DMO stuck in EU_CLONE_MIG_DT_RUN due to duplicate entries

[former_member182307]

Hi everyone,

I have a strange behavior with a DMO currently running for ECC.

The source ECC level is ECC6 EHP6 running on Oracle 11.2.0.3.

Target will be ECC EHP7 running on SAP HANA 1.00.111.00.1454183496.

SUM level is SP14.

We are in phase EU_CLONE_MIG_DT_RUN and have the following error message : ( file MIGRATE_DT_00201_IMP.LOG ) :

(IMP) INFO: starting create for object "EKPO~0" of type "primary index" #20160304191731
DbSl Trace:   execute(), rc=1, rcSQL=349
DbSl Trace: EXECUTE on connection 0, rc=349
ERROR exec_ddl_stmt: (DB) ERROR: DDL statement failed
(ALTER TABLE "EKPO" ADD CONSTRAINT "EKPO~0" PRIMARY KEY ( "MANDT", "EBELN", "EBELP" ) )
DbSlExecute: rc = 99
(SQL error 349)
error message returned by DbSl:

cannot CREATE UNIQUE INDEX; duplicate key found:  [5] Several documents with the same ID exist in the index;SAPSE2:EKPO.$trexexternalkey$ content not unique, cannot define unique constraint. rowCount != distinctCount

We have checked our source Oracle system and do not have any duplicate entry in it :

• On oracle this query brings no data back. Which proves the system does not have duplicate entries in this table in regards to the EKPO~0 key :

select alldata.rowid, alldata.*

from SAPSR3.EKPO alldata,

(select MANDT, EBELN, EBELP

from SAPSR3.EKPO group by MANDT, EBELN, EBELP

having count(*) > 1) keydata

where alldata.MANDT=keydata.MANDT

AND alldata.EBELN=keydata.EBELN

AND alldata.EBELP=keydata.EBELP

;

No rows selected.

• In the HANA database, the following query brings the following data back :

select T1."$rowid$", T1.*

from "EKPO" T1,

(SELECT MANDT, EBELN, EBELP

FROM "EKPO" GROUP BY MANDT, EBELN, EBELP

HAVING COUNT(*) > 1) T2

where T1.MANDT=T2.MANDT

and T1.EBELN=T2.EBELN

and T1.EBELP=T2.EBELP

order by EBELN;

	$rowid$	MANDT	EBELN	EBELP	LOEKZ	STATU	AEDAT	TXZ01	MATNR	EMATN
1	372 591	100	6500058866	2			20120423	CEFEPIME PM 256 WW B30	505000	505000
2	372 174	100	6500058866	2			20120528	5 V POWER.SUPPLY EU	MD5011	MD5011
3	370 279	100	6500059624	2			20120430	ACCUPROBE M GORDONAE	39005	39005
4	370 858	100	6500059624	2			20120430	BACILLUS CEREUS AGAR    500 GR	.00083	.00083
5	385 369	100	6500061247	24			20120517	COLOR GRAM2 R4 SAFRANIN BO2L	55548	55548
6	403 809	100	6500061247	24			20120525	PWR ENTRY MDL 115/230V	805-0017-02	805-0017-02
7	404 496	100	6500066511	4			20120710	STERILE COTTON SWABS (100)	70610	70610
8	444 828	100	6500066511	4			20120710	SEALER ASSEMBLY [R]	530111-1	530111-1
9	375 620	100	6500069747	2			20120806	YST TEST KIT VTK2 20CARDS	21343	21343
10	417 640	100	6500069747	2			20120810	NUCL. EXTR. BUFFER 3(4X1L)	280132	280132
11	426 563	100	6500070529	5			20120820	GELOSE MAC CONKEY   100 PLATES	43149	43149
12	483 095	100	6500070529	5			20120820	DIMETHYLFORMAMIDE	.4540266	.4540266
13	479 987	100	6500073722	2			20121010	SE THIO. + INDIC. 25X100ML	P1700	P1700
14	479 691	100	6500073722	2			20120924	SODIUM(DI-) HYDROG PHOSPHATE  -   (1000)	.4560125	.4560125
15	475 995	100	6500075946	2			20121010	VIDAS BARCODEREAD 3800GPDF USB	93628	93628
16	475 994	100	6500075946	2			20121010	VIDAS ANALYSER	410417	410417
17	560 158	100	6500087164	5			20130121	MUELLER HINTON 2 AGAR 90 20PL	43301	43301
18	560 650	100	6500087164	5			20130121	URETHRAL COLLECTION SWABS	39309	39309

So we have duplicate entries in the SAP HANA database for EKPO which do not exist on Oracle.

What could be the source of this issue ?

How can we continue the DMO ?

Thanks and best regards,

Steve.

[former_member182307 · Accepted solution]

Hello,

SAP connected to the system and solved the issue.

Here is what had to be done :

Analysis from SAP support :

Dear Steve,

I have found the root cause of the issue:

1. Table EKPO has more than 255 columns:

SQL> select count(*) from dba_tab_columns where table_name='EKPO';

  COUNT(*)

----------

       269

2. Table EKPO is compressed on the database level in oracle:

SQL> select table_name, compression, compress_for from dba_tables

where table_name='EKPO';

TABLE_NAME                     COMPRESS COMPRESS_FOR

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

EKPO                           ENABLED  OLTP

You are suffering from the bug described per note # 1970224 - Missing

or wrong data in OLTP compressed tables after expanding columns to be

>255.

The first step would be to decompress and reorganize the table

We did the requested actions :

- Applying the ORacle patch :

Patching component oracle.rdbms, 11.2.0.3.0...

Patch 16314254 successfully applied

Log file location: /oracle/VX1/112_64/cfgtoollogs/opatch/opatch2016-03-

24_10-29-53AM.log

- Decompress the EKPO table

SQL> alter table SAPSR3.EKPO nocompress;

Table altered.

- Reorg of the EKPO table :

BR1124I Starting 'online' reorganization of table SAPSR3.EKPO ...

BR0280I BRSPACE time stamp: 2016-03-24 10.50.31

BR1105I Table SAPSR3.EKPO reorganized successfully

BR0280I BRSPACE time stamp: 2016-03-24 10.50.31

BR1141I 1 of 1 table processed - 2227833 of 2227833 rows done

Then SAP support did this part :

the issue is resolved.

Table EKPO is imported into HANA and primary key is created.

What I have done is, I have exported EKPO and imorted it into oracle

with a different table name EKPO_BAK.

I think the crucial point was the new creation of the table

uncompressed.

Finally I renamed the EKPO_BAK to EKPO and repeated the export and

import of the table with SAPup. SAPup has continued now and is

proceeding with the comparision of table counts.

I#ll let you know as soon as this is finished.

Then we could proceed further.

Hope this helps,

Best regards,

Steve.

[Former Member]

Dear Steve,

I have the same error during System Copy,

and the table is not  compressed.
coulud I do the same with you ?

[former_member182307]

Hi,

Juste to let you know, SAP is currently looking at the issue on our side also.

Nothing new so far.

We'll probably get an answer soon.

BR,

Steve.

[Former Member]

Hello Steve,

We are facing the exact same problem with SUM SP 16 and migration of BW 7.40 to HANA SPS 10 1.00.102.04.

The oracle system do have any duplicate records, while the HANA DB have duplicate records in 8 tables.

Did you manage to solve the problem? or any solution from SAP?

Best regards,

Nahum

[former_member182307]

Hello,

The upgrade is still stuck with a message opened at SAP.

As it is aa sandbox system, it is not treated as an high priority ticket.

We did not manage to overcome this issue by oursleves so far.

We are still waiting for a solution from SAP.

I'll keep you posted when I have some more details around it.

Best regards,

Steve.

[former_member182307]

Hello again ,

I was wondering what twas the situation on your side ?

Did you open a call to SAP ?

What is the table impacted in your system ?

Thanks & BR,

Steve.

[Former Member]

Hi Steve,

The same as yours, it is the first migration on a sandbox system.

We got stuck almost 2 weeks ago. After double checking that our source system does not have a problem, I opened a ticket at SAP. Today finally, SAP requested to open the connection to the system.

The problematic tables are BW tables /BIC/*.

I will keep you update on that.

Best regards,

Nahum

[former_member182307]

Hi,

I was wondering also what is your source platform ?

In terms of OS / CPU etc ...

What is your source kernel and what are the patches you applied to the target kernel ?

On my side, I'm running Solaris 10 X86_64.

I had to add these patches on top of the SUM downloaded kernel :

• DBATL740O11_20-20012210.SAR
• R3check_310-20012208.SAR
• R3ldctl_313-20012210.SAR
• R3load_317-20012210.SAR
• R3szchk_317-20012210.SAR
• R3ta_314-20012208.SAR
• R3trans_322-20012208
• R3trans_322-20012208.SAR
• lib_dbsl_314-20012210.sar

Best regards,

Steve.

[Former Member]

Hi Steve,

We are migrating BW 7.40 SP8 running on Linux Suse X86_64, the database is  on Oracle Exadata 11.2.0.3.

Since the source system is not too old, I’ve just added

• R3load_322-20012214.SAR which was replaced later by R3load_325-20012214.SAR
• R3trans_322-20012215.SAR

Best regards,

Nahum

[former_member182307]

Hello,

Is anyone aware of a way to reload only that one table like the DMO would do but manually ?

I'd like to give a try to : delete the contents of this EKPO table and then re-upload the data .

Thanks & Best regards,

Steve.

[former_member182967]

Hello Steve,

I'm not sure if the SQL provided for Oracle is totally right. Can you take one duplicate record in HANA to search back to Oracle and check the result.

Please check this note 2262573 - Error in Phase EU_CLONE_MIG_DT_RUN cannot CREATE UNIQUE INDEX; duplicate key found

If needed, please contact with your application consultant and SAP AGS.

Regards,

Ning

[former_member182307]

Hello Ning,

I checked this note and checked the entries on the source system .

They are not duplicated on the source DB in regards to the primary index key fields.

I've opened a incident @ SAP .

Thanks & regards ,

Steve.

[former_member183326]

Please be aware this should be opened the incident under the component for ABAP Upgrade tools and not HAN-DB.

[former_member182307]

Hi Mickael,

It's opened under BC-UPG-TLS-TLA.

So I think this is O.K .

Thanks & best reagards,

Steve.