FLASH_RECOVERY_AREA and flashback database
PRISMJun 3 2013 — edited Jun 4 2013Basic question here. My db_recovery_file_dest_size is set to 65GB. This is usually enough to cover our archive log generation for 1 day, then we shut down and take a cold backup at night, after which we use RMAN to delete all archivelogs so we can start anew in the morning.
My question is what happens when I enable the flashback database. My db_flashback_retention_target is set to 1440 (1 day). These files are written to the same db_recovery_file_dest directory as my archive logs. Let's say flashback database needs 4 GB to meet the 1 day retention target. That means I have 61 GB of free space for archive log generation.
Is there a way to prevent the flashback database from using more space than it requires to meet its retention target? I don't want flashback to keep creating flashback logs, taking up space that I need for my archive logs. We've had many instances where flashback ends up using more than 50 GB of the flash_recovery_area, even though it doesn't need that much space. When that happens, I don't have enough room for new archive logs, which leads to the database stalling because it has no place to archive the full redo logs.
Let me know if I need to clarify. I understand that we aren't using the flash recovery area or RMAN the way they were intended for backups, but this method works for us (usually). We are running Oracle Enterprise 11.2.0.3 on Windows Server 2003 x64.
Thanks in advance,
Tom