Re: Clarification on rename dg command

From: Kumar Madduri <ksmadduri_at_gmail.com>
Date: Thu, 10 Feb 2011 21:01:35 -0800
Message-ID: <AANLkTik8MW_Mf=hw7XxhSp01ZDkyL31EdbSyHMAE8BDo_at_mail.gmail.com>



Update on this: Like Yong and Mathias suggested, the asm diskstring parameter in renamedg worked. Thank you for the suggestions. Martin: Actually I did try to bring up 2 asm instances on the same box earlier (in 11.1.0.7 cluster) and it allowed me to do so. With 11gR2 I was not able to bring it up and let clusterware also realize the other ASM instance. May be it was the combination in R2 (grid +asm) that did not allow me to bring it up this way.
  • kumar

On Tue, Feb 8, 2011 at 12:36 PM, Martin Berger <martin.a.berger_at_gmail.com>wrote:

> Kumar,
>
> Mathias' suggestions are quite good.
> If you want to make 100% sure, you do not touch the wrong disks, you can do
> something more complex, but quite funny.
>
> start a 2nd ASM instance on the same server (!)
> for this you only need to set a different ORACLE_SID and make sure in the
> new spfile you have another DB_UNIQUE_NAME<http://download.oracle.com/docs/cd/B28359_01/server.111/b28320/initparams065.htm#REFRN10242>
> .
> Make any dedicated directory you like, and set the ASM_DISKSTRING<http://download.oracle.com/docs/cd/B28359_01/server.111/b28320/initparams011.htm#REFRN10248> of
> that specific ASM instance to this directory.
> Now create a symlink of every disk you want to manipulate into this
> particular directory.
> To be soure, check v$asm_disks in this ASM to make sure you see only the
> disks you want to manipulate.
> Now you can run renamedg as Mathias described, just skip the asm_diskstrin
> parameter.
>
> Even it's uncommon to have 2 ASM instances on one host, i's really stable
> for such tasks and prevents some issues you cah face if you try the stunt
> with only one ASM.
>
> hth,
> Martin
>
> Am 07.02.2011 um 18:48 schrieb Kumar Madduri:
>
> Hi
> I am thinking about different scenarios that need to be tested in our new
> linux infrastructure which will have 11gR2 GI and this will be supporting
> our 11gR1 databases.
> One of the scenarios is some times to refresh a development instance from
> another development instance and I had this question
>
> (a) db1 and db2 are two databases on the same box node1.
> (b) db1 uses a diskgroup called db1_dg and db2 uses a diskgroup called
> db2_dg
> (c) Perform a bcv copy from db1_dg to db2_dg (of course db2_dg and the
> associated database will be down during this time)
> (d)At this point , he headers of the files/devices that make up db2_dg will
> have information about db1_dg
> (e) I cant bring up db2_dg unless I use the rename command and rename the
> headers of db2_dg (from the ones pointing to db1_dg) to db2_dg.
> Questions are:
>
> How can I accomplish this using rename command
> renamedg phase=both dgname=db1_dg newdgname=db2_dg
> If I use this, I am afraid that I will wipe out the source.
>
>
> Thank you
> Kumar
>
>
>

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Feb 10 2011 - 23:01:35 CST

Original text of this message