Troubleshooting the Indexing Service

[Previous] [Next]

This section lists common error messages and possible solutions. Some of the most common problems with the Indexing Service can be fixed quite easily. For more complex issues, search the Microsoft Knowledge Base on line at http://support.microsoft.com/search/ for solutions.

No Documents Matched the Query

• Instead of receiving a list of matching documents when you perform a query using the Indexing Service, you may receive the error message "No Documents Matched the Query." This common error means that the catalog is responding correctly but that there is an incorrect setting somewhere. To find the incorrect setting, try each of the following procedures in turn, testing after each one.
• Try the samples in the \inetpub\issamples\sdk directory. You may need to change the pointer to the catalog if you are not using the default catalog. In an .IDQ file, you should see a line like the following:

CiCatalog=d:\inetpub\wwwroot\tmjs_index

Make sure that the path is pointing to the directory that contains the catalog's Catalog.wci directory. In an Active Server Pages (ASP) file, you will need to add the Q.Catalog line in the code segment below:

set Q = Server.CreateObject("ixsso.Query") set util = Server.CreateObject("ixsso.Util") Q.Catalog="<catalog_name>" Q.Query = CompSearch Q.SortBy = "rank[d]" Q.Columns = "DocTitle, vpath, filename, size, write, characterization, rank" Q.MaxRecords = 300 

Replace <catalog_name> with the name of your catalog as it appears in the Indexing Service console. (Include the quotation marks.)

• In the code of your page, make sure that one of the following is true:

In an .IDQ:

CiScope=/ CiRestriction=%CiRestriction% CiCatalog=<path to the directory holding the Catalog.wci directory> 

In an ASP:

FormScope=/ Q.Query=compsearch Q.Catalog="<catalog_name>" 

Replace <catalog_name> with the name of the catalog as it appears in the Indexing Service console. This will ensure that you are running the widest possible search with the most chances of returning results against the correct catalog, with nothing on the query line to prevent results from appearing.

• Open Internet Services Manager from the Administrative Tools folder on the Programs menu. Right-click the root of your Web site and choose Properties. On the Home Directory tab, make sure that the Index This Resource check box is selected. Then try your query again.
• The Indexing Service does its indexing as the System account. If the System account does not have at least Read permissions on the files that are to be cataloged, the files will not be indexed. Also, make sure that the System account has Full Access permissions on the root of the drive that physically contains the catalog and on the Catalog.wci directory.
• Users are allowed to view only results that they have permissions to see. If you are accessing the query page as an anonymous user, make sure that the anonymous user has at least Read permissions on the document you are trying to find.
• Authenticate coming into the query page as an administrator if possible. Try setting the directory that the query page is located in to Basic/Clear Text authentication only, so that you receive a prompt when you attempt to connect to that page. If you get results, you have a permissions issue. If you log on as an administrator and still receive a "No Documents Matched the Query" message, you probably do not have a permissions issue.
• If you are querying on a specific page and it is returning in a #filename query, but it is not being filtered (the abstract is not populated), and it is an HTM, HTML, or ASP page, make sure you do not have a ROBOTS=NOINDEX metatag in the header of the document. This will not prevent indexing, but it will prevent filtering.

.PDF Files Aren't Indexed

Adobe makes a filter to enable the indexing of .PDF (Portable Document Format) files. After you install the filter, you may notice that .PDF files are no longer indexed after the Indexing Service is restarted or the computer is restarted.

The problem is caused when the Indexing Service restarts and reregisters all of the filter .DLL files. One of these .DLL files has a default association with the .PDF extension and therefore registers itself as the filter for these files. To fix this problem, ensure that the list of filter .DLL files includes the PDF filter by following these steps:

1. Start the Registry Editor (Regedt32.exe).
2. Locate the following key in the registry:

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\ ContentIndex

3. Add the path to the Pdffilt.dll to the DLLsToRegister key. The path to the .PDF filter should be added as the last entry in this key.
4. Close the Registry Editor.

Query Produces Inconsistent Results

When you use the Indexing Service to run a query and you set the sort method to anything other than Rank Descending, you do not receive the top matching records and you may get a different set of files on subsequent queries. Sorting on Rank Descending is the only way to get the top matching records from the catalog when the maximum number of returns is limited. Sorting on anything other than Rank Descending returns a subset of the total matching documents.

Catalog Is Reportedly Corrupted

If the Indexing Service reports that the catalog is corrupted after the indexing process is completed, it means that a file can't be filtered and that Filter Retries is set to a number greater than 4. This can occur as soon as the indexing process has completed or several minutes later.

When this happens, the information that the filter process sends to the Indexing Service causes a report that the catalog information is corrupted even though the data on the drive is fine. To fix this, start the Registry Editor (Regedt32.exe) and navigate to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\ContentIndex\FilterRetries. Change the value of this key to 4 or less, close Registry Editor, and restart the Indexing Service.

Indexing Is Slow and Some Documents Aren't Indexed

On occasion, it may take an extremely long time to index documents, and some documents may appear not to get indexed at all. In addition, the abstract for documents may be blank or contain incorrect information. This is usually caused by third-party software that places a lock on the Web content that you are attempting to index. Antivirus software programs and any other software that monitors or scans your Web content for extended periods of time can cause this problem. To work around this problem, disable any software that may be monitoring the Web content.