Run-time error "TSV_TNEW_PAGE_ALLOC_FAILED" occurred

[Former Member]

Hi all,

I am running BDLS report in background after system copy. Its getting failed. On checking the failed job log in SM37 am getting the TSV_TNEW_PAGE_ALLOC_FAILED runtime error and message as below.

I have checked and all memory parameters looks fine and similar to other systems, No other program/report is running on the server. Please help.

Memory low. Leave the transaction before taking a break!

Message no. 00072

Diagnosis

The memory on your application server is running low.

System Response

You are exclusively using one of the few work processes running on the application server. So that this does not hinder the other users too much, the work process is released if you do not work with it for a few minutes. The system terminates the transaction, causing you to lose data that you previously entered.

Procedure

Continue to work as normal and exit the transaction before you take a long break

memory parameter settings as below

ztta/roll_area 3000320 Byte

ztta/roll_first 1024 Byte

ztta/short_area 1600000 Byte

rdisp/ROLL_SHM 16384 8 kB

rdisp/PG_SHM 8192 8 kB

rdisp/PG_LOCAL 150 8 kB

em/initial_size_MB 6144 MB

em/blocksize_KB 4096 kB

em/address_space_MB 4092 MB

ztta/roll_extension 2000683008 Byte

abap/heap_area_dia 2147483648 Byte

abap/heap_area_nondia 2147483648 Byte

abap/heap_area_total 2147483648 Byte

abap/heaplimit 83886080 Byte

abap/use_paging 0

WP trace info

A *** ERROR => User SSeeeN reached avail. heapsize = 2047 MB: see ST22 [abstor.c 1766]

A *** ERROR => Increase abap/heap_area_total [abstor.c 1768]

A *** ERROR => heap memory WP0: 2047 MB [abstor.c 1771]

A *** ERROR => heap memory WP1: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP2: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP3: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP4: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP5: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP6: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP7: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP8: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP9: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP10: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP11: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP12: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP13: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP14: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP15: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP16: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP17: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP18: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP19: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP20: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP21: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP22: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP23: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP24: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP25: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP26: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP27: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP28: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP29: 0 MB [abstor.c 1771]

A *** ERROR => heap memory WP30: 0 MB [abstor.c 1771]

Please help.

Regards,

Sandeep

[Former Member]

Hai,

You have problem with the Memory, it is getting exhausted in order to execute the transaction.

Try increasing the Heap memory parameters, check SAP Note 20527,790099,146289 (if 64 bit).

Regards,

Yoganand.V

[Former Member]

You say, you started BDLS in background. We see WP0 consumed the heap memory. WP0 is usually NOT a background process.

So i suggest you check your current heap usage in transaction ST02 -> Heap memory (curr. use)

If it is low, then just retry BDLS.

Best regards, Michael

[Former Member]

Hi,

There is no problem with memory. I have tried changing the parameters. Again its the same error coming.

The BDLS background job is getting Cancelled after running for 2 hours and on checking the job log getting the dump..

I have got this info in SM21 logs.

During background processing, the system was unable to read the

specified job log. This may be due to the following:

- The instance from which you wanted to read the job log does not

have access to the central SAP directory containing the job logs:

/usr/sap/<Systemname>/SYS/global/<client>JOBLG, for example

/usr/sap/C11/SYS/global/000JOBLG for the job logs for client 000 in

system C11. Make sure your instance has access to this

directory. Examine the syslog for relevant entries.

- As job logs are written and read using the SAP TemSe subsystem,

there may be problems within TemSe.

Examine the syslog for relevant entries.

- The disk, on which the directory specified above is stored, is

full. Consequently, no new job logs can be read or written.

After which i checked the filesystem. There is no permission or access problem. I executed a small background job which has finished fine and job logs are fine. We can read via SM37.

Pls suggest.

Regards,

Sandeep

[Former Member]

Hi Mho,

Yes i started BDLS in background and its getting cancelled after running for 2.5 hours.

The below WP trace is when i am seeing BDLS job log via SM37 which is a dialog and is getting terminated with TSV_NEW_PAGE-allocation failed.

the job log file is quite huge. Is that a problem ??

-rwxrwxrwx 1 sidadm sapsys 1071026978 Jan 20 23:52 0001X21231200X62021

-rw-rw---- 1 sidadm sapsys 1071029688 Jan 21 05:28 0001X02591200X84343

I got this in SM21.

During background processing, the system was unable to read the

specified job log. This may be due to the following:

- The instance from which you wanted to read the job log does not

have access to the central SAP directory containing the job logs:

/usr/sap/<Systemname>/SYS/global/<client>JOBLG, for example

/usr/sap/C11/SYS/global/000JOBLG for the job logs for client 000 in

system C11. Make sure your instance has access to this

directory. Examine the syslog for relevant entries.

- As job logs are written and read using the SAP TemSe subsystem,

there may be problems within TemSe.

Examine the syslog for relevant entries.

- The disk, on which the directory specified above is stored, is

full. Consequently, no new job logs can be read or written.

However, i executed another small job which has finished fine and we can see the SM37 jobs

[Former Member]

Ok, so you get the TSV_TNEW_PAGE_ALLOC_FAILED when accessing the job log and not during the BDLS run. So the log is indeed to big to fit into memory.

But your BDLS was cancelled and that is why you wanted to check the log?

If yes, you can save your original log. And then split it into pieces and review it one after the other. I had a quick test on this procedure (wc -l to count the lines):

sidadm> wc -l 0001X12005102X57294
143921 0001X12005102X57294
sidadm> cp 0001X12005102X57294 0001X12005102X57294.ori
sidadm> head -70000 0001X12005102X57294.ori > 0001X12005102X57294.part1
sidadm> wc -l 0001X12005102X57294.part1
70000 0001X12005102X57294.part1
sidadm> cp 0001X12005102X57294.part1 0001X12005102X57294

Now review the first part in SAPGui.

sidadm> tail -73922 0001X12005102X57294.ori > 0001X12005102X57294.part2
sidadm> wc -l 0001X12005102X57294.part2
73922 0001X12005102X57294.part2
sidadm> cp 0001X12005102X57294.part2 0001X12005102X57294

Review the second part, and put the original file back if needed, cleanup the temp files. You may check the tail of the log first, because the error is most likely at the bottom.

Best regards, Michael

[Former Member]

Hi,

We have got the root cause of the issue. When you run BDLS for BW/APO systems where you have implemented process chains apart from changing logical system names in tables it recreates the function methods for DTP - Data transfer processes for the new logical system for which it asks for transport request numbers.

if you run BDLS in bg mode, then as it is not able to show the input screen, it goes into some kind of loop and keeps on writing to the job log file and when it exceeds 1 gb size its unable to write further and job terminates.

in such cases you have to run the BDLS in dialog mode only for inputting the transport request number.

Thanks for your help!!!

Regards,

Sandeep.