Unable to re-create BIA Index

[former_member1324371]

Hi Guru's,

We are trying to re-create BIA index on one of the cubes , but it is failing after running for long time. Below are the messages from Job Log. Worst part is this job runs for at least 3 hours ; finally failing in filling BIA indexes !!

-------------------------------------------------------------------------------------------------------------------------------------

Loading data to index BP1_BIC:F********* (records '0000191416', job '0' )

Background process '1' complete: number of indexed records '346.521'

Background process '2' complete: number of indexed records '337.509'

Background process '3' complete: number of indexed records '344.504'

Background process '4' complete: number of indexed records '186.930'

Parallel background process for indexing table '/BIC/F*********' complete ('1')

Total number of indexed reocrds in index 'BP1_BIC:F*********' after COMMIT: '194.937.733'

START SPARSE

END SPARSE

FILLING_FAILED  >CHECK LOG!!!!!

Job finished

--------------------------------------------------------------------------------------------------------------------------------------

As mentioned in numerous other posts in this forum, I figured out to check this issue in Application Log. Upon checking SLG1, found these below warning messages:

--------------------------------------------------------------------------------------------------------------------------------------

BIA server is overloaded

Loaded data/main memory available = '5.6058333547372841E-01'

Number of times main memory has been unloaded in last 24 hours: '560'

Error while determining the SPARSE indicators for index '/BIC/F********'

TCP/IP unknown 2971

Termination while setting the delta mode for index 'BP1_BIC:SF********'

A communication error occured, with the TREX TcpIp

Server;location=************:*****,index=bp1_bic:s

---------------------------------------------------------------------------------------------------------------------------------------

I deleted BIA for the cube, created BIA index (didn't fill BIA for cube) and used the program RSDDTREX_SINGLE_TABLE_INDEX to fill BIA for ' /BIC/SF******** ' Master Data table.

It is throwing a short dump MESSAGE_TYPE_X. Attaching the snapshot. It is very weird, I have tried almost all possible options provided here in various threads but no luck !

I have checked in SE13 and this master data table doesn't have any inconsistencies. Also looks fine in RSRV checks too. I had to do a BIA Re-org as the red flag popped up couple of times, even that I haven't done during BIA filling.

We are truly desperate for solution at this point and need your expertise here. I haven't mentioned exact names for cube and MD table due to restrictions, tried to give complete issue details. Please let me know if any further details are needed. Thanks a lot in advance.

Thanks & Regards,

Vijay Dasari.

[NatsukaR · Accepted solution]

Hello Vijay,

"A communication error occured, with the TREX TcpIp" can happen when TREXIndexServer was restarted, ICM timeout (TREXADMIN -> TREX ABAP Customizing) occurred or Word process was terminated.

Can you check if a huge memory was consumed during Sparse?

This can be checked in Services -> Load tab in TREX Administorator python tool.

If the memory consumption was huge, it may be the index size is huge. Do you see if the index is split?  If the index is not split, this can caused a high memory consumption during Optimization and Sparse phases.

I guess 'initialReorg' parameter is not set to '0' or the note 1905257 did not work.  Please follow the instruction in the note 1789310 then rebuild the index.

Regards,
Natsuka

[former_member1324371 · Accepted solution]

Hello Natsuka,

Thanks a lot for your response. 'Initial Reorg' meaning , the BIA re-org that has to be done when RSDDBIAMON2 tcode indicates a reorg is required ; is it ?

Fortunately this time (probably due to a lot less loading happening during night) BIA rebuild was successful for this cube.

But we are sure again these problems will arise again particularly during Book Closure. Your response will be quite handy then.

Looks like TREXIndexServer got restarted as per your analysis probably due to high usage of BIA. Thanks again. Cheers !

Regards,

Vijay Dasari.

[NatsukaR · Accepted solution]

Hello Vijay,

Glad to hear that you are able to index the InfoCube successfully.

initialReog is not Landscape Reorg in RSDDBIAMON2 but when an index is created in BWA, initialReorg decides if the index should be split or not (Note 1313260).

Kind Regards,
Natsuka Rastert

  SAP Development Support