Home » Server Options » RAC & Failsafe » storage migration on existing RAC nodes using OCFS2 (11gr1, linux64, ocfs2 v1.2)
storage migration on existing RAC nodes using OCFS2 [message #632656] Tue, 03 February 2015 22:41 Go to next message
oraclewannabe
Messages: 26
Registered: January 2008
Junior Member
Hi all,

I am involved in a project whereby there are a rac cluster with 2 nodes. please forgive me if i make any silly suggestions as i am still new to migration.

2) Both nodes are connected to shared disk using OCFS2.

3) The physical san storage is to be replace with another san storage from another vendor.

4) Oracle datafiles are stored in the OCFS2 shared disks.
Grid/Oracle binaries are stored on local disk

5) We are not allowed to setup new nodes to connect to the new storage. (it will be a pure storage change, nodes will remain)

6) New storage to be setup as OCFS2 filesystem as well as Customer is not willing to switch to ASM.

==================================================

q1) Are we able to create to create multiple database instances in a RAC environment ?

q2) I am thinking along this line ->

-- assuming the new storage volumes are already setup , formatted as ocfs2 and visible to the nodes


Method 1
1) create a new rac database instance (newdb) which will be pointing to the new storage location

2) shutdown the existing database instance (olddb),do a cold backup

3) restore the backup onto the new instance (newdb) using normal restore or duplicate/clone

=========================================
Method 2

-- assuming the new storage volumes are already setup , formatted as ocfs2 and visible to the nodes

1) Shutdown the current database

2) physically copy all the datafiles on the ocfs2 shared disk to the new storage mountpoints

3) unmount the current mountpoints on the old sorage (e.g. /u02 /u03)

4) mount the new storage (with the datafiles copied over earlier) as the old mountpoints /u02 /u03)

5) start the database

-- will this encounter any issue ?

================================================

q1) What should be the correct way to do this migration (without changing nodes, just storage) on a RAC 2-node environment with OCFS filesystem ?


q2) What if the grid/oracle binaries are also on the shared disk ? Do i have to backup and reinstall the whole grid/oracle installation on the new storage , before restoring the backup (since the old storage will be physically removed from the setup)

please advise =~~)

Regards,
Noob
Re: storage migration on existing RAC nodes using OCFS2 [message #632669 is a reply to message #632656] Wed, 04 February 2015 02:25 Go to previous message
John Watson
Messages: 8022
Registered: January 2010
Location: Global Village
Senior Member
You say that you are using 11gR1 and OCFS2 v1.2. That is a certified combination - but only just. At the very least, you will have to replace the old OCFS2 with v1.4 before you can upgrade your clusterware to GI, and then upgrade your database.
What attempts have you made to move your client off clusterware and on to GI? Even if they do have a valid reason for wanting to use OCFS2 (which I would say is insanity) there can be no valid reason for remaining on the old clusterware. GI is infinitely superior, and the client will never have a better opportunity to make the change.
Can you explain the background to this situation?
Previous Topic: Should we delete grid infrastructure logs if it gets too big
Next Topic: Shutdown asm instance options
Goto Forum:
  


Current Time: Tue Oct 15 16:44:13 CDT 2019