The name must not begin with "web," which the search appliance reserves for naming web content feeds. The database entry name must match the [a-zA-Z_][a-zA-Z0-9_-]* pattern, that is, you can use letters or an underscore for the first character, and alphanumeric characters, underscores, and dashes in the rest of the name.

Check that your database supports the JDBC driver that is listed in "Administering Crawl: Database Crawling and Serving," which is available from the Google Search Appliance help center.

The information retrieved from the Crawl Query provides the data for the indexing. For examples of crawl queries, refer to "Administering Crawl: Database Crawling and Serving," which is available from the Google Search Appliance help center.

To view the search results as raw XML, use the identity_transform.xsl stylesheet. When you are using this style sheet, the page that appears when a user clicks a link to fetch a particular database search result appears as unformatted HTML. To see the XML, view the source of the page in your web browser. For information about identity_transform.xsl, refer to "Administering Crawl: Database Crawling and Serving," which is available from the Google Search Appliance help center.

The base URL should have the following format: http://www.baseurl/docnum={docid} where {docid} represents the values in the column specified in the Document ID Field.

The Serve Query is used when the user clicks on a search result link to retrieve and display the desired document data from the database. The Serve Query displays result data using the ’?’ in the WHERE clause to allow for particular row selection and display. The Primary Key Fields must provide the column names for the field to substitute with the ’?’.

For examples of serve queries, refer to "Administering Crawl: Database Crawling and Serving," which is available from the Google Search Appliance help center.

The Primary Key Fields must provide a unique identifier for a database query result. This may be a combination of column names which produce a unique permutation from the corresponding values. The Primary Key allows each result row from a database query to be reliably identified by theServe Query. Primary keys must be listed in exactly the same order as they appear in the WHERE clause.

The value from this column is displayed as the link for each search result that refers to a row from the Crawl Query. For example, in a company directory, if an HTML page exists for each record, and the links are always in the same format (such as http://corp.company.com/hr/Joe_Employee.html), then the appliance displays that link when it serves results. Specify the name of the field that contains the URL, such as "Employee_name".

The search appliance usually transforms data from crawled pages, which protects against security vulnerabilities. If you cause the search appliance to crawl BLOB content by filling in these advanced settings, certain conditions could open a vulnerability. The vulnerability exists only if both of these conditions are true:

• A perpetrator has access to the database table.
• You are using secure search, which causes the search appliance to request usernames and passwords or other credentials.

Enter the name of the column that lists the modification type; valid values for Action Field are add or delete.

The database administrator populates the Action column using database triggers. The Action column need not be part of the source table, but instead part of a separate logging table which is joined with the source table holding the content by means of primary keys. The database administrator purges the logging information of all entries dated before the last successful incremental crawl.

If you select BLOB, enter the name of the database column that contains the standard Internet MIME type values of Binary Large Objects, such as text/plain and text/html.

Database feeds do support content in BLOB columns. The MIME type information must be supplied as a column. BLOBs use Base64 binary encoding. The XSL transformation from the specified stylesheet is not applied to BLOB data, or its associated row.

BLOBs are automatically binary encoded as Base64. BLOBs display HTML snippets but their links are to the original binary format (for example, MS Word, PDF). The cache link for the snippet provides an HTML representation of the binary data.

Multiple BLOBs in a single query are not supported. A CLOB can be treated as a BLOB column or as text.