Install Databridge Client on UNIX

To install the Client on UNIX

1. Log on using a userid that has root privileges.

2. If a previous version of Databridge Client is running on your system, stop it.

3. Create a new install directory for Databridge 6.6 (for example, " /opt/dbridge66").

4. Move the appropriate tar files (Client, Client Console) to the UNIX system using binary FTP. See Description of Files on the Installation Image.

5. Change the directory to the install directory you created in step 3 and extract the tar file:

   tar -xvf 
        filename
       

   where filename is the full tar filename, including location.

   The extract program creates a client subdirectory named after the database (for example, Oracle). Files are extracted to this subdirectory and to the directory you specified.

6. Next, change the owner and group for the files in the install directory to the userid (and the group to which the userid belongs) designated to run Databridge Client (that is, the USERID specified in the file globalprofile.ini). To do this, type the following command:

   chown -R dbridge: users *

   where dbridge is the user ID and users is the corresponding group.

   CAUTION:To prevent the files in the install directory from being accidentally deleted, we recommend that you leave them set to read-only (the default). Before you update the software, to prevent the extraction from failing, either remove all of the files from the install directory or make them writeable until the update is successfully installed.

7. In the editor, open the sample daemon script file ( dbdaemon) located on the root of the install directory, update the necessary environment variables, and then save dbdaemon to the following location:

   • On Linux and Solaris: /etc/init.d

   • On HP-UX: sbin/init.d

8. Create the directory /etc/Attachmate/DATABridge/6.6

9. In the editor, open the globalprofile.smp file on the root of the install directory and make the following changes:

   WARNING:This file is critical for client operations. If this file is missing or contains the wrong information, the client will not run.

   1. For INSTALLDIR, specify the install directory you created in step 3.

   2. For WORKINGDIR, specify the full name of the Client's working directory for 6.6. If you're upgrading from an older version than 6.2, make sure that the working directory for 6.6 is different from the older working directory. You create the actual directory later in this procedure.

   3. For USERID, specify the user under which you'll run the daemon. When you start the daemon, you must log in as this user or as the root user. Clients can only be run under this userid.

   4. To enable file security for the Client working directory, specify a value for umask. The bits specified get removed from the default file security bits. A umask of 027 (the default) indicates that the owner bits are unchanged, but the group's w and x bits are reset and all 3 bits are reset for other users. For a stronger mask, specify a value such as 077.

10. Save the file as globalprofile.ini to the directory /etc/Attachmate/DATABridge/6.6

11. Do one of the following:

    • (New installations) In the home directory, use the editor to update your profile (for example, ".profile" or ".bash_profile") and include the environment variable ORACLE_HOME. Consult the Oracle database administrator to determine what this variable should be set to (typically, it's "/opt/oracle..."). You must also include the directory $ORACLE_HOME/bin in the PATH to ensure that the bulk loader sqlldr can be located. You may want to also include the environment variable INSTALLDIR, which points to the directory created in step 3. You can then add the Client program directory $INSTALLDIR/Oracle to the PATH, making it a lot easier to run the Client from the command line.

    • (Upgrades - Optional) Update your profile so that the PATH points to the newly-installed Client.

12. For Databridge Client to find the Oracle shared libraries, add the Oracle lib (or lib32) directory to the appropriate environment variable for your system, as shown in the following table. The ORACLE_HOME environment variable is used instead of typing the complete Oracle lib directory name.

    To update	Add the following to your profile
    64-bit clients except on AIX 6.1 or newer	LD_LIBRARY_PATH = $LD_LIBRARY_PATH:$ORACLE_HOME/lib
    64-bit clients on AIX 6.1 or newer	LIBPATH = LIBPATH:$ORACLE_HOME/lib
    32-bit clients on Solaris or Linux	LD_LIBRARY_PATH = $LD_LIBRARY_PATH:$ORACLE_HOME/lib32
    32-bit clients on AIX	LIBPATH = LIBPATH:$ORACLE_HOME/lib32
    32-bit clients on HP-UX	SHLIB_PATH = SHLIB_PATH:$ORACLE_HOME/lib32
    NOTE:The UNIX shell you use may require that you add a line that exports the environment variable. For instructions, consult your UNIX documentation.

13. Using the value you specified in step 9b, create a Client working directory for 6.6 that contains the three subdirectories config, locks and logs. (For upgrades, the Migrate program creates a new, secure working directory and subdirectories from this value. See Run the Migrate Program on UNIX.)

14. When installation is complete, verify that the Client directory files are correctly installed. For a list of installed files, see Files Installed with Databridge Client and Client Console.

15. If you run the Client from a command line, you must also update the environment variable PATH in your profile to include the directory that contains Databridge Client. If you installed the Client to the directory /opt/dbridge66, add “ :/opt/dbridge66/Oracle” to the PATH. The PATH is defined as:

    PATH=$PATH: … $ORACLE_HOME/bin:  …

    where each specified directory is terminated by a colon, except for the last one on the line.

    Individual clients use the database names as the directories names (for example, Oracle).

    CAUTION:Make sure that the mknod utility’s directory (usually /usr/sbin) is included in the PATH. If the shell scripts used to clone data sets cannot find this utility, files will be used in place of named pipes resulting in bulk loader problems, such as broken pipes.

16. Do one of the following:

    • If you're upgrading, proceed to Run the Migrate Program on UNIX to complete the installation.

    • If you're installing for the first time, install the Client Console to configure the Client parameters. See Install the Client Console on UNIX.

    • The "auto_reclone" parameter has been deprecated, as this feature did not work correctly with online garbage collection reorgs. The two available options are to either stop the client when a garbage collection is detected, or, let the client deselect the affected data sets and reclone them later when all the reorgs are completed. The -y option for the process command makes it easy to reclone these data sets without having to do anything else.

    • • The " max_wait_secs" parameter now accepts two values, the second is optional and defaults to 0. When the second parameter is non-zero, it enables code that breaks up long wait-and-retry loops, implemented by the engine alone, into smaller wait-and-retry loops in the engine. The client repeatedly issues DBWait remote procedure calls until the period of inactivity exceeds the value specified by the first value. This allows you to break up a one hour loop into 60 one minute loops, which ensures that the line does not go idle for long periods of time.

    • The index suffixes for the Oracle client now can use the environment variable $(INDEX_NAME) which gets replaced by the actual index name when the suffix is applied. See the Databridge Client Administrator’s Guide section on SQL suffixes in Appendix C: Client Configuration.

    • The client implements row filtering for secondary tables derived from items (or GROUPS) that have OCCURS clauses that are not flattened. This is done by using a text file to specify the filtering criteria for such tables using a SQL like syntax.

    • The client implements the flattening of unsigned NUMBER items and ALPHA items with OCCURS clauses to a CHAR of VARCHAR column instead of making each occurrence of the item into a separate column. This process is referred to as flattening the OCCURS to a string. You can have fixed format strings and CSV format strings where the delimiter can be selected.

    • The DMSII time handling code now supports NUMBER(12) items that represents time in the format "HHMMSSmmmmmm", where the last 6 digits are the fractional part of the seconds.

    • The SQL Server client can now handle table names that are TRANSACT-SQL reserved words by enclosing them in square brackets in all SQL statements it uses. In order to maintain backward compatibility we implemented the configuration parameter " bracket_tabnames" that enables this feature.

    • The -k option was added to force the Client to drop tables rather than run cleanup scripts in a multi-source environments when recloning datasets. This is designed to be used after a DMSII reorganization that requires certain data sets to be recloned. Using the -k option for the first such clones makes the process a lot simpler, as manually dropping the tables to get them recreated with the new layouts are no longer required.