Rama
> >> Again to be on the safe side! If we lose all the 3 control files, we
> thought we could get the old control file from the RCAT and mount the
> database. Is there any otherway to mount the database, when all control
> files are lost...?
Not that I know of, from what I know if you loose all control files you would
have to restore them in a nomount mode, then once it is restored you mount your
target database. One you've gotten past that you can perform any type of other
recovery that is necessary. Here is an example of a recovery up to the latest
SCN for my TEST database. The SCN is is changed everytime I run my generation
script.
connect target rman/rman_at_TEST;
connect rcvcat rman/rman_at_RCAT;
startup nomount;
run
{
allocate channel d1 type disk;
setlimit channel d1 kbytes 2097150 maxopenfiles 32 readrate 200;
set until scn 155108;
restore controlfile;
alter database mount;
restore database;
recover database;
release channel d1;
}
sql 'alter database open resetlogs';
set snapshot controlfile name to
'/db0/orabackup/TEST/snapshot_controlfile_TEST.ctl';
reset database;
About the extra precautions with backing up a second copy of the control file,
no problem with doing that, I just was curious as to if you were doing it
because you weren't using a catalog. According to Oracle's documentation, that
was the only time they suggested making an extra backup of the controlfile. I
also had added into my scripts the code to make an extra backup as well.
One addition comment that I wanted to make, is everytime I perform a RMAN
backup I perform a full resync of the catalog to the controlfile. The reason I
do this, is if you do a backup, it only performs a partial resync which may not
get everything. It really can be a good thing to do, I would suggest
performing a full resync at least once, possibly multiple times throughout a
day.
Regards,
Tom Tyson
- Rama Malladi <rmalladi_at_inteliant.com> wrote:
> Tom,
> Thanks for your mail. As I mentioned in my mail, After having a long
> meeting, our team decided that to be on the safe side, it is good to have
> both Full and Level0 on day1. We restored successfully several times after
> that. But as you said it might have been redundant. Team decision at that
> time was, it was okay to be on the safe side, even if it meant being
> foolish!
>
> Coming to your questions...
>
> * Are you using a recovery catalog?
> >> YES
>
> * Why are you placing your database in restricted mode for both scripts, the
> "full" backup requires that your db in mount mode.. so no regular users can
> log on, and the "level 0" to me, most likely should let your users work,
> after all.. it is a hotbackup.. :)
> >> We are doing the Hotbackups at 4am, when there are only batch jobs
> running. No new sessions get started at that time. Infact, due to security
> if anybody tries to login at that time, triggers will log them out
> automatically. So we put it in Restricted mode.
>
> * Why are you backing up the control file after the full backup? Typically
> this is only needed if you aren't using a recovery catalog, and need a
> controlfile with the last backup info, is this the case?
> >> Again to be on the safe side! If we lose all the 3 control files, we
> thought we could get the old control file from the RCAT and mount the
> database. Is there any otherway to mount the database, when all control
> files are lost...?
>
> Thanks for your feedback
> Rama
> ======================
>
> -----Original Message-----
> Sent: Thursday, November 02, 2000 5:46 AM
> To: Multiple recipients of list ORACLE-L
>
>
> Just wanted to add my .02 cents...
>
> The only difference between a full backup and a incremental (differential /
> cummulative) backup is that one is consistent and the other is not. Another
> difference is that the full backup causes every used block to get backed up,
> but not unused blocks. But you only need to apply archive log files to your
> iconsistent database... Other than that, you can recover just as well. I
> have
> had clients running for months with only Level 0's and 1's using a
> incremental
> cummultative backups with no problems. I've had to perform db restores, and
> had no problems recovering any part of the database. A couple of questions
> on
> your scripts..
>
> * Are you using a recovery catalog?
>
> * Why are you placing your database in restricted mode for both scripts, the
> "full" backup requires that your db in mount mode.. so no reqular users can
> log
> on, and the "level 0" to me, most likely should let your users work, after
> all.. it is a hotbackup.. :)
>
> * Why are you backing up the control file after the full backup? Typically
> this is only needed if you aren't using a recovery catalog, and need a
> controlfile with the last backup info, is this the case?
>
> Anyway.. just thought I'd pipe in.. no offense meant..
>
> Tom Tyson
>
>
> --- Rama Malladi <rmalladi_at_inteliant.com> wrote:
> > Tom,
> > FYI.. I am posting both the Full as well as Level0 backup scripts that we
> > have ... Still I am not sure if Level0 is same as Full or are they
> > different.. After having a long meeting we all decided that to be on the
> > safe side, it is good to have both on day1. Since then we do only Level1,2
> > during the week and Level0 once a week.
> >
> > FULL
> > ====
> > {
> > sql 'alter system enable restricted session';
> > execute script alloc_disk;
> > execute script defin_maxcorrupt_for_system;
> > backup full
> > skip offline
> > skip readonly
> > skip inaccessible
> > tag Back_Fulldb_hsys
> > filesperset 5
> > format '/backup/oracle/arbor-106/Fulldb_arborora_%t_%s_%p.bs'
> > (database);
> > execute script rel_disk;
> > sql 'alter system switch logfile';
> > execute script arch_log_current;
> > allocate channel c1 type disk;
> > backup current controlfile format
> > '/backup/oracle/arbor-106/Control_arborora_%t_%s_%p.bs';
> > release channel c1;
> > sql 'alter system disable restricted session';
> > execute script back_all_alog;
> > }
> >
> > Level0
> > ======
> > {
> > sql 'alter system enable restricted session';
> > execute script alloc_disk;
> > execute script defin_maxcorrupt_for_system;
> > backup
> > incremental level 0
> > skip offline
> > skip readonly
> > skip inaccessible
> > tag Back_Hsys_Incremental_0
> > filesperset 5
> > format '/backup/oracle/arbor-106/Level0_arborora_%t_%s_%p.bs'
> > (database);
> > execute script rel_disk;
> > sql 'alter system switch logfile';
> > execute script arch_log_current;
> > allocate channel c1 type disk;
> > backup current controlfile format
> > '/backup/oracle/arbor-106/Control_arborora_%t_%s_%p.bs';
> > release channel c1;
Received on Thu Nov 02 2000 - 19:56:00 CST