Saturday 12 July, 2008 - 22:53
It has taken me three (3) attempts to get a good backup of the padstow cluster.
Attempt #1 failed because of corrupt blocks in the SYSAUX tablespace again! Since I did not have a valid backup, I could not recover from the block corruptions. Deleted and created the EXAMPLE database once more.
Attempt #2 failed because the ASM instance on padstow2 would not start. This backup could be okay because the archived redo log files are in the shared storage.
After a cluster reboot, both nodes of the cluster database came online and I was able to get a full backup.
Some interesting commands:
BACKUP VALIDATE DATABASE;
This RMAN command reads all of the disk blocks to find corrupted ones without stopping at the first corrupted one. These blocks are listed in the V$DATABASE_BLOCK_CORRUPTION view.
BLOCKRECOVER CORRUPTION LIST;
This RMAN command reads the list of blocks from V$DATABASE_BLOCK_CORRUPTION and attempts to fix them. How simple is this?