Installation on Cloudera

Analytic Server 3.2.2

Online installation

1. Navigate to the IBM Passport Advantage® Web Site and download the self-extracting binary file specific to your stack, stack version, and hardware architecture to a host within the Cloudera cluster. The available Cloudera binaries are:

   Table 1. Analytic Server self-extracting binary files

   Description	Binary filename
   IBM SPSS Analytic Server 3.2.2 for Cloudera 5.11, 5.12, 5.13, 5.14, 5.15, 6.0, 6.1, 6.2, and 6.3 Ubuntu English	spss_as-3.2.2.0-cdh5.11-6.3-ubun.bin
   IBM SPSS Analytic Server 3.2.2 for Cloudera 5.11, 5.12, 5.13, 5.14, 5.15, 6.0, 6.1, 6.2, and 6.3 Linux x86-64 English	spss_as-3.2.2.0-cdh5.11-6.3-lx86.bin

2. Run the Cloudera self-extracting *.bin installer on the Cloudera Manager primary cluster node. Follow the installation prompts by accepting the license agreement and keeping the default CSD installation directory.
   Note: You must specify a different CSD directory if it is altered from the default location.
3. Use the following command to restart Cloudera Manager after the installation is complete:

   service cloudera-scm-server restart

4. Open the Cloudera Manager interface (for example, http://${CM_HOST}:7180/cmf/login with the default login credentials of admin/admin), refresh the Remote Parcel Repository URLs (located in Host > Parcels > click Configuration), and verify that the URL is correct. For example:

   http://ibm-open-platform.ibm.com

   Note: The Parcel Update Frequency and Remote Parcel Repository URLs can be updated to suit your specific needs.
5. After Cloudera Manager refreshes the parcel files (you can manually refresh the parcel files by clicking Check for New Parcels), you will see that the AnalyticServer parcel status is set to Available Remotely.
6. Select Download > Distribute > Activate. The AnalyticServer parcel status is updated to Distributed, Activated.
7. In Cloudera Manager, add Analytic Server as a service and decide where to place the Analytic Server. You need to provide the following information in the Add Service Wizard:
   Note: The Add Service Wizard shows the overall progress during each phase of the service creation process, and provides a final confirmation message when the service is successfully installed and configured on the cluster.
   • Analytic Server metastore host name
   • Analytic Server metastore database name
   • Analytic Server metastore user name
   • Analytic Server metastore password

   MySQL as the Analytic Server metadata repository

   • Analytic Server metastore driver class: com.mysql.jdbc.Driver
   • Analytic Server metastore repository URL: jdbc:mysql://${MySQL_DB}/{DBName}?createDatabaseIfNotExist=true

     {MySQL_DB} is the hostname of the server where MySQL is installed

   Db2 as the Analytic Server metadata repository

   • Analytic Server metastore driver class: com.ibm.db2.jcc.DB2Driver
   • Analytic Server metastore repository URL: jdbc:db2://{Db2_HOST}:{PORT}/{DBName}:currentSchema={SchemaName};

     {Db2_HOST} is the hostname of the server where Db2 is installed.

     {PORT} is the port on which Db2 is listening.

     {SchemaName} is an available, unused schema.

     Work with your Db2 administrator if you are unsure of what values to enter.

   LDAP configuration

   Analytic Server uses an LDAP server to store and authenticate users and groups. You provide the required LDAP configuration information during Analytic Server installation.

   Table 2. LDAP configuration settings

   LDAP setting	Description
   as.ldap.type	LDAP type. The value can be ads, ad, or openladp. ads - Apache Directory Server (default setting) ad - Microsoft Active Directory openladp - OpenLDAP
   as.ldap.host	LDAP host
   as.ldap.port	LDAP port number
   as.ldap.binddn	LDAP bind DN
   as.ldap.bindpassword	LDAP bind DN password
   as.ldap.basedn	LDAP base DN
   as.ldap.filter	LDAP user and group filter rule Note: When this value contains vertical bar | characters, the characters must be escaped with backslash characters (for example, \|).
   as.ldap.ssl.enabled	Specifies whether to use SSL to communicate between Analytic Server and LDAP. The value can be true or false.
   as.ldap.ssl.reference	LDAP SSL reference ID
   as.ldap.ssl.content	LDAP SSL configuration

   • By default, as.ldap.type is set to ads and the other related settings contain default settings. The exception is you must provide a password for the as.ldap.bindpassword setting. Analytic Server uses the configuration settings to install an Apache Directory Server (ADS) and run the server initialization. The default ADS profile includes the user admin with a password of admin. You can conduct user management through the Analytic Server Console or import user and group information from an XML file via the importUser.sh script that is located in the <Analytic Root>/bin folder.
   • If you plan to use an external LDAP server, such as Microsoft Active Directory or OpenLDAP, you must define the configuration settings according to the actual LDAP values. For more information, see Configuring LDAP user registries in Liberty.
   • You can change the LDAP configuration after Analytic Server is installed (for example, changing from Apache Directory Server to OpenLDAP). However, if you initially start with Microsoft Active Directory or OpenLDAP, and decide to later switch to Apache Directory Server, Analytic Server will not install an Apache Directory Server during installation. The Apache Directory Server is only installed when it is selected during the initial Analytic Server installation.

   

8. When installing Analytic Server in a Kerberos enabled Cloudera environment, the following settings must also be configured in the Add Service Wizard:
   Note: Analytic Server inspects the HDFS configuration for Kerberos related values to use for authentication.
   • Select Kerberos as the Analytic Server security setting if you want to enable Kerberos authentication when logging into the Analytic Server console. When Kerberos is selected as the Analytic Server security setting, the Analytic Server console defaults to the Kerberos login mode.
   • Select Kerberos as the Analytic Server database data source connection method setting when you want to connect to Kerberos enabled databases. When Kerberos is selected as the Analytic Server database data source connection method setting, the Analytic Server console uses Kerberos mode when connecting to a database
   • The Kerberos Realm Name and KDC host settings are required. The Kerberos Realm Name (as.kdc.realms) and KDC host (kdcserver) values are located in the krb5.conf file on the Kerberos Key Distribution Center (KDC) server.

     Multiple realm names are supported when they are separated by comma characters. The specified Kerberos realm names correspond to, and are associated with, user names. For example the user names UserOne@us.ibm.com and UserTwo@eu.ibm.com would correspond with the realms us.ibm.com,eu.ibm.com.

     Kerberos cross-realm trusts must be configured when more than one realm is specified as a Kerberos Realm Name. The user name that is entered during the Analytic Server console login prompt is entered without the realm name suffix. As a result, when multiple-realms are specified, users are presented with a Realms drop-down list that allows them to select the realm.

     Note: When only one realm is specified, users are not presented with a Realms drop-down list when signing into Analytic Server.

     

     Notes:

     • The Analytic Server security and Analytic Server database data source connection method settings are applicable to IBM SPSS Modeler client and Analytic Server console authentication.
     • When Analytic Server database data source connection method is set to Kerberos, you must ensure that the target databases are also Kerberos enabled.
     • The Analytic Server security and Analytic Server database data source connection method settings do not configure Kerberos authentication on the Hadoop cluster. For more information, see the "Enabling Kerberos impersonation" section.
     • If you want Kerberos authentication to be enabled at login, you must deploy the IBM SPSS Modeler client as a valid Kerberos client. This is accomplished by using the addprinc command in the Kerberos Key Distribution Center (KDC) server. For more information, refer to your IBM SPSS Modeler documentation.

   When installing Analytic Server in a Kerberos enabled Cloudera environment you must also create the required accounts in Kerberos and enable Kerberos impersonation. For more information, see Configuring Kerberos.

Offline installation

The offline installation mode is the recommended method.

The offline installation steps are the same as the online steps except you must manually download the parcel files and metadata that are appropriate for your particular operating system.

1. Download and run the Cloudera self-extracting *.bin installer on the Cloudera Manager primary cluster node. Follow the installation prompts by accepting the license agreement and keeping the default CSD installation directory.
   Note: You must specify a different CSD directory if it differs from the default location.
2. Copy the required parcel and metadata files to your local Cloudera repo path on the Cloudera Manager primary cluster node. The default path is /opt/cloudera/parcel-repo (the path is configurable in the Cloudera Manager user interface). The required parcel and metadata files are located in the spss_as-3.2.2.0-cdh.tar.gz file.
   1. Download the spss_as-3.2.2.0-cdh.tar.gz file from the IBM Passport Advantage® site.
   2. Extract the tar -zxvf spss_as-3.2.2.0-cdh.tar.gz contents by running the following command.

      tar -zxvf spss_as-3.2.2.0-cdh.tar.gz

   The parcel and metadata files are located in the extracted IBM-SPSS-AnalyticServer/3.2.2.0/cloudera/ directory. Copy the appropriate parcel and metadata files from IBM-SPSS-AnalyticServer/3.2.2.0/cloudera/ to /opt/cloudera/parcel-repo on the Cloudera Manager primary cluster node.

3. Use the following command to restart Cloudera Manager:

   service cloudera-scm-server restart

   The AnalyticServer parcel shows as downloaded after Cloudera Manager refreshes the parcel. You can click Check for New Parcels to force a refresh.

4. Click Distribute > Activate.

   The AnalyticServer parcel shows as distributed and activated.

5. In Cloudera Manager, add Analytic Server as a service. Refer to steps 7 and 8 in the "Online installation" section for more information.