The following blog provides an introduction and additional information on the table comparison option of DMO: DMO: table comparison and migration tools
This blog focuses on analyzing the results of the table comparison option of DMO and discusses the possible options to handle typical checksum errors.

The table comparison option, which can be switched on as part of a DMO migration, compares the content of table rows before and after the migration.

Basically there are two general cases how differences between the source tables and the target tables can arise:

  • New entries in the source tables or in the target tables
  • Content changes in the source tables or in the target tables

Differences typically arise due to the following root causes:

  • Either the source or the target database tables are directly changed through database interfaces by other systems
  • Database inconsistencies
  • A job in the source system creates or changes entries in the source tables
  • The source system is started manually during the SUM run

Let us have a look at the last example in detail:

During the SUM run the source system is started manually. A user logs on to the system, creates a new user and adds a profile to this user.

If these actions are performed after the affected tables were migrated by DMO but before they are checked by the table comparison, which is executed after all tables have been copied, differences in the content of the tables will be detected.

2016 12 21 17 40 06 6732951

At this point we have several options to handle the checksum differences:

  1. Repeat phase: This simply repeats the calculation of the checksums for the tables where different checksums were detected. Usually, just repeating the phase will give us the same result. Anyhow, this option makes sense, if the differences on the tables were fixed manually, for example, by removing the problematic table entries on the source database and on the target database.
    To find out, which table entries lead to different checksums, we can have a look at the logfiles EUMIGRATEUTCRCRUN.LOG (in the phase EU_CLONE_MIG_UT_CRC_RUN) or EUMIGRATEDTCRCRUN.LOG (in the phase EU_CLONE_MIG_DT_CRC_RUN).In our example, we have CRC checksum differences for the table ‘USR04’:
  2. 2EETQ170 CRC differences for table 'USR04', key fields: "MANDT" "BNAME" 4EETQ171 Key: I '000' 'DDIC_TEST2'

    We can see, that the entry DDIC_TEST2 was inserted (letter I in front of the key).
    Other possible letters to describe the difference are U for updated and D for deleted.

    If the SUM finds more than 200 different rows in a certain table, the rest of the differences is summed up in blocks:

    2EETQ170 CRC differences for table 'USRBF2', key fields: "MANDT" "BNAME" "OBJCT" "AUTH"
     4EETQ171 Key: I '000' 'DDIC_TEST' 'S_USER_TCD' '&_SAP_ALL' - '000' 'DDIC_TEST2' 'C_KLAH_BSE' '&_SAP_ALL'
     4EETQ171 Key: I '000' 'DDIC_TEST2' 'C_KLAH_BSE' '&_SAP_ALL' - '000' 'DDIC_TEST2' 'S_BRAN_ADM' '&_SAP_ALL'
     4EETQ171 Key: I '000' 'DDIC_TEST2' 'S_BRAN_ADM' '&_SAP_ALL' - '000' 'DDIC_TEST2' 'S_LANG_ADM' 'S_LANG_ALL'
     4EETQ171 Key: I '000' 'DDIC_TEST2' 'S_LANG_ADM' 'S_LANG_ALL' - '000' 'DDIC_TEST2' 'S_RS_PLST' '&_SAP_ALL'
     4EETQ171 Key: I '000' 'DDIC_TEST2' 'S_RS_PPM' '&_SAP_ALL'
     4EETQ171 Key: I '000' 'DDIC_TEST2' 'S_RS_PPMAD' '&_SAP_ALL'
     4EETQ171 Key: I '000' 'DDIC_TEST2' 'S_RS_RSFC' '&_SAP_ALL'
     …

    This is done to keep the log from becoming too verbose. When a table has more than 200 rows with different checksums, there is probably a general issue with this table and we will most likely not try to correct these differences manually.

  3. For a more detailed handling of the tables with checksum differences, we can additionally select the checkbox “Accept errors and repeat phase”. Then the SUM will evaluate the files MIGRATE_UT_CRC_CHECKDIFF.LST or MIGRATE_DT_CRC_CHECKDIFF.LST, which are referred to in the dialogue:
  4. MIGRATE_UT_CRC_CHECKDIFF.LST # # By choosing "Accept errors" in the repeat dialogue, # the following lines in the list below will be handled as follows: # Line | Repeat cloning | Check again # | table contents | # ----------------------------+----------------+--------------- #  | yes | yes # # | no | yes #  ignchecks | no | no # # Remarks: # - Commenting out the line is useful if the wrong count/checksum was # fixed on the DB manually. # - In case the cloning is to be repeated, one can change the # behaviour during cloning by adding phrases like 'nosplit' # after the table name. # # Tables with different row counts/checksums: INDX # 1 bad checksum(s) CCMSBITHRH # 1 bad checksum(s) USREFUS # 1 bad checksum(s) TEMSGU # 1 bad checksum(s) USRSTAMP # 1 bad checksum(s) CCMSBIDATA # 1 bad checksum(s) USR04 # 1 bad checksum(s) NRIV # 1 bad checksum(s) ALTSTLOD # 1 bad checksum(s) TUCON # 1 bad checksum(s) CDPOS # 1 bad checksum(s) CCMSBIMETH # 1 bad checksum(s) USRBF2 # 1 bad checksum(s) USRBF3 # 1 bad checksum(s)

    The lines will be handled with the following logic:

    • The SUM will repeat the cloning of the table contents and check them again for differences for all lines, which just contain the table names. However, special cloning options can be added after the table name, like ‘nosplit’, which will prevent the splitting of the table during the cloning procedure.
    • If a # is added in the line directly before the table name, the SUM will only check the table contents again by recalculating and comparing the checksums.
    • To completely exclude a certain table from the table content check, we can add a space character and the word ignchecks after the table name. In this case, the table will not be cloned again and the table contents will not be re-checked.

    Example:

    INDX # 1 bad checksum(s) CCMSBITHRH # 1 bad checksum(s) USREFUS nosplit # 1 bad checksum(s) #TEMSGU # 1 bad checksum(s) #USRSTAMP # 1 bad checksum(s) #CCMSBIDATA # 1 bad checksum(s) #USR04 # 1 bad checksum(s) #NRIV # 1 bad checksum(s) #ALTSTLOD # 1 bad checksum(s) TUCON ignchecks # 1 bad checksum(s) CDPOS ignchecks # 1 bad checksum(s) #CCMSBIMETH # 1 bad checksum(s) #USRBF2 # 1 bad checksum(s) #USRBF3 # 1 bad checksum(s)

    In this example the tables INDX, CCMSBITHRH and USREFUS are cloned and checked again.
    The table USREFUS is cloned with the special option ‘nosplit’, so it will not be splitted during the cloning procedure.

    The tables TEMSGU, USRSTAMP, CCMSBIDATA, USR04, NRIV, ALTSTLOD, CCMSBIMETH, USRBF2 and USRBF3 will only be checked again.
    The tables TUCON and CDPOS will be ignored and excluded from the checks.

    After the execution of “Accept errors and repeat phase”, the file MIGRATE_*_CHECKDIFF.LST will only contain the remaining tables with checksum errors, if any. Otherwise it will be deleted after the successful execution of the phase.
    If certain tables have been excluded from the checks by adding ignchecks after the table name, they are listed in a new file *_CHECKIGN.LST.

    In our example, it has the following content:

    CDPOS igncount igncrcdiffs TUCON igncount igncrcdiffs
  5. Ignore phase errors and proceed to the next phase: This will skip the processing of the current check phase and directly proceed to the next phase. This option is not recommended.
    It should only be used if instructed to do so by the SAP support.

Note:
The table comparison option is useful as a preparation step in non-productive DMO runs to reveal potential issues before a productive run.
However, comparing the table content will extend the downtime, so it is not recommended to enable table comparison for a DMO run on a productive system.
As a safeguard, DMO will in any case compare the number of rows for each table (count *) before and after the migration. This cannot be switched off.

New NetWeaver Information at SAP.com

Very Helpfull

 

 

User Rating: Be the first one !