Re: Prob with softraid on sparc64
Raymond Lillard <[email protected]>
| Newsgroups | gmane.os.openbsd.sparc |
|---|---|
| Message-ID | <[email protected]> |
On 04/13/2013 06:01 PM, Stuart Henderson wrote:
> On 2013/04/14 01:27, Stuart Henderson wrote:
>> On 2013/04/13 16:50, Raymond Lillard wrote:
>>> I have 4 140GB drives and want to configure sd0 & sd1 as root-on-softraid
>>>
>>> I have done this in the past with amd64 machines, but this is not working
>>> for me on this machine.
>>
>> This is only supported in certain bootloaders, I believe only amd64/i386 at present.
>>
>
> Ah sorry I was confusing boot-from-softraid (which it looks like sparc64
> can do) with root-on-softraid (which I still believe is just amd64/i386)..
I believe you have the cases reversed. Isn't what I trying to do root-on-softraid?
I am booting from the "sd0a" partition which is a normal 4.2BSD type partition,
yet my root is on device sd4 (softraid) partition "a" (4.2BSD).
Getting back to this...
http://brycv.com/blog/2012/openbsd-sparc64-and-root-on-softraid/
The only difference between what he has done and what I am trying to
do is different machines, v120 vs v240 and four disks instead of two.
>>> First problem is:
>>> When I boot in step 3, all four disks appear in the roll call as expected
>>> in dmesg, but when I try the command "disklabel sd3", rsd3 is not found.
>>>
>>> Why ??? This seems like a bug to me.
>
> The device nodes are normally created in the installer when needed,
> I think this is to save inodes on the ramdisk.
>
>>> 1: I zero the first 10MB of all four disks with dd
>>> 2: I power cycle the box and boot form a cdrom snapshot from April 10
>>> 3: I use disklabel check that all disks have only C partitions
>
> I'm not sure offhand where softraid metadata are stored to suggest
> how much needs blanking, but obviously this must be further into the
> disk than the partition table, you probably just need to go a bit
> further with blanking. When I'm doing this I normally don't specify
> a count and just hit ^C after a few seconds...not exactly scientific
> but works for me.
Man page softraid(4) gives an example that indicates zero'ing the
first megabyte is sufficient.