Question concerning Raid Manager default partition table

[James McPherson]

>Andreas wrote:
>>James McPherson writes:
>> > Andreas Sindermann writes:
>> > > Why not simply use slice 2 which covers the whole LUN?
>> > The short answer to this is, because if you do then your configuration is 
>not 
>> > supported. The longer answer is that there are some applications (dd, 
>backups 
>> > etc) for which you need to be able to grab the _whole_ disk all at once 
>without 
>> > worrying about partitions. 
>>Hmm, I can't see why slice 2 is special in this case. It's a slice
>>that covers all blocks, just as intended (she wanted to have a slice
>>covering the complete disk). What's the difference between slice 0
>>covering all blocks and slice 2 covering all blocks?
>
I think (apart from the historical reasons) that the difference is that slice 2 
is supposed to be transparent to the OS. I don't know whether there's some hard 
coding of this in the solaris kernel or not. 

I'll be chasing this one internally for my own peace of mind (and I'll post the 
answers here too).

In the mean time, the answer(s) are that you just don't use slice 2 to put a 
filesystem on, no way, no how, because it renders your system unsupported by 
Sun; and there are applications out there which depend on slice 2 being a 
transparent whole of disk slice.

I know that's not terribly informative, but the first reason at least can be 
pretty important to people, especially if they have support contracts with Sun 
>;)


Regards,
James C. McPherson

--
Customer Technical Support Engineer      828 Pacific Highway 
Sun Microsystems Australia Pty Ltd       Gordon NSW 2072
Support Helpline: 1-800-555-786