Thanks Bobby,
It works!
The last thing to do is to create the redo thread for the 2nd instance and to start it.
Now my concern is how to duplicate the database using an old backup removed from the recovery catalog, specialy when the server hosting the auxiliary instance cannot have access to the volume containing the backup.
I tried this:
- create a similar volume on the server hosting the auxiliary instance
and copy the backup in that volume
- catalog the backup in the recovery catalog
- run the duplicate command after issuing the set until time command.
Is there any way to tell RMAN which backup use to make the recovery?
Regards,
Jean
Bobby Keegan <bkeegan@(protected) :
Correct only start one instance and duplicate to it. You can use the same
initfiles just set cluster_database to false during the first instance
startup for the rman duplicate. Also the there are 2 init parameters that
are made for the duplication they are
db_file_name_convert - datafile name convert patterns and strings for
standby/clone db
log_file_name_convert - logfile name convert patterns and strings for
standby/clone db
With these parameters you can relocate the files during the duplication
process if you don't have exact filesystem layouts between the source and
target servers. As far as the listener goes it depends on how your
connecting to the source and target. The one thing I did learn is that in
order to connect remotely to a database in nomount mode (which is required
for the auxiliary database), then you must have a listener.ora file with a
SID ( not service_name) specified on the host that houses the auxiliary
database. Also the server that is connecting to it must have a TNS alias
that points to this SID and not the service_name. When the database is
started in nomount mode it registers with the listener but according to the
lsnrctl stat , its in a blocked state, hence the requirement for the
listener.ora stanza. So the remote_listener and local_listener have to be
set, so the database can start, but the rman connection doesn't use them
because of the blocked state. It took me a while to figure that one out
too.
-----Original Message-----
From: Jean-Hilaire Yapi [mailto:jhyapi@(protected)]
Sent: Monday, July 10, 2006 6:57 PM
To: Bobby Keegan; suse-oracle@(protected)
Subject: RE : RE: [suse-oracle] Help for using RMAN to duplicate a two-nodes
RA C database
Thanks Bobby,
If I understand your point, I need to start only one instance for the new
database and duplicate the database to this instance before starting the
second instance?
Do I use the same initfiles as the the database that I want to duplicate?
What about the parameters remote_listener and local_listener.
What about the listener?
regards
Jean
Bobby Keegan a ?crit :
I've had to tackle this one a few months ago. If you haven't got the RMAN
Backup & Recovery book by Robert Freeman from Oracle Press its quite good
and has saved me a few times, particularly with RAC. But on page 435 you
get this great nugget of truth. "Media recovery in a RAC environment has one
rule that you must never forget: only one node can perform recovery". So
you set cluster_database=false. Do you duplicate, make sure you have
backups of both archive log threads because rman knows to apply both archive
log threads to the single instance. Then you have to basically re-cluster
you db by recreating the redo logs for thread 2, enable thread 2, set
cluster_database to true.
Here is the post duplicate sql script I have.
alter database add logfile thread 2 group 3
('/u06/oradata/dwprd/redo03.log') size 100M reuse, group 4
('/u06/oradata/dwprd/redo04.log') size 100M reuse;
alter database enable thread 2;
alter system set cluster_database=true scope=spfile;
alter database rename global_name to dwprd.tgic.com;
Then restart instance on node 2.
Hope that helps,
Bobby
-----Original Message-----
From: Jean-Hilaire Yapi [mailto:jhyapi@(protected)]
Sent: Monday, July 10, 2006 4:13 PM
To: suse-oracle@(protected)
Subject: [suse-oracle] Help for using RMAN to duplicate a two-nodes RAC
database
Hi everyone,
I have a RAC database with instances papp1 et papp2
I would like to create on a new set of server a duplicate of this database
(with instances pprod1 and pprod2) to serve as preproduction.
We are doing hot backup of production database with RMAN and a recovery
catalog. I would like to use the duplicate feature of RMAN to make the
duplication.
The duplication of a single instance is explained in Oracle Documentation
but I could not find anything about how to duplicate a RAC database using
RMAN backups.
Any help will be appreciated since I need the new database asap.
Regards
---------------------------------
L?che-vitrine ou l?che-?cran ? Yahoo! Magasinage.
-----------------------------------------
The information contained in this email and any attachments may be
privileged and/or confidential and protected from disclosure. It
is intended solely for the use of the recipient(s) to whom it is
addressed, or an employee or agent responsible for delivering this
message to the intended recipient(s). You are hereby notified that
any dissemination, distribution, copying or other use of this
communication is strictly prohibited. If you have received this
communication in error, please notify the sender by reply email and
delete this email and all attachments from your system. Thank you.
_____
L?che-vitrine ou l?che-?cran ? Yahoo!
Magasinage.
---------------------------------
L?che-vitrine ou l?che-?cran ? Yahoo! Magasinage.