Critical: Data Recovery

[sandr]

I have an urgent and serious problem here with my Oracle instance. I know that I have no real clue what I am doing here and I promise to make myself familiar with the stuff I am working with and appropriate backup strategies but for the time being, any helpful hint is greatly appreciated.
I am running Oracle 8.1.7 on Linux. Our file system somehow crashed and we restored the files from our RAID setup. I was able to recover the redo logs and two of three control files and all the data files. It looks like we are running in NOARCHIVELOG mode.

We also export the database daily. We never had a problem before with re-importing data to the database. With the current dump file, imp crashes after a certain table. With another dump file, it just skips some tables.

Now my next idea was to recover the database from a cold backup. When I startup the database, I get

ORA-00227: corrupt block detected in controlfile: (block 1, # blocks 1)
ORA-00202: controlfile: '/var/spool/ora8/m01/oracle/oradata/ora8/control02.ctl'

I assume that this is due to recovery from RAID. So I created a new control file with NORESETLOGS and startup mount'ed the database.

But now I am a bit clueless on how to proceed. I think I have to recover the database (in)completely? When I do a recover database, it always gives me

ORA-00279: change 21200170 generated at 08/22/2002 06:54:13 needed for thread 1
ORA-00289: suggestion : /var/spool/ora8/m01/oracle/product/817/dbs/arch1_3522.dbf
ORA-00280: change 21200170 for thread 1 is in sequence #3522

I then tried

alter database recover database until cancel

but I get the same error message but no prompt. Obviously, media recovery has been started but I don't see whether there is something going on since there is no prompt.

Can anybody help me. Thanks so much in advance.

Denis

[ramaral]

Sandr,

you mentioned that this database is in NOARCHIVELOG mode. Is this correct? If so then you can't recover the database as it stands and you cannot just switch archivelog modes to and hope to apply archivelogs. In NOARCHIVELOG mode no archives are generated that's why it appears to you that the nothing is happening when you are trying to recover.

Best you can do is recreate the database completely (not from backups) Re-create it using dbca or sql whichever you are most comfortable with. Then import. And keep your fingers crossed that the import is fine and wasn't affected by the crash either. Oh before doing this do an fsck on the filesystem. Make sure it really is clean before proceeding.

HOH

[Mnemonical]

I believe your control_file parameter is '/var/spool/ora8/m01/oracle/oradata/ora8/control01.ctl', '/var/spool/ora8/m01/oracle/oradata/ora8/control02.ctl'

so, change it to '/var/spool/ora8/m01/oracle/oradata/ora8/control01.ctl'

then delete control02.ctl and recreate it.

about the recover, try to recover from your redo if your are not in archivelog. This should help if your cold backup is new and your redo log file is big, otherwise it wont.


F.