1. All Databases:
UNIX: Adjustment of directory structures
When is it essential to adjust the directory structure before starting the export?
One should adjust the directory structure before starting the export, if the source system is upgraded from a SAP system with a release level earlier than SAP NetWeaver 2004s, and is not a new installation.
How to get the directory structure adjusted?
SAP 1104735 should be applied.
Table Splitting Preparation: Missing *.WHR files
What is the common issue which occurs during the table splitting preparation?
For every table which is to be split, the corresponding *.WHR files are not created in the export directory /ABAP/DATA.
How to get this issue resolved?
Fewer packages should be created for the tables for which no *.WHR files could be produced, using the following steps:
1. A new and empty installation directory should be created.
2. A new and empty export dump directory should be defined.
3. The Prepare Table Splitting service should be Run again, and an input file should be provided which must contain only the missing tables with a lower number of packages for each defined table.
These results should be merged with the results from the first Prepare Table Splitting run, if the *.WHR files have been created for the missing tables:
4. The *.WHR files should be copied from /ABAP/DATA to /ABAP/DATA.
5. The lines should be added from /ABAP/DATA/whr.txt to /ABAP/DATA/whr.txt.
Table splitting: perform export/import with MigMon
When the table splitting feature is used, then the export and import must be performed with the Migration Monitor.
The name of the export directory that is specified during the export, should not end with /DATA.
Otherwise, the .SQL files that may be located in the export directory /DB/ are not found by the SAPinst.
Specifying the Export Directory
It should be ensured that the path does not contain any blanks, while specifying the 'Export Directory' on dialog screen 'ABAP System > Export'.
Error during export of DDLOG package
The SAP 928474 should be checked if a separate package for table DDLOG has been created while splitting the STR files.
Ignore "- No upgrade-prepare is performed"
The phrase "- No upgrade-prepare is performed" in section: Preparations -> Technical Preparations can be ignored in the documentation "Homogeneous and Heterogeneous System Copy for SAP Systems based on SAP Web Application Server 6.40 SR1":
Prerequisites for an export:
: "- No upgrade-prepare is performed"
# characters in R3load *STR, *.cmd, *.TOC files
The resulting STR files may contain '#' characters, if Java based STR splitter from the NW04 or NW 04 SR1 installation kit or version 1 from the SAP service marketplace are used. The export should not be performed with these STR files and a corrected version of the SPLIT.SAR archive should be downloaded as described in SAP 784118.
The SPLIT.SAR archive should be extracted into the 'installation' directory for the export after SAPinst copies the installation files (xml files, libraries, ...) into this directory.
Unicode Conversion: additional space requirement
While performing a unicode conversion, the calculated size for table DYNPSOURCE is wrong. This causes the data import into the target system to get aborted due to the missing space in the database respectively tablespace or dbspace. The database or database container, in which the table DYNPSOURCE is to be created in the target database, should be enlarged. The required size for the table will be fifteen times larger than in the non-unicode source system.
2. SAP DB / MaxDB:
Heterogeneous copy: Missing BLOB entries
What is the issue which occurs during a heterogeneous system copy with MaxDB assource system database?
During a heterogeneous system copy with MaxDB assource system database, often corrupt BLOBs occur.
How can this be eliminated?
SAP 1043735 should be applied, before performing the export.
SAPINST ERROR (example Windows):
ERROR 2005-05-06 13:59:23
MOS-01012 PROBLEM: 'E:usrsapT11SYSexerun/R3szchk.exe -s DB -f
G:export/DATA/SAP0000.STR G:export/DATA/SAPAPPL0.STR
G:export/DATA/SAPAPPL1.STR G:export/DATA/SAPAPPL2.STR
G:export/DATA/SAPCLUST.STR G:export/DATA/SAPDDIM.STR
G:export/DATA/SAPDFACT.STR G:export/DATA/SAPDODS.STR
G:export/DATA/SAPPOOL.STR G:export/DATA/SAPSDIC.STR
G:export/DATA/SAPSDOCU.STR G:export/DATA/SAPSLEXC.STR
G:export/DATA/SAPSLOAD.STR G:export/DATA/SAPSPROT.STR
G:export/DATA/SAPSSEXC.STR G:export/DATA/SAPSSRC.STR
G:export/DATA/SAPUSER.STR G:export/DATA/SAPUSER1.STR -l
./R3szchkExport.log -p G:export/DATA/ -t ada T11 640' returned with '-1073741819' which is not defined as a success code.
Solution: see 846757
Unicode conversion only:
The encrypted passwords for building secondary connections become invalid and must be maintained again, if it is converted from non-Unicode to Unicode system.
Solution: see SAP 842788
Unicode migration only:
SAP 1043735 should be read before starting the unicode migration.
4. MS SQL Server:
Heterogeneous BW system copy: Missing P indexes:
The latest version of R3load should always be used for DB export.
5. R3load procedure:
Export: Choosing the right data file codepage.
The default 'Data File Codepage' is '1100'. This value is suitable only if a non-unicode system is to be exported and the target system is also non-unicode. A different data file codepage needs to be chosen, if the target system is a unicode system or a user wants to export from a unicode system. It is recommended to convert during the export, and enter the data file codepage which corresponds to the endianness of the target system. One can see SAP 552464 for more information.
System copy of MDMP system:
What is the common problem associated with copying an MDMP system using the R3load unload/load method?
When a user wants to copy an MDMP system using the R3load unload/load method, then it becomes mandatory to enter 'MDMP' as DB code page on the dialog screen 'General Load Parameters' during the import phase. Non-numeric values (like MDMP) are not allowed to be entered as DB code page, by the XML files delivered with the installation pack.
How to get this issue resolved?
The file dialog.xml should be edited in the installation directory before starting the installation:
1. The string 'fiDBCodePage_i' should be searched.
2. The two lines should further be corrected by removing the
'type="numeric"' instruction:
OLD: 1100
NEW: 1100
3. Since the correction needs to be done twice, hence step 1+2 should be repeated.
6. Oracle:
Oracle 10.2 for products based on SAP Web AS 6.40 - backward release
An installation directly with Oracle 10.2 is possible only with the SAP versions described in SAP 940794.
Exporting the source system:
The export can be performed with any of the available installation master DVDs released for Oracle 10.2, if the source system runs with Oracle 10.2 but there is no new installation master DVD for the product.
Installing the target system:
The target system has to be installed with Oracle 9.20, if there is no direct installation with Oracle 10.2 available for the product to be installed. The up gradation of the database to Oracle 10.2 should be done afterwards.
Error in sapinst_dev.log: (...)runCatprocSql
It only occurs during the installation of the oracle patch 10.2.04.
Error:
FCO-00011 The step runCatprocSql with step key NW_Onehost ind ind ind ind 0 0 NW_Onehost_System ind ind ind ind 1 0 NW_CreateDBandLoad ind inind ind 10 0 NW_CreateDB ind ind ind ind 0 0 NW_OraDBCheck ind ind ind ind 0 0 NW_OraDBMain ind ind ind ind 0 0 NW_OraDBStd ind ind ind ind 3 NW_OraDbBuild ind ind ind ind 5 0 runCatprocSql was executed with status ERROR .
How to get this error resolved?
One of the following options can be chosen:
* The latest SAPinst version from SAP Service Marketplace should be downloaded and the installation must be repeated.
* The file control.xml should be edited in the installation directory as follows:
a) Search for "this.runCatproc =
b) Go to "script.execute_AbortOnFailure();" some lines below.
c) Replace "script.execute_AbortOnFailure();" with "script.execute();"
d) Restart sapinst and continue with the installation.
CJS-00081 Assertion failed: Component OraDbBuild: Context parameter tDirectoryToDriveMapping, no drive for directory origlogA found:
This error occurs for the installation method "System Copy/Migration by Reload: Refresh the database content of a non-MCOD system (R3load)".
How can this error get resolved?
Change control.xml:
old:
kdb.remove("tORA_DbServerDirectories_context", "WHERE dbSid = ''")
if (installer.getGlobalParameter("MigType") != "OBR" &&
installer.getGlobalParameter("MigType") != "RLD" &&
installer.getGlobalParameter("MigType") != "RLDMMN" )
new:
kdb.remove("tORA_DbServerDirectories_context", "WHERE dbSid = ''")
if (installer.getGlobalParameter("MigType") != "OBR" )
The installation should be restarted with this new control.xml file.
Windows Oracle: In chapter "Generating the Control File Structure" of the system copy guide for the database specific system copy procedure the following information should be added as Step 1:
1. If the database needs to be migrated from 32-bit to 64-bit or vice versa, the following lines should be added at the bottom of the control.sql file for Oracle 9.2 or Oracle 10.2:
Oracle 9.2: shutdown immediate
startup restrict
spool utlirp.log
@?/rdbms/admin/utlirp.sql
spool off
alter system disable restricted session
Oracle 10.2 shutdown immediate
startup upgrade
spool utlirp.log
@?/rdbms/admin/utlirp.sql
spool off
Shutdown immediate
Startup
Spool utlrp.log
@?/rdbms/admin/utlrp.sql
Spool off
Exit
Then the other steps of the procedure should be continued.
R3szchk: Correct calculation of tablespace sizes in DBSIZE.XML:
How to get this error resolved?
* Get the latest R3szchk from service marketplace.
* start R3szchk to recreate the DBSIZE.XML with the options R3szchk -t ora
Export with Migration Monitor, but without sockets:
Why does this error occur?
It occurs if the user chooses 'Export using Migration Monitor' on the SAPinst dialog 'ABAP System > Database Export', without selecting the 'Using Sockets' option.
What is the problem which occurs due to this error?
SAPinst tries to export the packages again, after the Migration Monitor exports all packages successfully. SAPinst is not able to recognize certain packages, if a package has been split into more than 10 packages, but creates new TSK files and exports these packages again.
How can this problem be solved?
* The export steps should be performed with SAPinst until the message box 'Configure and the Export Monitor in server mode' gets started and appears.
* Choose 'Cancel', then 'Stop' and exit SAPinst.
* The Migration Monitor should be started and configured to perform the export.
* The file keydb.xml should be edited before restarting SAPinst, when the export has been performed successfully.
* The string 'tWhat' should be searched from the beginning of the file. One will reach the line '
* The string 'MMN' should be searched and replaced by 'MMN_SC'.
* SAPinst (SAP NetWeaver '04 Support Release 1 / ABAP System / Oracle / {Unicode|Non-Unicode} / ABAP Database Content Export) should be restarted and 'Continue old installation' should be performed for the missing steps of export process.
* ‘OK’ should be chosen when the message box 'Configure and start the Export Monitor in server mode' appears.
Wrong initsid.ora with new tablespace PSAPUNDO:
When the system copy is performed using OraBrCopy the correct initsid.ora gets overwritten by SAPinst after the files CONTROL.SQL and initsid.ora get copied.
How to get this error resolved?
The control.xml file should be corrected:
1. Close SAPinst while SAPinst stops for installing Oracle software with the Oracle RUNINSTALLER.
2. Search in control.xml file for:
installer.getGlobalParameter("MigType") "OBR" &&
installer.onWindows();
3. Replace with
installer.getGlobalParameter("MigType") "OBR"
4. Restart SAPinst and continue the old installation.
Now SAPinst stops again after the RUNINSTALLER step and you can copy again the correct initsid.ora to the Oracle folder $ORACLE_HOME/dbs/init.ora or windows $ORACLE_HOMEdatabaseinit.ora
5. The installation should be continued.
The next problem also gets fixed with this correction.
error "ORA-00205: error in identifying controlfile,...":
The target database is created with only one control file under /oracle//920_/dbs, when the system copy is performed using OraBrCopy. One needs to copy cntrl to those directories which are defined in init.ora for the parameter control_files. Else one can get the error "ORA-00205: error in identifying controlfile, the alert log should be checked for more information when the system is started.
Migration Monitor: Mistakable popup text:
What is the common problem which occurs while using Migration Monitor for system copy?
The Migration Monitor starts the R3load processes (in server mode), when the Migration Monitor is used for a system copy. The popup in SAPinst displays (for export): "Configure and start the Export Monitor in server mode". On pressing the OK button the SAPinst will execute 'PostExportActivities' step though the export is not yet ready.
How can this issue be solved?
When the Migration Monitor finishes the data load/unload successfully, then only the system copy with SAPinst should be continued.
7. IBM DB2 UDB for UNIX and Windows:
R/3 Load Method:
Export:
1. If the target database is IBM DB2 UDB for UNIX and Windows, then new R3szchk should be downloaded.
Before the export of the existing SAP System is started, the current version of R3szchk should be downloaded from SAP Service Marketplace.
2. The downloaded file should be copied into directory /usr/sap//SYS/exe/run/
3. Windows only:
One should make sure that he has logged on to his source system that is running on DB2 UDB for UNIX and Windows, as user adm BEFORE you start SAPinst. Otherwise, the user might encounter the following error when switching the user context:
ERROR 2005-03-01 11:49:40
FSL-06002 Error 1326 (Logon failure: unknown user name or bad password) in execution of a 'LogonUser' function, line (554), with parameter (adm).
Import:
1. Prior DB2 UDB Version 8 Fixpack 7: Log File full error on large tables might occur.
2. Before one starts SAPinst, the file DDLDB6.TPL should be edited on the export directory: AFTER_LOAD should be replaced with BEFORE_LOAD.
Backup Restore Method:
Preparation
It is recommended to use the brdb6brt-tool because of its features to backup the database and generate the redirected restore script. (See Database Administration Guide: SAP on IBM DB2 Universal Database for Unix and Windows )
1. Get the latest version of brdb6brt from SAP Service Marketplace.
2. A backup of source-database should be performed.
(If your source-database consists of multiple nodes you have to perform a backup of all database nodes)
Note: useful options for brdb6brt are
-nn Nodenumber
-replace = ... if you change th SID
-bm 3 ... backup and generate
redirected restore script
Note
"Database Connect User" is an input parameter during the dialog phase on the screen "Database Specific Paramters". The default value should be replaced by the name of the connect user from your source system. As the schema cannot be changed during restore, hence this is necessary.
Normal procedure (Single Partition-Systems)
See "System Copy for SAP Systems based on SAP Web AS 6.40"
Procedure for Multipartition-Systems:
The following steps should be performed in addition to normal procedure:
At Exit-Step "Restore of Database"
7.1 Additional participating DB-Hosts should be installed via sapinst
(if appropriate)
7.2 Platform dependent actions should be performed.
7.2.1 Unix-Platform
Check/Adjust Directory Ownership and Permissions of the directories in /db2/ on the Target System
to owner : db2
group : dbadm
Permission : rwxr-xr-x (755)
Adjust db2nodes.cfg and .rhosts
(For Details about db2nodes.cfg and .rhosts refer to
DB2 V8 Administration Guide [Implementation].)
7.2.2 Windows-Platform
The additional Nodes should be created via db2ncrt
(For details about creation of additional nodes on NT via db2ncrt
refer to
1. DB2 V8 Administration Guide [Implementation]
2. SAP Web Application Server ABAP 6.40 on Windows :
IBM DB2 UDB for Unix and Windows )
7.3 Stop/restart DB2 and verify , that all nodes are starting
7.4 Database Nodes beginning with Node 0 should be restored.
Possible Errors during restore
NT-only
If the following error occurs
SQL2036N The Path for the file or device <...> is invalid and device is a network drive
How to get the above mentioned issue fixed?
UNC names should be used in from clause of restore command in the generated restore script "from hostsharedirectory"
Possible Errors during RUNSTATS
If one encounters Problems during runstats, then this step should be skipped or newest dmdb6srp executable should be installed on the system.
Postprocessing
If one wants to adjust his tablespace-names to the standard conventions, the Instructions described in the documentation "System Copy for SAP Systems based on SAP Web AS 6.40" should be followed.
8. IBM DB2 UDB for OS/390, z/OS:
This step fails with an error reported in the file R3ldctlExport.log, if the environment variable DB2DB6EKEY has a different value from the default value.
ERROR: DbSlConnect rc= 29
DbSl Trace: DB2 Call 'SQLConnect' Error:
sqlcode = -30082 : [IBM][CLI Driver] SQL30082N
Attempt to establish connection failed with security reason "24"
("USERNAME AND/OR PASSWORD INVALID").
What is the reason for the above mentioned failure?
The reason for the failure is that sapinst uses the default value for DB2DB6EKEY. The default value of DB2DB6EKEY is "value of environment variable dbs_db2_ssid + value of environment variable SAPDBHOST".
The default value of DB2DB6EKEY would be D8H0ihsapac, if dbs_db2_ssid = D8H0 and SAPDBHOST = ihsapac.
DB2DB6EKEY should be set back to the default value and run dbdb2pwd -create , in order to be able to continue the export. This SAPinst step should be repeated as described in SAP 731937.