Assumptions
-
There are at least two servers in the deployment running the Search service.
-
The Search service is stopped after IBM Connections is installed.
-
The following steps are based on a IBM Connections installed on a medium topology.
Procedure
1. Identify the nodes in the Search cluster.
a. Log in to the Integrated Solutions Console and click
Servers >
Clusters >
WebSphere application server clusters.
b. Click
cluster_name, where
cluster_name is the name of the Search cluster.
c. In the Additional Properties area, expand
Cluster members and then click
Details.
d. In the table of cluster members, make a note of the nodes that host the cluster members.
2. Copy the contents of the shared_data_directory_root/search/stellent directory from the shared content folder to a local directory on each node.
Use exactly the same path on each node. The following path is an example only and might be different on your operating system:
AIX or Linux: /opt/IBM/Connections/data/local/search/stellent
IBM i: /QIBM/ProdData/IBM/Connections/data/local/search/stellent/
Windows: C:\IBM\Connections\data\local\search\stellent
The new directory contains the exporter executable file.
a. IBM i: In qshell, run the command:
chown -R QEJBSVR /QIBM/ProdData/IBM/Connections/data/local/search/stellent
b. IBM i: Run the CL command to add a system level environment variable:
LIBPATH: ADDENVVAR ENVVAR(LIBPATH)
VALUE('/QIBM/ProdData/IBM/Connections/data/local/search/stellent/dcs/oiexport')
LEVEL(*SYS) REPLACE(*YES)
3. On the Deployment Manager, update the FILE_CONTENT_CONVERSION WebSphere variable to point to the exporter file in the local directory on each node.
For more information about WebSphere® variables, see the Changing WebSphere Application Server environment variables topic.
-
For example: /opt/IBM/Connections/data/local/search/stellent/dcs/oiexport/exporter
-
The exporter file must be in the same file path on all nodes.
On AIX and Linux: Add the following entry as a WebSphere variable of the cluster hosting the Search application.
This must be done for all nodes in the deployment.
Note that SearchBinariesHome is the path to the directory that you specified in Step 2. In the example above, this is
/opt3/IBM/ICLocal/search/stellent/dcs/oiexport
4. Start WebSphere Application Server on each node.
5. Run the exporter command. The exporter command has the following syntax:
exporter inputpath=
outputpath=
outputid=FI_SEARCHTEXT
For example:
$ ./exporter inputpath=/root/translation.pdf \ outputpath=/root/testStellent.txt outputid=FI_SEARCHTEXT
If the text extraction worked successfully, you will see the following output:
6. If this test fails please review the following technote.
7. Stop the Search application on all nodes.
8. The Search index and persistence folders should be empty before creating the initial index.
Verify that the folders defined by the IBM WebSphere® Application Server variables SEARCH_INDEX_DIR, SEARCH_INDEX_BACKUP_DIR
and SEARCH_INDEX_DIR SEARCH_INDEX_SHARED_COPY_LOCATION are empty. If any files exists in these folders, remove them before proceeding.
9. Stop ALL Clusters from the WebSphere administration console.
10. Optional : Enable the following trace strings for the Search service.
*=info:com.ibm.lotus.connections.search.admin.lucene.IndexCreation=all:com.ibm.connections.search.seedlist.crawler.impl.SeedlistCrawlerImpl=all:com.ibm.connections.search.process.work.IndexingWork=all:com.ibm.connections.search.process.initial.InitialProcess=all:com.ibm.connections.search.process.incremental.IndexBuilder=all:com.ibm.lotus.connections.search.admin.index.impl.IndexBuilderQueue=all:
11. Start all the IBM Connections services. Shortly after the Search service starts, it will begin crawling the IBM Connections services.
12. To verify that Search is crawling on a regular basis, open the SystemOut.log file that corresponds to the application server instance on
which Search is running and look for the following log messages:
CLFRW0297I: Search is starting to crawl the {0} component and
CLFRW0294I: Search has finished crawling the {0} component
13. In a deployment with multiple Search nodes:
Note that only one node in the cluster does the crawling although all the nodes do the incremental index building (based on the crawling that Search node does). Because the crawling is performed by a single Search node, you only see the log messages on that node in the cluster.
14. To verify that Search is building the index incrementally, open the SystemOut.log file that corresponds to the application server instance on which Search is running and look for the following log messages:
CLFRW0285I: Search is starting to build the index for {0}
CLFRW0282I: Search has finished building the index for {0}
In a deployment with multiple Search nodes: Check that you can see this log message listed on all the Search nodes in the cluster.
15. After indexing takes place the Post Processors are run.
16. After the Post Processors are finished running the social analytics indexers begin.
17. The final task to run as part of the initial index build process is the optimization task.
18. When the index is being built as expected, these messages display for each of the IBM Connections applications that you installed and
configured as part of the scheduled crawling task. By default, this task is scheduled to run every 15 minutes and includes all the
IBM Connections applications that you installed.
19. In a deployment with multiple Search nodes: Check that you can see the log messages listed on all the Search nodes in the cluster.
20. After the index is built on the node, the following message appears in the logs:
CLFRW0869I: Search indexing of services [profiles, dogear, communities, activities,
blogs, forums, wikis, ecm_files, files, status_updates, calendar,] to directory
C:\IBM\Connections\data\local\search\index has completed successfully.
The index is now copied to shared location. The other Search servers will then be able to copy and restore this index copy as their default index.
21. The index is then copied to each node in the cluster and the following messages appear on the nodes:
CLFRW0584I: Search is now using the search index at C:\IBM\Connections\data\local\search\index
to handle search queries.
CLFRW0723I: This server has been registered with the index management process.