cancel
Showing results for 
Search instead for 
Did you mean: 

configuration of memory parameters - TSV_TNEW_PAGE_ALLOC_FAILED

Former Member
0 Kudos

Dear Community,

my Users get the Error TSV_TNEW_PAGE_ALLOC_FAILED when performing several tasks or Transactions in SAP.

Our SAP System is on Windows 2008 r2 and the DB is Microsoft SQL

The VM has 48 GB Memory and 24 GB are assigned to the SQL Server.

The memory parameters in SAP are:

abap/heap_area_total                  6000000000

abap/heap_area_dia                    6000000000

abap/heap_area_nondia              0

abap/heaplimit                              40894464

em/initial_size_MB                       20480

ztta/roll_area                                3000000

ztta/roll_extension                        2000000000

ztta/max_memreq_MB                 2047

abap/shared_objects_size_MB    20

PHYS_MEMSIZE                         20480


I hope I didn’t forget any relevant parameters.

When in look in st02 I see the max use values as follows:

Do you have any suggestions for me, how to configure the parameters?

I don't want to just increase one value an causing a lack of perfomance at other circumstances.

I already ordered a book about SAP Performance analysis, but the Problem is to serious to wait for it.

Do you need any other information?

thanks for your help

Kind regards

Franz

Accepted Solutions (0)

Answers (6)

Answers (6)

Former Member
0 Kudos

Thank you for your support.

I changed the

I decreased the parameter em/initial_size_MB and changed the parameters as Alwina suggested.


The problems are gone so far.


Kind Regards


Franz

former_member221372
Participant
0 Kudos

Hello Franz,

Usally this error comes because of some points like,

1.Memory issue, so please check in SM04 t.code whether any of other programs are running which is taking too much of memory.

2. Check your query means how many records it is fetching from the server.

Cheers,

Bharat

former_member221372
Participant
0 Kudos

Its solved by TMG Events....

Cheers,

Bharat

Former Member
0 Kudos

This dump occurs in various transactions.

For example in Transaktion fbl5n when a user tries to change a long list he has gererated. But there are other Standard SAP Transactions, where it occurs too.

thanks for your suggestion nirmal , I will decrease the value em/initial_size_MB as soon as possible.But this is probably not the reason for the error, as you can see below. Or maybe I didnt quite understand.


thanks for your tip mercedes, i tried to find our scenarion in the wiki, but when couldnt.


When I look in the dump in ST22, i find following Ammount of memory used in the moment of the interruption:


Roll-Bereich................... 2718272

erweiterter Speicherbereich(EM) 2002720576

fest allokierter Speicher(HEAP) 1706998000

Short-Bereich.................. " "

Paging-Bereich................. 40960

Max. Adreßraum................. " "

but thoese values are lower then in those in our parameters.

Thanks for your help and kind regards,

Franz

Brindavan_M
Contributor
0 Kudos

Hello Ertl,

The best solution to identified the issue , do one restart and clean the shared memory using " cleanipc <nr> remove. and then check if the dumps occurred.  If you still facing the same issue then kindly provide all the information about the parameters.

how much free space left in RAM?.

em/initial_size_MB

em/blocksize_KB

em/max_size_MB

em/address_space_MB

ztta/roll_first

ztta/roll_area

ztta/roll_extension

abap/heap_area_dia

abap/heap_area_nondia

abap/heap_area_total

abap/heaplimit

rdisp/ROLL_MAXFS

rdisp/ROLL_SHM

rdisp/PG_MAXFS

rdisp/PG_SHM

abap/buffersize

zcsa/presentation_buffer_area

zcsa/table_buffer_area

zcsa/db_max_buftab

rsdb/obj/buffersize

rsdb/obj/large_object_size

em/initial_size_MB

rsdb/ntab/irbdsize

ztta/roll_area

ztta/roll_extension MB

will suggest some parameter to tune.

Thanks,

Brindavan

alwina_enns
Employee
Employee
0 Kudos

Hello Franz,

from your parameters a user can allocate 2GB in extended memory:

ztta/roll_extension                        2000000000

and from your dump is seen 2GB in extended memory were really allocated:

erweiterter Speicherbereich(EM) 2002720576

The other amount of memory in your dump (heap):

fest allokierter Speicher(HEAP) 1706998000

is limited by these parameters:

abap/heap_area_total                  6000000000

abap/heap_area_dia                    6000000000

abap/heap_area_nondia              0              

A user in a dialog process can allocate 6GB and in a non-dialog process can allocate unlimited amount od memory. But the users altogether cannot allocate more than 6GB (parameter  abap/heap_area_total). Since you see less than 6GB in the heap in dump, then you should check, how many user were on the system at that time and if they altogether have allocated 6GB in heap. You need to increase the value for abap/heap_area_total if you have enough RAM and space in SWAP on the server.

To get a better overview about memory consumption please take a look at the history in ST02 part "Extended and heap memory (MEM)". You will have then an idea, if the application server requires more memory and in which area.

Regards, Alwina

Brindavan_M
Contributor
0 Kudos

Hi Eril,

kindly let me know this dump is occurring for z*program or any other standard program. let me know the details. If its Z*program than followup with user there must be a recent variant implemented in system and searching for million of records. check with ABAPers to split the variant for the same. This might be also temporary bottleneck neck ask run the job in background and check the same.


Give more details on this issue.


Thanks,

Brindavan

mercedes_fernandez
Active Participant
0 Kudos

Hello Franz,


you could find information about how to analyze this scenario in

:TSV_TNEW_PAGE_ALLOC_FAILED - Application Server Infrastructure - SCN Wiki

kind regards,
Mercedes

nirmal_konchada
Active Contributor
0 Kudos

Hello Franz,

You need to check what exactly the user is trying to do. Without knowing it and proper understanding its not suggested to increase any parameter.

As per the information provided by you, you have assigned 24 GB to the SQL and the parameter em/initial_size_MB   is set to 20 GB. By looking at all this it clear that you dont have any space for increasing any parameter until you add some RAM to it or else based on the performance history you can reduce the size of EM is its not optimally used and then increase few other buffers.


First thing you can do it try to get the clear information on what the user is trying to do.

Regards,

Nirmal.K