Re: logfile questions

Start with three or four moderately-sized logfiles. Then check your alert.log files for messages that say "unable to advance to logfile filename", "Checkpoint not complete". If you see this message, try increasing the number, then the size, of logfiles and continue to monitor.

Logfiles are like journal files. They record all database changes and commits. When the database is restarted after an abnormal termination, the changes that hadn't yet been written to the database are applied. A rollback segment stores the previous version of deleted or updated rows, to allow rollbacks. They are also used to provide read consistency, i.e., if you execute and query, and rows in the result set are changed before you fetch them, the data in the rollback segment is returned to you when you do fetch those rows.

---
Paul Baumgartel, Principal Consultant   
paul.baumgartel_at_worldnet.att.net
Adept Computer Associates, Inc.          (203) 735-3524