The recovery procedures, known simply as “recovery,” rebuild the server’s databases from the transaction logs. The following situations cause recovery to run:
Adaptive Server start-up
Use of the load database command
Use of the load transaction command
The recovery isolation mode setting controls how recovery behaves when it detects corrupt data while reversing or reapplying a transaction in a database.
If an index is marked as suspect, the system administrator can repair this by dropping and re-creating the index.
Recovery fault isolation provides the ability to:
Configure whether an entire database or just the suspect pages become inaccessible when recovery detects corruption
Configure whether an entire database with suspect pages comes online in read_only mode or whether the online pages are accessible for modification
List databases that have suspect pages
List the suspect pages in a specified database by page ID, index ID, and object name
Bring suspect pages online for the system administrator while they are being repaired
Bring suspect pages online for all database users after they have been repaired
The ability to isolate only the suspect pages while bringing the rest of the database online provides a greater degree of flexibility in dealing with data corruption. You can diagnose problems, and sometimes correct them, while most of the database is accessible to users. You can assess the extent of the damage and schedule emergency repairs or reload for a convenient time.
Recovery fault isolation applies only to user databases. Recovery always takes a system database entirely offline if it has any corrupt pages. You cannot recover a system database until you have repaired or removed all of its corrupt pages.
