Re: database revovery issue(my system is down)

"utkanbir" <hopehope_123_at_yahoo.com> wrote in message news:f6c90ebe.0310282028.8f8834c_at_posting.google.com...
> Hi Gurus,
>
>
> I am using oracle 9i rac on linux itanium , with emc disk + raid 5. We
> are building a datawarehouse system .Today ,
> we got error in one of the disks , unfortunately that disk contains
> system.dbf.
> Than the disk is replaced , when we try to start the database , we got
> corrupted blocks . DB verify also reported corrupted blocks in
> system.dbf and user.dbf. So we decided to recover database BUT:

You should have "decided" to restore the two datafiles that were reported as problematic, and recovered the *datafiles*, not the database. The database isn't at fault here, only two specific data files.

> our data backups does not contain all the tablespaces , it contains
> system,undo,user ... but not our data tablespaces. (This is beacuse of
> the itanium machine .Our data backup software has no version for
> itanium , the only way to backup data is first store it in disk than
> tape. We dont have enough disk space to store all the data archive.)

Let's see if I got this right. You are building a data warehouse RAC on a machine for which you have no reliable and functioning backup software and for which you have no backup medium capable of storing the stuff you actually care about -namely your data. Are you quite sure this is the direction you want to be going in??

> Is it possible to recover this database by using this backup?

I think so. You report corruption in SYSTEM and USER, and you state you have backups for system.dbf and user.dbf. Since you only need to restore those two files, then you should be able to do the resotre. Can you then do the recovery? So long as you have all the archived redo logs produced since the time of those files' backups, and the online redo logs, yes.

>If i
> recover system,user,undo tablespace , will i use the other tablespaces
> again? I know that other tablespaces have no corrupted blocks.
>
> 'recover database using backup controlfile'

Oh, why do people "experiment" with recovery commands.

Did you restore a binary version of the controlfile? No?? Then why in Heaven's name would you even think you need the 'using backup control file' command?

And why are you restoring the undo tablespace? Did dbverify report that to be corrupt?? Not according to your post it didn't. So why are you restoring things that don't need restoring??

>
> We tried the command , it asks some redo log file numbers to recover ,
> than we
> resetted the redo logs. But we still get data block corruption
> messages and database cant be opened.
>
> If i restore system.dbf , undo and user , can i use other tablespaces?
> May be we made a great m&#305;stake of resetting logs.Instead of
> resetting logs , we must recover system.dbf from backup ??

Your mistake is to blunder about, issuing inappropriate restore and recovery commands, and hoping you'll muddle through somehow.

Learn how to backup a single instance database, and to recover it, under a variety of circumstances, and then apply that knowledge (without any fundamental change) to a RAC database. But you need to learn what it is you're supposed to do *before* you try doing it.

And if this is a serious production database, then get Management to resource backup and data storage issues before you go any further, because otherwise you'll just be wasting their (and your) time.

>
> **********
> By the way , i can open database as read only , and get my data. But
> some queries against dictionary gets corruption messsages , suc as
> tab$, dba_tables,etc..
> ***********************
>
>
> So , within this current situation , if i do full restore and returns
> the tablespaces that are inside archive but not the others , what will
> be the status of other tablespaces? I mean if dbserver stores the
> information about tablespaces inside system or control files , can
> recovering them solve my problem since those tablespaces data are
> valid? Or will the status of other tablespace be offline? Than may be
> a patch can be applied?(if exists) Is there any number inside
> tablspaces (may be in headers) which describes the redo log number
> which is necessary for that tablespace? What is the relation between
> redo log file numbers and tablespaces?

So many questions. Learn about backup and recovery, and the answers will be obvious to you. You've royally stuffed things up by throwing ad hoc recovery commands at a database that didn't need them. You should just have restored the two datafiles that dbv reported to be corrupt, and said 'recover datafile 'system.dbf', followed by 'recover datafile user.dbf', rollowed by 'alter database open'. End of story. Once you start restoring other things, and issuing recover commands that aren't needed, Lord knows what state your database is in, and since you don't have a complete database backup you can't reverse the process to find out.

Bye-bye database.

Regards
HJR
>
> Any help will be appreciated,
> Kind Regards,
> hope (although it remains less ...)
Received on Wed Oct 29 2003 - 01:56:21 CST