and system behavior.

The value 28800 would be 8 hours.

If this period is ok to you, and you do not see the server reaching the maximum memory / swapping, you could keep this value.

If the memory leak is faster, you can consider reducing the value. 14400 would be 4 hours.

Regards,

Isaías

[Former Member]

Hi Isaias,

Thanks for the information.

But you know that memory leak is a big concern in SAP kernel 740 areas as 742/50 was having some memory leak issue .

Though I feel DBSL patch should be upgraded as I faced few other issues related to DBSL patches .

But let me know if the rdisp/wp_auto_restart = 14400 is Ok for BW system or not ?

Can you please elaborate the oracle issue and what solution and how it was related to work process?

Thanks

Biswajit 

[isaias_freitas]

The DBSL is a library loaded by the work process.

Ifthe DBSL has a memory leak, you will see that the work process has a memory leak...

Now, to see what in the work process is leaking memory (e.g., the SAP kernel itself, the DBSL, the DB client, OS libraries, ...) a deep analysis has to be performed.

About the parameter value, disregarding whether it is a BW, ECC, or any other system, what really matters is how the system is behaving, and how fast the memory leak is occurring (if it is really a memory leak).

[Former Member]

Hi Biswajit,

Apart from what isaias suggest you may want to have look on below oss note regarding 742 kernel and sap memory management.

2180736 - TSV_TNEW_PAGE_ALLOC_FAILED

2148571 - Explanation for higher Extended Memory (EM) and Extended Global Memory (EG) consumption af...

Regards,

Ram

[Former Member]

Hi Ramkrishna.

I have gone through the SAP notes but my issue is different - where in my case Work process are not releasing the memory and after manual restart the memory released .

rdisp/wp_auto_restart = 28800,10 set but hope before 8 hours also the memory consumption is happening .

But this is a pattern where I cant not justify my comment that why work process are not releasing memory ?

I need to find that justification. beside Zero memory administration is already implemented .

Thanks

Biswajit

[Former Member]

Hi Biswajit,

Then apart from the wp_restart parameter you can check on below parameter as well.

rdisp/os_heap_for_restart = 1G, which will control the Heap limit usages and the further memory consumption.

Refer below note for same.

2065315 - Memory leak handling

And you can refer below wiki for further monitoring and then justifying the high memory consumption.

High memory allocation or Memory not released - Application Server Infrastructure - SCN Wiki

also you may need to check which jobs or processes are causing the heap limit to exceed.

and based on the abap program or job you can check with your functional guys for the selection criteria or how they are executing the jobs so that you can limit those high memory consumptions.

Regards,

Ram

[Sriram2009]

Hi Biswajit.

1.  Have updated the latest Windows Service pack & Secure updates from Microsoft site? kindly refer the SAP Note for Windows Service pack

30478 - Windows Service Packs.

2. Could you share the you Windows & DB version details?

Regards

SS