Hot Backup Question

[Andrew E]

I just took over DBA responsibilities for an Oracle 8i database
running on Linux. Although I've been working with relational
databases for some time, I'm a bit green on Oracle so forgive me.

My first task was to implement a backup strategy (currently there is
none). I did a great deal of research both on this board and using
the Oracle Press 8i handbook. I made a cold backup (with oracle
shutdown) and sucessfully moved it from my production machine to my
test machine and this works fine. My question is around "hot
backups".

I sucessfully switched the database into ARCHIVELOG mode. I am then
running this script (abbreviated for brevity) to back it up:

connect internal as sysdba
alter system switch logfile;
alter tablespace SYSTEM begin backup;
cp $ORACLE_DATA/system*.dbf /home/oracle/backup
alter tablespace SYSTEM end backup;
....I do the above for each tablespace...
archive log stop
....Run procedure to archive logfiles...
archive log start
alter database backup controlfile to
'/home/oracle/backup/controlfile.bck';

This all appears to run fine, but I have two questions on my
unsuccessful restore:

#1. Something appears to be wrong with my control file backup. Upon
moving the backed up files to my "test" machine, I place them in the
data directory, and rename my control file backup to the correct name
and run this procedure:

- I start the instance
- I mount the database
- I issue a recover database
I then receive the following error:

ORA-00283: recovery session canceled due to errors
ORA-01610: recovery using the BACKUP CONTROLFILE option must be done

Does anyone have any thoughts around this?

#2.

If I use 'recover database using backup controlfile', I get further,
but I am prompted for an archived log that is 1 sequence number
greater then I have! I checked both in my archived log area that my
backup script makes as well as the "live" area on the production
server.

Why would this be happening?

Thanks so much for everyone's time.

Andrew

[Joel Garry]

dr****@materiamagica.com (Andrew E) wrote in message news:<1e*************************@posting.google.c om>...

I just took over DBA responsibilities for an Oracle 8i database
running on Linux. Although I've been working with relational
databases for some time, I'm a bit green on Oracle so forgive me.

My first task was to implement a backup strategy (currently there is
none). I did a great deal of research both on this board and using
the Oracle Press 8i handbook. I made a cold backup (with oracle
shutdown) and sucessfully moved it from my production machine to my
test machine and this works fine. My question is around "hot
backups".
This board is obsolete, use google to search for Oracle newsgroup
Charter.
You should be researching on comp.databases.oracle.server for
administration issues.

I sucessfully switched the database into ARCHIVELOG mode. I am then
running this script (abbreviated for brevity) to back it up:

connect internal as sysdba
alter system switch logfile;
Do switch for as many logfile groups as you have.
alter tablespace SYSTEM begin backup;
cp $ORACLE_DATA/system*.dbf /home/oracle/backup
alter tablespace SYSTEM end backup;
...I do the above for each tablespace...
archive log stop
...Run procedure to archive logfiles...
archive log start
No, don't stop archive logging. You could grind everything to a halt.
alter database backup controlfile to
'/home/oracle/backup/controlfile.bck';

This all appears to run fine, but I have two questions on my
unsuccessful restore:

#1. Something appears to be wrong with my control file backup. Upon
moving the backed up files to my "test" machine, I place them in the
data directory, and rename my control file backup to the correct name
and run this procedure:

- I start the instance
- I mount the database
- I issue a recover database
I then receive the following error:

ORA-00283: recovery session canceled due to errors
ORA-01610: recovery using the BACKUP CONTROLFILE option must be done

Does anyone have any thoughts around this?
01610, 00000, "recovery using the BACKUP CONTROLFILE option must be
done"
// *Cause: Either an earlier database recovery session specified
BACKUP
// CONTROLFILE, or the controlfile was recreated with the
RESETLOGS
// option, or the controlfile being used is a backup
controlfile.
// After that only BACKUP CONTROLFILE recovery is allowed
// and it must be followed by a log reset at the next
database open.
// *Action: Perform recovery using the BACKUP CONTROFILE option.

#2.

If I use 'recover database using backup controlfile', I get further,
but I am prompted for an archived log that is 1 sequence number
greater then I have! I checked both in my archived log area that my
backup script makes as well as the "live" area on the production
server.

Why would this be happening?

Could be you need to recover until cancel, and then just cancel. Or
maybe it's the lack of enough switches earlier. Or maybe you need to
supply the redo log file names.

See http://metalink.oracle.com/metalink/...&p_id=131865.1

jg
--
@home.com is bogus.
http://www.bayarea.com/mld/mercurynews/news/5676110.htm