IBM BPM, V8.0.1, All platforms > Migrating and upgrading your IBM BPM environment > Upgrading from IBM BPM V8.0 to IBM BPM V8.0.1 > Upgrading an ND environment > Upgrading clusters

Upgrading clusters with minimum downtime

Special precautions are required if minimum downtime is required during an upgrade in an ND environment.

Attention: The Installation Manager on the Microsoft Windows XP operating system does not recognize profile update errors. When you upgrade an installation on the Windows XP operating system, you must look at the "Identifying Profile Update Errors" section in the "Recovering after profile update errors" topic, even if Installation Manager ended with the following message: "The packages are updated".

These instructions assume the following sample setup; you might need to adapt them to your environment:


Procedure

Complete the following steps to upgrade your clusters.

  1. Stop the dmgr.

  2. Install the refresh pack onto the dmgr installation using either IBM Installation Manager or the silent installation method.
  3. Check for any errors, as explained in the "Identifying Profile Update Errors" section of the "Recovering from profile update errors" topic, before continuing.

    Attention: The Installation Manager on the Microsoft Windows XP operating system does not recognize profile update errors. When you upgrade an installation on the Windows XP operating system, you must look at the "Identifying Profile Update Errors" section in the "Recovering after profile update errors" topic, even if Installation Manager ended with the following message: "The packages are updated".

  4. Before you update the databases, ensure that you have applied Mandatory Interim Fix JR44669. See Required interim fix for APAR JR44669. To download the fix, see Required interim fixes for IBM BPM.

  5. Run the following command on the dmgr to generate the SQL scripts that are needed for updating the database schemas:

    • INSTALL_ROOT\bin\BPMGenerateUpgradeSchemaScripts.bat -upgrade source_version profile_name
    • INSTALL_ROOT/bin/BPMGenerateUpgradeSchemaScripts.sh -upgrade source_version profile_name

    where:

    • source_version is the version from which you are upgrading; for example, 8.0.0.0. This value must be specified as four digits separated by periods, with the third digit representing the fix pack level.
    • profile_name is the name of the dmgr profile.

    If you are prompted for information about your database setup, specify the relevant values for your environment. When prompted, enter Y to confirm that you want to generate the scripts.

    If you enter incorrect information, you can run the BPMGenerateUpgradeSchemaScripts command again before proceeding.

    When the BPMGenerateUpgradeSchemaScripts script is run from the dmgr profile, it will generate the SQL scripts that are needed to update all of the databases in the ND environment.

  6. Start the dmgr. This step is necessary because the 8.0.1 installer must connect to the central configuration repository on the dmgr to make the updates.
  7. Upgrade the clusters.

    Do not upgrade more than one remote installation at a time. The fix pack installer connects to the central configuration repository on the dmgr to make the updates. Updating multiple installations that host cluster members at the same time might cause conflicts in the central configuration repository and cause one or more of the profile updates to fail.

    For each cluster:

    1. Stop half of the cluster members and their node agents.

      For example, if you have two members, stop one.

    2. Install the refresh pack onto the corresponding installations that host cluster members using either IBM Installation Manager or the silent installation method.
    3. Check for any errors, as explained in the "Identifying Profile Update Errors" section of the "Recovering from profile update errors" topic, before continuing.

      Attention: The Installation Manager on the Microsoft Windows XP operating system does not recognize profile update errors. When you upgrade an installation on the Windows XP operating system, you must look at the "Identifying Profile Update Errors" section in the "Recovering after profile update errors" topic, even if Installation Manager ended with the following message: The packages are updated.

      Attention: Do not restart the servers at this point.

    4. Stop the remaining set of cluster members and their node agents. This short downtime cannot be avoided.

      Different versions of the cluster members cannot be up and running at the same time. After an upgraded member has been started you must no longer start members of the "old" version.

    5. Stop the dmgr.

    6. On the dmgr, change directories to INSTALL_ROOT/.

    7. Run the following command:

        • For English locales: bin\ws_ant.bat -f util\BPMProfileUpgrade.ant -profileName profile_name -Dupgrade=true –Dcluster= cluster_name

        • For non-English locales: bin\ws_ant.bat -f util\BPMProfileUpgrade.ant -profileName profile_name -Dupgrade=true "–Dcluster= cluster_name"

      • bin/ws_ant.sh -f util/BPMProfileUpgrade.ant -profileName profile_name -Dupgrade=true –Dcluster= cluster_name

      where profile_name is the name of the dmgr profile and cluster_name is the name of the cluster.

    8. Check the following log file for errors: profile_root/logs/BPMProfileUpgrade. profile_name. cluster_name. timestamp.log where profile_root is the root directory of the dmgr profile. If there is an error, fix the cause, then run the command again.

    9. If the Business Process Definition engine is configured on the cluster, update the Process Server database.

      If you are using DB2 on z/OS , you must manually alter a set of table spaces and upgrade the database schema before you upgrade the database data:

      1. To ensure that you can successfully run the SQL files for the DB2 for z/OS schema upgrade, alter the following table spaces to increase the buffer pool size to 8K:

        • WLPT36
        • WLPT44
        • WLPT64

        For information about altering table spaces, see

      2. From the INSTALL_ROOT/profiles/ profile_name/dbscripts/ProcessServer/DB2zOS/ database_name directory, copy the script named upgradeSchema800_ProcessServer.sql to your working directory.

        Review the script, and, if necessary, edit the file to replace the following symbolic variables with the actual values for the schema name, database name, storage group, and buffer pool for large objects and tables: @SCHEMA@, @DB_NAME@, @STOGRP@, @BPLOB4K@, @BPTABLE4K@, and @BPTABLE8K@. Then connect to the DB2 for z/OS database, and run the script against the database by using your preferred tool.

      3. From the INSTALL_ROOT/profiles/ profile_name/dbscripts/PerformanceDW/DB2zOS/ database_name directory, copy the script named upgradeSchema800_PerformanceDW.sql to your working directory.

        Review the script, and, if necessary, edit the file to replace the following symbolic variables with the actual values for the schema name, database name, storage group, and buffer pool for large objects and tables: @SCHEMA@, @DB_NAME@, @STOGRP@, @BPLOB4K@, @BPTABLE4K@, and @BPTABLE8K@. Then connect to the DB2 for z/OS database, and run the script against the database by using your preferred tool.

      4. Go to the [IBM_BPM_home]/BPM/Lombardi/tools/upgrade/UpgradeProcessData directory and set the database.is.db2zos property to true in the upgrade.properties file.

        For example: 

        database.is.db2zos=true

      5. Run the following command on the installation that hosts the dmgr:

        • BPM\Lombardi\tools\upgrade\UpgradeProcessData\upgradeProcessData.bat -profileName profile_name -nodeName node_name -serverName server_name
        • BPM/Lombardi/tools/upgrade/UpgradeProcessData/upgradeProcessData.sh -profileName profile_name -nodeName node_name -serverName server_name

        where profile_name is the name of the dmgr profile, node_name is the name of any cluster node, and server_name is the name of any member of the cluster on that node. For more information about the parameters that you can use with the upgradeProcessData command, see upgradeProcessData command-line utility.

      For all other database types on the installation that hosts the dmgr:

      • BPM\Lombardi\tools\upgrade\UpgradeProcessData\upgradeProcessData.bat -profileName profile_name -nodeName node_name -serverName server_name
      • BPM/Lombardi/tools/upgrade/UpgradeProcessData/upgradeProcessData.sh -profileName profile_name -nodeName node_name -serverName server_name

      where profile_name is the name of the dmgr profile, node_name is the name of any cluster node, and server_name is the name of any member of the cluster on that node. For more information about the parameters that you can use with the upgradeProcessData command, see upgradeProcessData command-line utility.

    10. If you have configured Business Process Choreographer, you must upgrade the BPEDB database manually to create the SCHEMA_STATUS table.
      1. Change to the directory where the Business Process Choreographer database scripts are generated. Run the following command on the installation that hosts the dmgr:

        • cd profile_root\dbscripts\ProcessChoreographer\ db_type\ db_name\ schema_name
        • cd profile_root/dbscripts/ProcessChoreographer/ db_type/ db_name/ schema_name

        where profile_root is the root directory of the dmgr profile, db_type is the database type, db_name is the database name, and schema_name is the (possibly empty) schema name.

      2. If the database server is remote and you do not have a local database client, copy the upgradeSchema_SchemaStatus.sql script to the database server so that you can run it there.

      3. Run the upgradeSchema_SchemaStatus.sql script to create the SCHEMA_STATUS table in the Business Process Choreographer database.

        If the table already exists, there will be an error that you can ignore.

    11. When the cluster has been successfully upgraded, restart the dmgr.
    12. Restart the node agents of those nodes where the refresh pack has already been applied and wait for the node synchronization to complete.
    13. Restart the cluster members of those nodes where the refresh pack has already been applied.

    14. Install the refresh pack onto the installations of the remaining cluster members using either IBM Installation Manager or the silent installation method.
    15. Check for any errors as explained in the "Identifying Profile Update Errors" section of Recovering from profile update errors before continuing.

      Attention: The Installation Manager on the Microsoft Windows XP operating system does not recognize profile update errors. When you upgrade an installation on the Windows XP operating system, you must look at the "Identifying profile update errors" section of Recovering from profile update errors, even if Installation Manager ended with the following message: "The packages are updated".

    16. Restart the node agents and cluster members after the refresh pack has been installed.

Upgrading clusters


Related concepts:
Topologies of a ND environment


Related tasks:
Executing SQL upgrade scripts


Related reference:
Recovering from profile update errors
BPMGenerateUpgradeSchemaScripts command-line utility