disaster recovery

[steeve123]

At my client site:
One of his disk went down on a raid 4, the raid reconstruct the disk or what ever its supose to do when a disk goes down.(i dont know that much hardware)

Now when i start the database, i have :
ORA-01122: database file 1 failed verification check
ORA-01110: data file 1: 'C:\ORACLE\ORADATA\ORCL\SYSTEM01.DBF'
ORA-01207: file is more recent than controlfile - old controlfile

If i check the dates of my 2 control files, they seem fine, but oracle dont see it that way, anyways, now i have to recover from backup. Il be ok in 1 hour. But the real problem is the following.

One of the major reason to have a raid configuration, is that if a disk goes down, the raid reconstruct the failed disk and life goes on right?
But then its does something to my oracle files and when i try to startup the database, its says my files are srewed.

If someone has links or info on how to configure raid so it wont mess my oracle files, please dont hesitate.

[abhaysk]

Ok, you may try something like this

Shutdown
Startup Mount
Alter database backup controlfile to trace;

use the create contole file statement(generated in trace file) and for all datafiles, logfiles use REUSE command.

Shutdown
Startup nomount
Run Create ControlFile statement
OPEN DB

I think this should work..

before this make sure you have offline backup of all DB files.

Abhay.

[TomazZ]

pure guessing:
this storage of yours probably has some sort of write cache. I think it's possible that cache was not completely cleaned out on the disk prior to failure, so you've got some previous state of control files together with succesfully updated state of datafiles.

[steeve123]

Good observation, so what raid configuration would be ok so when one disk failed, i dont have that
ORA-01122: database file 1 failed verification check
ORA-01110: data file 1: 'C:\ORACLE\ORADATA\ORCL\SYSTEM01.DBF'
ORA-01207: file is more recent than controlfile - old controlfile



I know my raid 0 + 1 would be cool cuse the appication is write intensive. With no cache?

[marist89]

Because of this very reason, I prefer to place one of my control files on the local disk and two somewhere on the RAID (different controllers). In fact, during the power outage last week this saved my A** for three databases.

[TomazZ]

since your raid is probably not failing every day, you can as well do some recovery "once in few years". that's why you have backups, right?

[SANJAY_G]

Did you have sparesets available in the array when the disk failed?

[marist89]

Originally posted by SANJAY_G
Did you have sparesets available in the array when the disk failed?

absolutely, but the spares don't kick in when the power goes out.