David Lehman wrote:
Just curious, how accurate does this size need to be, because this
The more accurate, the better.
calculation will almost always be bigger than the actual size? At a
minimum the non-raid1 raid levels will be rounded down to be a multiple
of the chunk size. Intel(R) Matrix arrays are a bit weird in that they
will be rounded down to the nearest 1MB boundary.
Oh, great. What we need is more special cases.
Yes, I was not thrilled to discover this...
Currently, this class is
only used to represent MD raid arrays, which does not include the Matrix
(they are handled by dm-raid).
Interestingly (or depressingly depending on your viewpoint) dmraid ends
up with a different size which is neither the correct 1MB rounded size,
nor the MD chunk size rounded version of the per-device size.
DM does num_data_stripes * stripe_size * (num_data_members)
MD does per_device_size & ~(stripe_size - 1) * (num_data_members)
The metadata expects MD_SIZE & ~(1MB - 1)
MD gains support for this odd sizing in 2.6.30 [1]
--
Dan
[1] http://marc.info/?l=linux-raid&m=123733796707891&w=2
_______________________________________________
Anaconda-devel-list mailing list
Anaconda-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/anaconda-devel-list
