Lost File Notifications
Dropped Network Connections
Corrupted Files and Faulty Filter DLLs
Disk-Full Condition
Corrupted Property Cache
Data Corruption and Internal Inconsistencies

Several kinds of errors are automatically detected by Microsoft Index Server and recovered. Other than hardware failures and situations where the catalog drive runs out of disk space, no manual intervention is needed.

Lost File Notifications

During normal operation of Index Server, all changes to the documents in the indexed directories are automatically tracked if the indexed directories are on computers running Windows NT. If the rate of file modification is very high, it is possible to lose some notifications due to buffer overflows. In that case, Index Server automatically schedules an incremental scan on the scope that lost notifications. No manual intervention is needed.

Dropped Network Connections

If a virtual root is pointing to a remote share and the connection to the remote share is lost, it is called a disconnected path. Index Server detects this situation and periodically scans the remote shares to see if the connection is active. No manual intervention is needed.

Corrupted Files and Faulty Filter DLLs

If corrupted files are detected by the CiDaemon process, the files will be marked as unfiltered. You can get a list of unfiltered files by issuing an adminstrative command to list the unfiltered files. There might also be a problem with the filter DLL used to filter those files. If a particular type of file is consistently not being filtered, you should contact the vendor of that filter DLL.

Disk-Full Condition

If the catalog drive starts getting full, indexing is temporarily paused. Filtering cannot continue unless disk space is freed up on the catalog drive. A message is written to the event log when the disk starts getting full. Administrators should monitor the event log for these messages and take corrective action. To avoid a disk-full condition, follow the configuration guidelines given in the installation page.

Corrupted Property Cache

If Index Server is shut down abruptly or a corruption is detected during normal operation, the property cache will be marked as corrupted. A recovery action will be performed on the property cache when Index Server is restarted. One event log message is written at the beginning of the recovery action and another at the end of recovery. During recovery, queries will be allowed but filtering will not start until the recovery is completed. No manual intervention is needed.

Data Corruption and Internal Inconsistencies

As a result of a power failure or other catastophic event, the index data may become corrupted beyond repair. In that situation, existing Index Server data is deleted and all the documents are refiltered. If corruption is detected during Index Server startup, the refiltering will happen automatically. However, if corruption is detected during normal operation, an event log message is written and no further queries are allowed. When this event is written to the event log, administrators should stop and restart Index Server.

When automatic recovery is performed by Index Server, an event is written to the event log. If the catalog specified in the event log message had default settings and all the virtual roots were indexed, no action is required.

Note If some virtual roots were marked not to be indexed, administrators have to explicitly disable indexing on those virtual roots again. Use the virtual root configuration option from the administrative Web page.