3 Installing Oracle Ultra Search

Hardware Requirements for Oracle Ultra Search

Oracle Ultra Search hardware requirements are based on the quantity of data that you plan to process using Oracle Ultra Search. Oracle Ultra Search uses Oracle Text as its indexing engine and the Oracle Database as its repository.

Sufficient RAM Along with the resource requirements for the database and the Text indexing engine, also consider the memory requirements of the Oracle Ultra Search crawler. The Oracle Ultra Search crawler is a pure Java program. When the crawler is launched, the Java Virtual Machine (JVM) is configured to start with 25MB and grow to 256MB. When crawling very large amounts of data, these values might need to be adjusted.

The Oracle Ultra Search administration tool is a J2EE 1.2 standard Web application. It can be installed and run on a separate host from the Oracle Ultra Search backend. You might want to install and run this on the same host as the Oracle Ultra Search backend. Regardless of your choice, allocate enough memory for the J2EE engine. Oracle recommends using the Oracle HTTP Server with the Oracle Application Server Containers for J2EE (OC4J). Allocate enough memory for the HTTP Server as well as for the Java Development Kit (JDK) that runs the J2EE engine.

Sufficient Disk Space Because customer requirements vary widely, Oracle cannot recommend a specific amount of disk space. As a general guideline, the minimum requirements are as follows:

• Approximately 3GB of disk space for the Oracle Application Server Infrastructure or database and the Oracle Ultra Search backend.

• 15MB of disk space for the Oracle Ultra Search middle tier on top of the Web server's disk requirements.

• For each remote crawler host, the same amount of disk space as needed to install the Oracle Ultra Search backend.

• Disk space for a large TEMPORARY tablespace. Create a TEMPORARY tablespace as large as possible, depending on the amount of RAM on the host.

• Disk space for the Oracle Ultra Search instance user's tablespace.

  • The Oracle Ultra Search instance user is a database user that you must explicitly create. All data that is collected and processed as part of crawling and indexing is stored in this user's schema.

  • You should create the tablespace as large as the total amount of data that you want to index. For example, if you estimate that the total amount of data to be crawled and indexed is 10GB, then create a tablespace that is at least 10GB for the Oracle Ultra Search instance user. Make sure to assign that tablespace as the default tablespace of the Oracle Ultra Search instance user.

Software Requirements for Oracle Ultra Search

The Oracle Ultra Search middle tier components are Web applications. Therefore, they require a Web server to run. Oracle recommends using Oracle Application Server.

Installing the Middle Tier with the Oracle Database

The Oracle Ultra Search middle tier is installed as part of the Oracle Database Server installation, which is accomplished using the Oracle Universal Installer.

Start the Oracle Ultra Search Middle Tier

For the Oracle Database release, use the following command to start the Oracle Ultra Search middle tier. You must run this command manually to bring up the Oracle Ultra Search middle tier after installation.

$ORACLE_HOME/bin/searchctl start

For the Oracle Application Server release, use Oracle Process Manager and Notification Server (OPMN) to start the OC4J_Portal instance. For example:

$ORACLE_HOME/opmn/bin/opmnctl startproc instancename=OC4J_Portal

Unlock WK_TEST

The Oracle Ultra Search installer creates a default Oracle Ultra Search instance based on the default Oracle Ultra Search test user. You can test Oracle Ultra Search functionality based on the default instance after installation.

The default instance name is WK_INST. It is created based on the database user WK_TEST. The default user password is WK_TEST.

For security purposes, WK_TEST is locked after the installation. The administrator should log on to the database as DBA role, unlock the WK_TEST user account, and set the password to be WK_TEST. The password expires after the installation. If the password is changed to anything other than WK_TEST, then you must also update the cached schema password using administration tool Edit Instance page after you change the password in the database.

The default instance is also used by the Oracle Ultra Search query application. Make sure to update the data-sources.xml file.

Enable the Oracle Ultra Search Query Applications

The data-sources.xml file is located in the $ORACLE_HOME/oc4j/j2ee/OC4J_SEARCH/config directory. Under tag <data-sources> add the following:

<data-source 
   class="oracle.jdbc.pool.OracleConnectionCacheImpl" 
   name="UltraSearchDS" 
   location="jdbc/UltraSearchPooledDS" 
   username="username" 
   password="password" 
   url="jdbc:oracle:thin:@database_host:oracle_port:oracle_sid"
/> 

In the preceding syntax, the following variables were used:

• username and password are the Oracle Ultra Search instance owner's database user name and password.

• database_host is the host name of the back end database computer.

• oracle_port is the port to the user's Oracle Database.

• oracle_sid is the SID of the user's Oracle Database.

In addition to user name, password, and JDBC URL, data-sources.xml allows configuration of the connection cache size, and the cache scheme.

The following tag specifies the minimum and maximum limits of the cache size, the inactivity time out interval, and the cache scheme.

<data-source 
   class="oracle.jdbc.pool.OracleConnectionCacheImpl" 
   name="UltraSearchDS" 
   location="jdbc/UltraSearchPooledDS" 
   username="wk_test" 
   password="wk_test" 
   url="jdbc:oracle:thin:@localhost:1521:isearch" 
   min-connections="3" 
   max-connections="30" 
   inactivity-timeout="30">
 <property name="cacheScheme"  value="1"/>
</data-source>

If you are adding the data source for the default Oracle Ultra Search instance user WK_TEST, then make sure to unlock WK_TEST first.

There are three values for the caching schemes:

• 1 = DYNAMIC_SCHEME

• 2 = FIXED_WAIT_SCHEME

• 3 = FIXED_RETURN_NULL_SCHEME

Restart the Oracle Ultra Search Middle Tier

For the database release, stop and restart the Oracle Ultra Search middle tier. For example:

For the Oracle Application Server release, use Oracle Process Manager and Notification Server (OPMN) to start the OC4J_Portal instance. For example:

$ORACLE_HOME/opmn/bin/opmnctl startproc instancename=OC4J_Portal

Reconfigure the Oracle Ultra Search Backend for the Database Character Set

You must reconfigure the Oracle Ultra Search backend to adapt to the new character set if one of the following has occurred:

• You choose a database character set (for example, UTF8) other than the default WE8ISO8859P1 character set during the Oracle Ultra Search installation.

• The database character set is changed after installation.

Two SQL scripts (wk0prefcheck.sql and wk0idxcheck.sql), located in $ORACLE_HOME/ultrasearch/admin/, are used for reconfiguration:

• wk0prefcheck.sql is run by the wksys user to reconfigure default cache character set and index preferences.

• wk0idxcheck.sql is needed for reconfiguring instances created before the database character set change (for example, the default instance). This script must be run by the instance owner, and wk0prefcheck.sql must be run first because it depends on reconfigured default settings generated by wk0prefcheck.sql.

  Running wk0idxcheck.sql also drops and recreates the Oracle Text index used by Oracle Ultra Search. If there are already data sources indexed, then you must force a recrawl of all of the data sources.

  wk0idxcheck.sql must be run once for each instance. For example, if there are two instances, inst1 and inst2, owned by owner1 and owner2, respectively, then wk0idxcheck.sql should be run twice, once by owner1 and once by owner2.

Testing the Oracle Ultra Search Administration Tool

If you log on to the Oracle Ultra Search administration tool successfully, then you have completed the Oracle Ultra Search administration tool configuration process. Do the following to check the Oracle Ultra Search Administration Tool:

1. Check that the Web Server is running.

2. Attempt to log on to the administration tool:

   1. Visit the following URL

      http://hostname.domainname:port/ultrasearch/admin/index.jsp

      In the preceding URL, hostname.domainname is the full name of the host where you have installed the Oracle Ultra Search middle tier, and port is the default Web server port.

   2. Log on to the Oracle Ultra Search administration tool by entering the Oracle Ultra Search instance owner's database user name and password.During the installation of the Oracle Ultra Search backend, a new Ultra Search instance owner, WK_TEST is created.

   The first time any JSP page is accessed, it takes a few seconds to compile. Subsequent accesses are much faster.

Testing the Oracle Ultra Search Query Applications

After you verify that the Oracle Ultra Search administration tool is working, you should be able to run the Oracle Ultra Search query applications.

To test the Oracle Ultra Search query applications, do one of the following:

• Visit the following URL:

  http://hostname.domainname:port/ultrasearch/query/search.jsp

• Follow the links in the Oracle Ultra Search welcome page: http://hostname.domainname:port/ultrasearch/index.html

Locations for query applications are listed in the following section. Access the query source code by going to the directories list. You can also see a working demonstration of each query JSP page with the URL root, and you can append the correct JSP file name at the end of the URL root.

The query application is shipped as $ORACLE_HOME/ultrasearch/ultrasearch_query.ear.

Portlet is shipped as $ORACLE_HOME/ultrasearch/webapp/ultrasearch_portlet.ear.

Installing the Backend on Remote Crawler Hosts

The Oracle Ultra Search remote crawler is part of the Oracle Ultra Search backend. The crawler installation procedure is the same as installing the Oracle Ultra Search backend.

On each remote crawler host, the Oracle Ultra Search backend is installed under a common directory known as ORACLE_HOME. You should have been prompted by the Oracle Universal Installer to enter this directory. The remote ORACLE_HOME directory is referred to as $REMOTE_ORACLE_HOME.

If you choose not to install the Oracle HTTP Server during the Oracle Application Server installation, then you must perform the following steps manually for remote crawling:

• Locate the file that defines the environment.

  • On UNIX, $REMOTE_ORACLE_HOME/ultrasearch/tools/remotecrawler/scripts/unix/define_env

  • On Windows, $REMOTE_ORACLE_HOME/ultrasearch/tools/remotecrawler/scripts/winnt/define_env.bat

• Replace %ORACLE_HOME% with the value of the REMOTE_ORACLE_HOME environment variable.

• Replace %s_jreLocation% with the directory path of a Java runtime environment (JRE) version 1.2.2 and higher. You should specify the root directory of the JRE.

• Replace %s_jreJDBCclassfile% with the full path and file name of the Oracle JDBC Thin driver version 12.

Configuring the Remote Crawler

The remote crawler requires a communication channel between the backend database and the remote crawler host.

The mechanisms of communication are RMI and JDBC. Configuration of the remote crawler differs depending on which mechanism you use. The JDBC-based mechanism requires you to supply a database user (or role) during the registration process.

The registration process is done by running a SQL script on the Oracle Ultra Search remote crawler host. The SQL script connects over SQL*Plus to the Oracle backend database and registers the remote crawler host.

1. Locate the correct ORACLE_HOME.

   The Oracle Ultra Search middle tier is installed under a common directory known as ORACLE_HOME. If you have installed other Oracle products prior to the Oracle Ultra Search middle tier, then you could have multiple ORACLE_HOME directories on your host. The registration script requires that you enter the ORACLE_HOME directory in which the Oracle Ultra Search middle tier is installed.

2. Locate the WKSYS super-user password.

   You must run the registration script as the WKSYS super-user or as a database user that has been granted super-user privileges.

3. Start SQL*Plus.

   Be sure to run the correct version of SQL*Plus, because multiple versions can reside on the same host if you have previously installed some Oracle products. On UNIX platforms, make sure that the correct values for PATH, ORACLE_HOME and TNS_ADMIN variables are set. On Windows platforms, choose the correct menu item from the Start menu.

   After you have identified how to run the correct SQL*Plus client, you must log on to the Oracle Ultra Search database. To do this, you might need to configure an Oracle Net service setting for the Oracle Ultra Search database.

4. After SQL*Plus is running, log on to the database using the schema and password that you located in Step 2.

5. Run the registration script.

   Start up SQL*Plus as the WKSYS super-user and enter the following:

   @full_path_of_registration_script
   

   • The registration script for RMI-based remote crawling is the following:

     $REMOTE_ORACLE_HOME/ultrasearch/tools/remotecrawler/scripts/<platform>/register.sql
     
     

     For example, if the value for $REMOTE_ORACLE_HOME on a UNIX host is /home/oracle10g, then enter the following at the SQL*Plus prompt to register an RMI-based remote crawler:

     @/home/oracle10g/ultrasearch/tools/remotecrawler/scripts/unix/register.sql
     
     

     The RMI-based registration script prompts you for three variables:

     • RMI_HOSTNAME: The remote hostname. This is where the RMI registry/daemon will run.

     • RMI_REGISTRY_PORT: The port that the RMI registry is listening on.

     • ORACLE_HOME: The Oracle home located in Step 1.

       For example, /u01/oracle10g on a UNIX host or d:/u01/oracle10g on a Windows host. Remember to use forward slashes for Windows hosts.

   • The registration script for JDBC-based remote crawling is the following:

     $REMOTE_ORACLE_HOME/ultrasearch/tools/remotecrawler/scripts/<platform>/register_jdbc.sql
     
     

     For example, if you are running SQL*Plus on Windows, and $REMOTE_ORACLE_HOME is in d:\Oracle\Oracle10g, then enter the following at the SQL*Plus prompt to register a JDBC-based remote crawler:

     @d:\Oracle\Oracle10g\ultrasearch\tools\remotecrawler\scripts\winnt\register_jdbc.sql
     
     

     The JDBC-based registration script prompts for three variables:

     • LAUNCHER_NAME: An arbitrary string used to identify a JDBC-based remote crawler launcher, which is needed when you start up the JDBC-based remote crawler launcher.

     • CONNECTUSER: The database user (or role) that the JDBC-based remote crawler launcher will use to establish a database connection and listen for launch events.

     • ORACLE_HOME: The Oracle home located in Step 1.

   The registration script invokes the wk_crw.register_remote_crawler PL/SQL API. The REMOTE_CRAWLER_HOSTNAME and ORACLE_HOME variables are used to compose arguments for the wk_crw.register_remote_crawler API. You may optionally choose to call this API, especially if you need to register multiple remote crawlers programatically.

6. Verify and complete the remote crawler profile configuration. Be sure to enter the correct values for both variables. To verify that the registration has completed correctly, do the following:

   1. Log on to the Oracle Ultra Search administration tool.

   2. Click the Remote Crawler Profiles tab in the Crawler tab. You should see the remote crawler launcher you registered in the remote crawler profile list.

      • For RMI-based remote crawlers, you will see the host:port combination that uniquely identifies the RMI-subsystem.

      • For JDBC-based remote crawlers, you will see the Launcher name.

   3. Click Edit to complete the configuration process for the remote crawler profile.

Unregistering a Remote Crawler

If you enter any wrong values for the register.sql script, then you should unregister the remote crawler using the unregister.sql script. Run the unregister script the same way as you ran the registration script. The unregister.sql script calls the wk_crw.unregister_remote_crawler PL/SQL API. After you have successfully unregistered the remote crawler, you can rerun the register.sql script.

Post-Upgrade Configuration Steps

After upgrading to the current release, follow these post-upgrade configuration steps:

1. Set the ORACLE_HOME and ORACLE_SID environment variables to Oracle Database 10g.

2. Change directories to ORACLE_HOME/ultrasearch/admin/.

3. Run the following statement:

   sqlplus "sys/password as sysdba"
   
   

4. Run the following statement:

   @wk0config.sql WKSYSPW JDBC_CONNSTR LAUNCH_ANYWHERE NET_SERVICE_NAME
   
   

   In the preceding statement, the following parameters were used:

   • WKSYSPW is the password for the WKSYS schema.

   • JDBC_CONNSTR is the JDBC connection string. Use the format hostname:port:sid, such as machine1:1521:iasdb, if the database is not in the Oracle Real Application Clusters environment.

   If the database is in a Oracle Real Application Cluster environment, then use the TNS keyword-value format instead, because it allows connection to any node of the system:

   (DESCRIPTION=(LOAD_BALANCE=yes) 
    (ADDRESS_LIST= 
       (ADDRESS=(PROTOCOL=TCP)(HOST=cls02a)(PORT=3001)) 
       (ADDRESS=(PROTOCOL=TCP)(HOST=cls02b)(PORT=3001))) 
    (CONNECT_DATA=(SERVICE_NAME=sales.us.acme.com))) 
   
   

   In the preceding syntax, the following parameters were used:

   • LAUNCH_ANYWHERE is the mode of the database. Setting it to TRUE indicates that the database is in Oracle Real Application Cluster mode; FALSE indicates that the database is not in Oracle Real Application Cluster mode.

   • NET_SERVICE_NAME is the network service name used by wk0config.sql to establish the database connection. Setting it to ""(empty string) while running wk0config.sql from the database host eliminates the need to specify the network service name.

@wk0config.sql welcome1 machine:1521:iasdb FALSE""

@wk0config.sql welcome1
"(DESCRIPTION=(LOAD_BALANCE=yes) 
   (ADDRESS_LIST= 
      (ADDRESS=(PROTOCOL=TCP)(HOST=cls02a)(PORT=3001)) 
      (ADDRESS=(PROTOCOL=TCP)(HOST=cls02b)(PORT=3001))) 
   (CONNECT_DATA=(SERVICE_NAME=sales.us.acme.com)))" FALSE ""