Re: patch for probing raid

[Mridul Jain]

hi,
Sorry I didn't mention that the patch is for
parted-1.4.24.
 
> I'm interested in this, since I was going to make a
> much more generic
> method for probing RAID devices for parted 1.6.  

This may be possible for RAID only...earlier I was
thinking of making a generic method for any device not
just RAID ,which I found is not possible because of
the  naming conventions of the various devices.They
simply don't follow any standard I guess.

> The
> one thing I'm
> wondering about is whether or not the RAID devices
> showed up in the
> /proc/partitions output, and what it looks like?

On my machine with 2 RAID disks it looks something
like this:
  major minor  #blocks  name

  48     0   35876864 rd/c0d0
  48     1    1953776 rd/c0d0p1
  48     2    4882432 rd/c0d0p2
  48     3    4882432 rd/c0d0p3
  48     4          1 rd/c0d0p4
  48     5    4882416 rd/c0d0p5
  48     6    4882416 rd/c0d0p6
  48     8  125511680 rd/c0d1
  48     9    2050032 rd/c0d1p1
  48    10     526336 rd/c0d1p2
  48    11     514048 rd/c0d1p3
  48    12          1 rd/c0d1p4
  48    13     514032 rd/c0d1p5
  48    14     264176 rd/c0d1p6
  48    15     206832 rd/c0d1p7


> Also, do you know whether RAID disks are
> automatically numbered from 0?
> So, do you always have rd/c0d0 before you have
> rd/c0d1, or is it possible
> to have rd/c0d1 and no rd/c0d0?  Similarly, are the
> controllers always
> started with c0, or can you have rd/c1d* without
> having rd/c0d*?

I will have to find out that...as in my machine
everything was listed in order..as you can see from
the /proc/partitions output above.
But I enquired with some people and they confirmed
that you cannot have rd/c1d* without rd/c0d*.
But you can have rd/c0d1 without rd/c0d0.
I need to confirm it though thro' a written doc.

> The reason I'm asking this is that there are
> potentially a _lot_ of
> RAID devices with rd/*, ida/*, cciss/*, ideraid/*
> and I don't want
> to probe all of them if we can bail out early (e.g.
> don't probe any
> controllers c1-c7 if c0 doesn't exist, and possibly
> don't probe any
> disks d1-d31 if d0 doesn't exist).

That is a good idea for optimization.
I will have to try on my other machines and find out.
I will also confirm the things I enquired with people
thro'some written doc or tutorial of RAID and then may
be write this type of optimization.

cheers again!
Mridul.

=====
Inspired by GNU
www.gnu.org

__________________________________________________
Do You Yahoo!?
Send your FREE holiday greetings online!
http://greetings.yahoo.com