Re: Simple archive question

Hi Friends ,

thanks for your help. After reading your postings and backup concept manual ,

i have come to a point:

Last night , i have tried this:

1. shutdown immediate ;
2. copy all the files including redo , control , undo
3. start dbserver and do something : insert into a .....
4. shutdown the db .
5. copy back only data files and undo not redo or control
6. startup mount
7. recover database

This finished successfuly and changes after the cold backup are in db.

Based on your postings , i understand that i can also copy the control files back to the original location and issue recover database using backup control file:

after shutting down the db:

1. copy data files and control files back.
2. startup mount
3. recover database using backup controlfile

But indeed this is what i tried to complish at the beginning. This fails after some point during redo log applying , and i am unable to start the dbserver.
I will try this once again but i want to be sure whether the above steps are
correct in order to recover full database .

And finally , these are the sql commands which i want to be sure whether they are logged and can be recovered:

insert /*+append*/ into a select * from dual;

table a is logged.

insert /*+append*/ into b select * from dual;

table b is not logged.

after recovering the db :

 table a has all the rows available. :
** If insert append is logged does it still by pass the buffer cache?

table b returns : ora-01578

and data is not available. It is intereseting to get this result for table b.
I hope to see the table b data without the last insert append statement , but i think it is reasonable since append directly writes to the table data bypassing the buffer cache.

In this case , my live system must be carefully examined which is a datawarehouse and full of nologgng tables with insert append statements.

I will be appreciated for all the corrections, comments ,... Kind Regards,
tolga   

sfitzgerald_at_centurytel.net wrote in message news:<1110482469.862326.179470_at_f14g2000cwb.googlegroups.com>...
> You are missing some important points.
>
> First, what type of recovery are you trying to simulate? Loss of only
> some db files? Loss of entire machine? The type loss dictates what
> you would restore. You just don't go copying all backup files into
> production if you don't need them for the recovery.
>
> Second, you should not be backing up the online redo log files. These
> should be multiplexed, but not included in a backup scenario (for just
> the reason you discovered, you over-wrote the production ones when you
> copied your files back).
>
> Third, your control files should be multiplexed as well. You should
> also back them up, but should only restore them if absolutely
> necessary.
>
> Read up on backup/recovery concepts because there are so many different
> ways to backup and recover and you need to be familar with all of them.
Received on Fri Mar 11 2005 - 01:13:44 CST