This article is an extension of SAP Note 1775293 – Migration/system copy to SAP HANA using latest SWPM 1.0

Whenever we update the SAP Note, we will also write a few lines in this article. So, by using the feature “Receive email notifications”, you will be updated as soon as the SAP Note is updated.

09. April 2015

In case you are using HANA Client software SP9 and you want to use virtual hostname. Please check oss-note 2141442.

06. February 2015

An error occurs during the execution of the initial HANA Landscape Reorg:

Error during execution of HdbCmdClazz ‘EXECUTE_DATABASE_REORG’ : Fri Feb 06 14:31:10 CET 2015
| com.sap.hdb.core.main.cmd.HdbCmdMain | For input string: “undefined”

This error only occurs for scale-out system in case the file “HDB_TABLE_CLASSIFICATION.TXT” does not exist on the export medium in the directory “ABAP/DB”. This is the case in older versions of the ABAP report SMIGR_CREATE_DDL. In case the error occurs just create an empty file like:

touch <exportMedium/ABAP/DB

and hit the retry-button on the error pop up of the software provisioning manager . The error will be fixed with version SWPM 1.0 SP7 PL5.

05. February 2015

SAP HANA Multi Tenant Database Containers and SAP System on the same host.

This scenario is supported with software provisioning manager SWPM 1.0 SP7 PL4.

26. January 2015

software provisioning manager SWPM 1.0 SP7 PL3

The new version of software provisioning manager version SWPM 1.0 SP7 PL 3 offers the following new features:

  • Create, recreate and drop Tenant Databases.
    SWPM now directly creates new tenant databases during the installation. The tenants are “simple” tenants (one node, no port specification).
    It is also possible to drop tenant databases during uninstall.
    In case you want to have a more specific tenant database (using several nodes, special ports) you still have to create them by hand before the installation. In this case you don’t want to use the option “recreate tenant database” as it only recreates a simple tenant. In one of the next versions the software provisioning manager will reuse the host- and port specification during the recreation of an existing tenant database.
  • Using tenant database backups for system copy.
    It is now possible to use an existing backup of oyur tenant database for system copy.
  • Distributed export.
    Using the option “System copy towards SAP HANA” it is now possible to use several SAP application server to export the source database. Please check the SAP Systemcopy guide for more information. It should be updated within the next few days.
  • New system copy features:
    NewSystemCopy.PNG
    Use DDL mapping file
    This feature is already available within the migration monitor for a long time. We have decided to also offer it through the software provisioning manager.
    The feature offers the possibility to use a different TPL file for a specified set of tables or packages. In case of HANA this is usefull for tables where you want to have the primary key created differently.

    Manual Execution of Migration Monitor.
    This feature is an alternative to the existing feature “Start Migration Monitor Manually”:
    StartMigrationManualy.PNG

    In case you want to use the new feature, do not set the checkbox for “Start Migration Monitor Manually”.
    The differences between the old and the new option are:
    – Only the step about import or exporting data (the feature is also available for export) is omitted. All other steps are executed.
    – You will receive a Pop Up once the step is reached.
    – The migration monitor properties file is fully configured. So you can also use this to find out how SWPM configures the migration monitor.
    – In case the option is used for a SAP HANA scale-out system, the table creation phase and the landscape reorg is still executed by the software provisioning manager.  It will then stop before the data import.

    Omit tables from import
    With this feature you have the possibility to skip tables or packages from the main import. You might want to use this feature in case you have special tables you want to load in a different way or on a different server.

    You feedback is always wecome.

14. January 2015

Error during an installation/system copy involving a scale-out system whereby the tenant is not located on the master host.
Error message: System (Last error reported by the step:Process call

‘/usr/sap/ERP/SYS/exe/uc/linuxx86_64/R3load -testconnect’ exits with error code 2. For details see log file(s) R3load.exe.log.).

The hdbuserstore on the SAP host shows the following entries:

<sid>adm 1> hdbuserstore list KEY DEFAULT
  ENV : ld9968.wdf.sap.corp:30044;ld2432.wdf.sap.corp:30044;ld9970.wdf.sap.corp:30044
  USER: SAPERP

In this example, the tenant is located on the node ld2431. The software provisioning manager enters the potential master hosts of the
scale-out system in hdbuserstore in descending order.If the tenant’s master service is located on a node other than the first node in hdbuserstore, and if
this first node can be accessed, the client tries to connect to this host. This action fails because the tenant is running on a different node.
The first node in hdbuserstore must be the node on which the tenant’s master service is running. To determine the relevant node, you can execute the following SQL statement on the system database:

select * from SYS_DATABASES.M_SERVICES where database_name = ‘<TENANT>’ and COORDINATOR_TYPE = ‘MASTER’

The user store is then corrected using this command:

hdbuserstore set DEFAULT “ld2431.wdf.sap.corp:30044;ld9968.wdf.sap.corp:30044;ld2432.wdf.sap.corp:30044;ld9970.wdf.sap.corp:30044” SAPERP <password>

The client should then be able to successfully connect to the host. As soon as the node ld2431 fails, all services are moved to the next free potential
master host. The client automatically connects to this node.

The fix will be available in February.

22. December 2014

In case you are using software provisioning manager 1.0 SP7 you will recognize that some roles are missing after the installation or system copy.
The HANA roles ABAP_* and AFL* are not created due to a bug within HdbSlLib. You can create them manually after the installation by running the following command from the installation directory of the software provisioning manager:

./sapjvm/sapjvm_6/bin/java -classpath com.sap.hdb.sl.lib.jar:com.sap.hdb.core.jar:ngdbc.jar com.sap.hdb.core.main.cmd.HdbCmdMain HDB_CMD_CLAZZ_NAME=UPDATE_REPAIR_ROLES_AND_PRIVILEGES JDBC_PATH=. DATABASE_SID=<DBSID> DATABASE_HOSTNAME=<hostName> INSTANCE_NUMBER=<Nr> DATABASE_SYSTEM_USER_PASSWORD=<Password> DATABASE_SQL_USER_NAME=<NAME> DATABASE_SQL_USER_PASSWORD=<Password>

The fix will be available with SWPM 1.0 SP7 PL3.

21. November 2014

SAP HANA SP 9, Multitenant database containers

Starting with SP9, SAP HANA supports so-called multitenant database containers. With this feature, it is possible to run multiple instances (tenants) within a single database. Tenants are fully seperated from each other.

The software provisioning manager also supports this new feature.
With the first version of software provisioning manager 1.0 SP7 (provided as part of Software Logistics Toolset 1.0 SPS12), the following procedures are possible :

  • Installation and deinstallation of an SAP System using an existent tenant
  • System copy/migration of an SAP system via R3load using an existent tenant

We plan an update of software provisioning manager 1.0 SP7 in January 2015 to support further procedures like:

  • Creating and dropping tenants during installation and deinstallation
  • System copy using Backup/Restore of a tenant
  • System Rename of an SAP System which uses a tenant as database backend

11. July 2014

In case you are installing/migration into a SAP HANA Scale Out System please check the new note: 2041412

26. May 2014

In case you are migrating a big system, the output of ABAP report SMIGR_CREATE_DDL might also become quite big.

In these cases software provisioning manager can run into an Out Of Memory error when checking those files.

Please check oss-note: 2022386 – Migration nach HANA. Abbruch des Software Provisioning
Managers beim Ausführen der HdbSlLib: java.lang.OutOfMemoryError: Java heap
space
for further information.

19. March 2014

R3load Massloader error, new version of software provisioning manager disables Massloader functionality

The new versions of software provisioning manager 1.0 SP5 PL0 and software provisioning manager 1.0 SP4 PL8 disable the Massloader functionality by default using the R3load option “-loadprocedure dbsl“. To enable it again, you have to set the environment variable HDB_MASSIMPORT=YES before you start software provisioning manager. Before enabling the functionality again, make sure you have the latest R3load that contains the corresponding fix.

As long as there is no new SAPEXEDB archive available, you can manualy put the R3load into the current archive, as described in SAP Note 1826587 – SAP NetWeaver installation using SWPM 1.0:exchange R3load.

17. March 2014

R3load Massloader error, ABAP check report, new version of software provisioning manager

In addition to the R3load fix, there is now an ABAP report available that checks potential tables in the target system. The report is attached to SAP Note 1991638 – Supplementary SAP Note to SAP Note 1990894.

The SAP Note also contains a PDF document that describes how to check the log files of R3load to find out if the Massloader functionality was enabled during import. The document also describes how to exchange the current R3load executable with the new version from SAP Service Marketplace. We recommend to use the latest R3load version for any import.

With the next version of software provisioning manager SWPM 1.0 SP4 PL8, the Massloader functionality will be switched off by using the R3load parameter:

-loadprocedure dbsl

To enable the Massloader functionality again, in case you are using the latest R3load executable, set the environment variable HDB_MASSIMPORT=YES before you start software provisioning manager.

14. March 2014

Fix for R3load Massloader error

A fix is now available. Right now, it is only available as single R3load executable. You can download it from SAP Service Marketplace where you download SAP kernel patches. The patchtext of the fix is:

( 0.059) R3load/HANA correction data lost fields datatype decfloat (note 1990126)

For further information, see SAP Note 1680045 – Release Note for Software Provisioning Manager 1.0 SP04.

Unfortunately, setting the environment variable HDB_MASSIMPORT=NO doesn’t switch off the Massloader functionality in all cases. To switch it off completely, you have to add the argument -loadprocedure dbsl to the R3load call.

In case you need further assistance, open an OSS message.

12. March 2014

Error in R3load Massloader functionality

This only affects R3load executables when they are used for an import into an SAP HANA database.

Due to a problem inside of the R3load Massloader functionality, we strongly recommend to not use the functionality until a fixed version is available. To make sure the functionality is switched off, set the following environment variable before you start the software provisioning manager:

HDB_MASSIMPORT=NO

We also recommend not using any R3load version for SAP HANA database which was compiled after November 2013. To check the version, you can call ‘R3load -version‘. Check the entry “compile time“.

We will update this blog as soon as a fixed version is available.

12. February 2014

Import jobs for group ROWSTORE don’t get started

In case of an import into a non-scale out SAP HANA database, software provisioning manager runs into a state where 0 jobs are running, but a lot of jobs are waiting.

rowstore_does_not_start.JPG

The migmonctrl part of migmon regulary checks, if the jobs for the Column Store have finished – although, it does not check the “pre” job of a splitted table. In case a “pre” job fails, the dataload jobs don’t get started and remain in state “0”. Due to this, migmonctrl assumes that the Column Store related jobs have not yet finished.

In migmonctrl.log, this looks as follows:

Number of all jobs in process: 0

Column store package: REPOLOAD-2 has state: 0

Jobs for specific Column Store tables are still active. Will not start Rowstore group yet.

In this case, migmon would wait forever, so this is a bug. As a workaround, once you made sure you are facing exactly this issue, edit the order_by.txt file as follows:

# table group for rowstore tables

[ Rowstore ]

jobNum = 0

jobNum = 100

After that, migmon will start the group with the rowstore-related jobs. At the end, migmon will raise an error, because the “pre” job has failed. After solving the problem, you can hit the Retry button of the software provisioning manager.

10. February 2014

If the platform AIX is used please check the following OSS-note: 1976256 – AIX: TCP/IP connections hang

04. February 2014

In case the new statisticsserver of SAP HANA is activated, the installation fails with an error during the creation of role DBA_COCKPIT.

Logfile HdbCmdOut.log:

| com.sap.hdb.core.main.cmd.HdbCmdMain | 0 Unexpected result size. Expected 1

for sql query: SELECT ACTIVE_STATUS …

The new statisticsserver is part of the indexserver. Due to this, it doesn’t have its own service anymore and the check inside of software provisioning manager aborts in case there is no such service. This problem is planned to be fixed very soon (planned for software provisioning manager 1.0 SP4 PL5). Check the patch text that is available via the Info link besides the software provisioning manager download link. It should be:
SAP HANA, error during creation of role DBA_COCKPIT….

For more information about the new statisticsserver, see this short video:
Monitoring SAP HANA – New Statistics Server | SAP HANA

15. November 2013

The import terminates and the system issues the
following message: “Exchange directory must be an existing directory”.

This is corrected with SWPM 1.0 Support Package 4 Patch Level 1.

The problem occurs only when loading scale out systems for a parallel export / import.

24. October 2013

Performance problems can occur on Scale Out Systems during data import.

Fixed with SWPM 1.0 SP4 PL0 which is now available.

Please check SAP Note 1930853 – HdbUserStore contains
too many SAP HANA nodes

21. October 2013

Problem during system copy from SAP HANA to SAP HANA using R3load.

Please check SAP Note 1928043 – SAP HANA, homogeneous
system copy of a declustered system

16. October 2013

Release of SWPM 1.0 SP3 PL8

Patch Level 8 contains the following bugfixes related to SAP HANA.

  1. Long running R3szchk
    Depending on the size of the system, the runtime of R3szck can be very long.
    To fix this, there is now a dialog where you can choose between the size check variants ‘DD’ or ‘DB’:
    NewR3szchkDialog.JPG
    The second option uses the R3szchk option “-s DB”. In case you have very long runtimes for R3szchk, you might want to try this option.
  2. Error: Insufficient privilege during landscape reorg.

    Error executing database reorg: The following tables were not reorganized successfully: /BIC/B0001893000
    (Reason: insufficient privilege)

    The ABAP schema needs some addtional privileges. These are now added to the role USER.

  3. Migmonctrl puts column store tables into group rowstore
    With SAP kernel 7.40, migmonctrl checks the STR files to decide, if a table is put into group rowstore or not in file order_by.txt.
    In case the ABAP report SMIGR_CREATE_DDL has decided to change the location of the table from row to column store, the table was still in group rowstore in the file order_by.txt. With PL8, this has changed as migmonctrl now also checks the SQL files to find out in which store the table will be put.
To report this post you need to login first.

Be the first to leave a comment

You must be Logged on to comment or reply to a post.

Leave a Reply