Naming of Filesystems for a DB Clon

[ Edited ]


I've successfuly cloned a DB (AUMT) to DB KLON. Since I've to be able to unmount/mount the Filesystems in a Script I would prefer to

not have a Timestamp in the generated Filesystem Name.

Is there a way to configure the bold Part (see below) of the Filesystem Name ?

oracle@krebs :~ $ df -ef | grep KLON

Filesystem                           size       used      avail   used%  Mounted

n11a:/vol/SnapManager_20120927092549127_v11_s041_004/AUMT/u01        10G       1.5G       8.5G   15%   /KLON/u01

n11a:/vol/SnapManager_20120927092549127_v11_s041_004/AUMT/u02        10G       1.5G       8.5G   15%   /KLON/u02

Thanks Eric

Re: Naming of Filesystems for a DB Clon

I don't fully understand what you are trying to do,

if you simply want to unmount/remount the filesystems from the host but not destroy the clone, then you can simply issue umount /KLON/u01 /KLON/u02, and mount them the same way.

if you are going to destroy the clone, then its advisable that this is mounted and the database open, as SMO will:

a) gather the list of datafiles,redo logs and control files from the clone

b) stop the database

c) remove the files

d) umount filesystems

e) remove the entries from /etc/fstab

f) remove the flexclone and update the repository.

There are only a few situations where renaming the flexclone vol is necessary, but this is not one of them.

Re: Naming of Filesystems for a DB Clon

We are runing the Clone DB in a Veritas Cluster Environment (3 Node Cluster) where Veritas should be able to handle the failover of the DB and all it's resources.
For the Parant DB (in this Case AUMT) the Failover of the Mounts and Volumes is done in a Script writen by our System Admin. Here is a part from the script
we're using for the Parent DB AUMT :

mount -F nfs -o ${MOUNTOPTION_DATA} ${NFSSERVER}:/vol/v11_s041_004/AUMT/u01 /${INSTANCE}/u01
    mount -F nfs -o ${MOUNTOPTION_DATA} ${NFSSERVER}:/vol/v11_s041_004/AUMT/u02 /${INSTANCE}/u02

We would need a similar Script for our KLON DB. It would simplify the script if the Flexclon Vol Name would be the same after every Clon Operation.

Since you mention that it is possible to rename the flexclon vol could you/someone elaborate on this process ?

Re: Naming of Filesystems for a DB Clon

Try a different approach. Instead of renaming the flexclone which will break some smo operations try to add a post clone script to save the mount point and nfs flexvol name to a file.

If you use a filesystem under vcs control to keep that file then you only need to read its contents to mount the filesystem when you online the package.

You could even go one step further and include hares modify commands to the post clone script so that vcs can do it all automatically. And still monitor the resource.

Re: Naming of Filesystems for a DB Clon

Yes, a post Clone script seems the way to go. I was hoping that there is a Configuration Parameter to tell SMO how to name the Volumes to avoid future scripting.

Thanks for your help !