Preparing to install on a Unicode database

BMC supports BMC Helix Service Management deployment only with Unicode and case insensitive databases. Before you install BMC Helix Service Management, make sure that your database is configured to use Unicode.

Before you begin

Each database client library has special mechanisms for specifying the codeset in which database clients attempt to communicate with the AR System server. If these mechanisms specify a codeset that is inconsistent with the codeset that the AR System server and upgrade programs require, errors and data corruption can occur.Perform the following procedure to avoid this problem.

To prepare your host computer for a Unicode BMC Helix Service Management installation or upgrade

• (For Linux) Set the LANG environment variable for the locale you want to use:
  
  • Make sure that you have installed the UTF-8 locales in which you plan to run AR System programs.
  • Make sure that you use the correct spelling and capitalization for your particular system.
    To find the locales that correspond to the language you want to use, use the locale -a command.
  • To set the locale of the installation, the server installation script uses the locale of the shell from where it is run. 
  • During installation, the arsystem script is set up with the correct values for the LANG variable. The arsystem script launches armonitor, which launches the programs mentioned in the armonitor.conf file; each of these programs inherit the environment variables established in the arsystem script.

• (For Windows) If you are installing an Oracle Unicode data base on a Microsoft Windows operating system, set the value of the NLS_LANG in environment variables.

  1. Go to Start > My Computer/Computer.
  2. Right-click anywhere on the My Computer window and select Properties > Advanced system settings > Environment variables.
  3. Under System variables, click New and enter the values for the following variables:
     • Variable name (NLS_LANG)
     • Variable value (<LANGUAGE>_<TERRITORY>.<NLS_CHARACTERSET>)  
  4. Reboot the machine after creating the environment variable.

  Oracle defines the NLS_LANG value as LANGUAGE_TERRITORY.CHARACTERSET, for example, AMERICAN_AMERICA.AL32UTF8.
  The CHARACTERSET value AL32UTF8 tells the Oracle client library to send and receive character data as UTF-8. (Do not use the CHARACTERSET value UTF8, which is obsolete.) 

  1. Verify that the NLS_LANG environment variable is set correctly.

  2. Set the NLS_LENGTH_SEMANTICS=BYTE initialization parameter on the Oracle database instance.
     Although the AR System server enables you to request that character fields be measured in characters, the server still communicates with the database in bytes.
     For more information, see Oracle Metalink Note 144808.1, "Examples and limits of BYTE and CHAR semantics usage."

     Warning

     AR System does not support the CHAR setting. If your database administrator changes the NLS_LENGTH_SEMANTICS value from BYTE to CHAR when reconfiguring NLS_INSTANCE_PARAMETERS , the data in the AR System server gets corrupted. The NLS_LENGTH_SEMANTICS parameter must be set to BYTE in NLS_INSTANCE_PARAMETERS when the Oracle database server is started.

To upgrade a server with a Unicode database

1. Back up the database components, objects, forms, and data.
2. (On Windows systems) Perform the upgrade in the same locale in which you installed the original AR System Server.
   For example, if the original server has a Japanese locale, perform the upgrade in the same Japanese locale.
   (On Linux systems) Perform the upgrade in the Unicode version of the locale in which you ran the original AR System Server.
   For example, if the server has the ja_JP.eucJP locale, perform the upgrade in the ja_JP.UTF-8 locale. Locale names vary across UNIX variants and versions.