Recovery question
Gilberto Casiraghi
Hello all!
Imagine this scenario: I have a db running in archive mode. I do
a physical (offline) backup avery week (e.g. Saturday) and every
day I save the archived redo log (suppose every hour). All these
files are archived to tape.
Now suppose a disaster happens and I lose ALL the physical files
of my database.
The only chance I have it is to recover the files from the weekly
backup and then apply all the archived redo log saved to tape.
But after I start (mount) the db and I give the 'recover database'
command the system return this error: ORA-00264: no recovery required.
(maybe becouse the weekly backup is consistent !)
How is it possible to recover the database till the last saved redo
log ? Or I have to do a different weekly backup ?
PS: I am running Oracle Server 7.2.
Thanks in advanced.
Bye.
Gilberto Casiraghi.
Robin Bishop
I imagine this happens because the control file is in sync with the data
file headers when a 'consistent' offline backup is performed. My
understanding is that you need to restore the data files (and redo logs)
from the backup but keep the control file from the time of the crash. That
way Oracle knows the database is inconsistent and can roll forward to the
point in time recorded in the control file.
I'd be interested to know the outcome, since I am planning to run some
tests of this nature myself when I have a backup machine I can crash
without risk to my production database!
--
Robin Bishop
University of Brighton Computer Centre
Gilberto Casiraghi <ed...@candy.it> wrote in article
<3584F996...@candy.it>...
<snip>
ORACLE_DBA
HI, You have to copy only DATAFILES FROM YOUR BACKUP.( DO NOT COPY
CONTROL FILE ) .
Once copied, then
svrmgrl
connect internal
startup mount
set autorecovery on
recover database... ( This will bring your database in sink as if on the
day it failed...
but please remember that you have to have all ARCHIVE_LOG FILES and
redo_log_files as well..other wise you have to do INCOMPLETE RECOVERY )
Hope this Helps
VIjay K Banda
Robin Bishop <rm...@bton.ac.uk> wrote in article
<01bd9ad2$263ec720$4baa...@regan.bton.ac.uk>...
Saar
Gilberto Casiraghi wrote:
> Now suppose a disaster happens and I lose ALL the physical files
> of my database.
ok.. he also lost the control files...
> The only chance I have it is to recover the files from the weekly
> backup and then apply all the archived redo log saved to tape.
right...
> But after I start (mount) the db and I give the 'recover database'
> command the system return this error: ORA-00264: no recovery required.
That's normal.. you just brought a complete cold backup including
logfiles... database hasno way to know it needs recovery...
> (maybe becouse the weekly backup is consistent !)
>
it is, just like all cold backups should be.
> How is it possible to recover the database till the last saved redo
> log ?
you have to tell the database you are using a backed up controlfile using
the command:recover database using backup controlfile;
<suggestion of log>
Specify log: {<RET>=suggested | filename | AUTO | CANCEL}
it will give a suggestion of a log, if it looks good, just type auto.
when all logs applied and you try to issue:
alter database open;
you will get:
ORA-01589: must use RESETLOGS or NORESETLOGS option for database open
no problem you say, we'll try with the NORESETLOGS (sounds less harmful)
it comes back and says:
ORA-01588: must use RESETLOGS option for database open
well... beurocracy... fine put the RESETLOGS... database opens, however
you've just obsoleted all your backups, since you are starting to
generate redo with thread #1...
take a cold backup as soon as you can, because the one you have from
before the RESETLOGS is no good.
you will not be able to rollforward (applying redo) beyond this RESETLOGS
point...
> Or I have to do a different weekly backup ?
your backups are ok, just try not to lose your controlfiles and you won't
have to deal with this "using backup controlfile" option that then forces
you to RESETLOGS...
--
__ _ _ __ _ _ _ _ ___ ______________________________
(( /\\ /\\ ||) |\V/| /\\ /\\ >/ Principal Performance Engineer
_))//-\\//-\\||\ |||||//-\\\\//<_ Oracle Corporation Digital SBU
////////////////// Drop x's in email (spam) //////////////////////
John P. Higgins
Oracle thinks no recovery is required because the SCN's in all the file
headers match the SCN in the control files. This is what you always get
when you restore the control files from the same backup as the data files.
To get around this, you must add 'using backup controlfile' to your
recover database command.
Gilberto Casiraghi wrote:
> Hello all!
> Imagine this scenario: I have a db running in archive mode. I do
> a physical (offline) backup avery week (e.g. Saturday) and every
> day I save the archived redo log (suppose every hour). All these
> files are archived to tape.
> Now suppose a disaster happens and I lose ALL the physical files
> of my database.
> The only chance I have it is to recover the files from the weekly
> backup and then apply all the archived redo log saved to tape.
> But after I start (mount) the db and I give the 'recover database'
> command the system return this error: ORA-00264: no recovery required.
> (maybe becouse the weekly backup is consistent !)
>
> How is it possible to recover the database till the last saved redo
JHY
Try backing up your current control file when you backup your archive
logs.