Re: 10gR2 RAC: Mirroring OCR and voting disks

From: DA Morgan <damorgan_at_psoug.org>
Date: Wed, 14 Jun 2006 12:27:47 -0700
Message-ID: <1150313271.513508@bubbleator.drizzle.com>

Andreas Sheriff wrote:
> "DA Morgan" <damorgan_at_psoug.org> wrote in message
> news:1150304122.71218_at_bubbleator.drizzle.com...

>> Andreas Sheriff wrote:
>>> Where does it state in the manual that the majority must survive?
>>> I love how they leave these little tidbits out.  'Makes it exciting!
>> I don't know that it states it explicitly: Perhaps it does. But that
>> was my assumption based on the fact that an uneven number of resources
>> was required. With control files any one can survive and thus two
>> control files or three it makes no difference. If an uneven number is
>> required then it is consistent that a vote takes place and that a tie
>> is not allowed.
>> -- 
>> Daniel A. Morgan
>> University of Washington
>> damorgan_at_x.washington.edu
>> (replace x with u to respond)
>> Puget Sound Oracle Users Group
>> www.psoug.org

>
> Just a hypothesis here...
>
> Could it be that if a majority of the votedisks survive then nodes seeing a
> majority share in common at least one of the votedisks.
> Nodes that see 1/2 the vote disks have a 50/50 chance of getting it right
> that the other nodes share at least one of the votedisks.
> Finally, nodes that see less than 1/2 have an even lower chance of all
> seeing a shared votedisk.
>
> Hence, a majority of the votedisks have to survive for a proper quorum to
> exist.
>
> Couldn't they just have said this?

I'll forward your comment to the person that writes the docs.

Because I agree.

-- 
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu
(replace x with u to respond)
Puget Sound Oracle Users Group
www.psoug.org

Received on Wed Jun 14 2006 - 14:27:47 CDT