Troubleshooting full-text indexing

Meridian Enterprise Server runs background tasks to extract text from documents for full-text indexing.

The task processes each document as follows

1. Copies the document content to the folder %TEMP%\BCTemp\
2. Starts a separate process to generate the full-text data.
3. Retrieves the result and stores it in the database
4. Removes the temporary file

The Meridian Explorer server uses the normal SQL Server or Oracle full-text search feature.

It uses a full-text index on two columns in two tables:

• [DocumentRevisionContentIndex].[Data] contains the text extracted from the document content
• [DocumentRevisionIndex].[Data] contains the combined text from the properties selected for full-text indexing

To populate the first table, generate the full-text indexes as described in Configuring full-text indexing. This should be done using a scheduled job to make sure that new documents are indexed. This will run the background task. This executes the IFilter for each document based on its file extension. The result depends on the design of the IFilter.

To populate the second table, rebuild the property indexes as described in Configuring item names and indexing. This is only required after including new properties in the search.

To troubleshoot full-text searches:

1. Confirm that the full text index feature of SQL Server or Oracle is enabled and the catalog (DocContentIndex) is created.

2. Confirm that the values you searched for are present in one of the Document*Index tables. If the values are not present, confirm that the IFilter is working properly.

   To check the IFilter result from the command prompt:

   1. Run the BCSyncUtil.exe program with the following arguments. By default, it resides in the folder C:\Program Files\BC-Meridian\Enterprise Server:

      BCSyncUtil.exe /CIdx /InFile:"C:\temp\MyTestFile" /OutFile:"C:\temp\output.txt"

      Or if the IFilter is 32-bit when shown as described in Viewing the full-text filters

      BCSyncUtil32.exe /CIdx /InFile:"C:\temp\MyTestFile" /OutFile:"C:\temp\output.txt"

If the IFilter is working properly, the output file will contain the text extracted from the file.

Meridian Explorer combines the full-text search results with a filter on the custom property search results. When both produce large results, the combination can cause searches to timeout.

To optimize complex or large searches consider enabling the following settings:

1. The Optimize grid queries option as described in Configuring the application options.
2. The following settings in the file C:\ProgramData\BlueCieloECM\Hyperion\WebConfigDto.dat. After changing settings, press F5 to reload the application.

Full-text search settings

Setting	Description
UseUnionForFullTextSelect	Set to False to not combine the full-text and custom property search results.
UseUnionForFullTextCount	Set to False to not combine the full-text and custom property search results in the total result count. This often reduces the server load for large search results.
RowLimitOnFullTextCount	Limits the maximum number of rows that are evaluated when getting the total result count. This may cause the total row count displayed in the paging toolbar to be incorrect (too low) for large search results but it can dramatically improve search performance. An alternative is to enable the Optimize grid queries option.