Re: In-kernel units for block numbers, etc ...

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Thu, 26 Nov 2015 16:12:34 -0800
    From:        John Nemeth <jnemeth%cue.bc.ca@localhost>
    Message-ID:  <201511270012.tAR0CYUf007052%server.CornerstoneService.ca@localhost>

  |      I would hope the answer is no, considering how much that would
  | complicate things, not to mention the slow down (i.e. doing a single
  | sector write on one drive would require an RMW cycle on the other).

I would have thought that wouldn't be supported, a mirror of 512 and 4K
devices would present as a device with 4K sectors - whatever is most
restrictive of all the components.

The 4K sector device just gets pass through, the 512 byte device gets
groups of 8 sectors for all of its I/Os.

That's why I asked about what happens if the 4K device goes away and is
replaced by a 512 byte device.

Of course, the even harder case would be a mirror of t 512 byte sector
devices, where one dies, and is replaced by a 4K sector drive.

What's more, in a few years, I'd kind of expect the demand for that to
be fairly high, as new (bigger) drives start being the norm and getting
smaller (today's sizes) drives get hard to find, and existing ones start
failing.

I have no idea how to make that work sensibly though.

kre