Prev: Patch availability change
Next: Peculiar change in available patches
From: mannu on 11 Feb 2010 00:07
Hello,
I am not able to understand what is difference between
"/dev/did" and "/dev/global" devices in sun cluster.

Can some one explain in simple terms?

I have a sun cluster where I am seeing that
"/dev/did" is populated where as "/dev/global"
is not. Is this normal (or) abnormal?

Also should oracle RAC use "/dev/did" devices or "/dev/global"
when running on sun cluster.

Thanks and Regards,
Mannu.

From: mannu on 12 Feb 2010 02:48
On Feb 11, 10:07 am, mannu <manishsingh...(a)gmail.com> wrote:
> Hello,
>          I am not able to understand what is difference between
> "/dev/did" and "/dev/global" devices in sun cluster.
>
> Can some one explain in simple terms?
>
> I have a sun cluster where I am seeing that
> "/dev/did" is populated where as "/dev/global"
> is not. Is this normal (or) abnormal?
>
> Also should oracle RAC use "/dev/did" devices or "/dev/global"
> when running on sun cluster.
>
> Thanks and Regards,
> Mannu.

I saw oracle rac document for solaris from sun, but
it mentions to use "/dev/did". I am not able to understand the reason
for this.
Can some one please explain difference between
"/dev/did" and "/dev/global"?
From: wisnios on 12 Feb 2010 08:31
Hello Mannu.

> I have a sun cluster where I am seeing that
> "/dev/did" is populated where as "/dev/global"
> is not. Is this normal (or) abnormal?
Yes, it is completely normal.

AFAIK, '/dev/did' device name space is synchronized through the
cluster nodes to make sure
the specific device is the one and only, for example to choose the
quorum device correctly.
So, the IO operations of specified 'did' device are under control of
its node.

On the contrary '/dev/global' is under control of primary cluster
node.
So, any request to use the 'global' resources is going to and through
the primary node.

> Also should oracle RAC use "/dev/did" devices or "/dev/global"
> when running on sun cluster.
'/dev/did' is the right choice. Oracle RAC takes care of its I/O
operations on every node by itself
and write directly to its disks.
Usage of HA agent makes a defference. When agent takes control of
oracle's instances it also
should use the 'global' disk device.

It's very general comparison, but hope you understand.
/dev/did - global naming, local access
/dev/global - global scope, primary node control

Anybody, please correct me if I am wrong.

Regards.
Marcin
 | 
Pages: 1
Prev: Patch availability change
Next: Peculiar change in available patches