Re: [PATCH] BTRFS: Adds an option to select RAID Stripe size

[Sanidhya Solanki <jpage.lkml@gmail.com>]

On Tue, 29 Dec 2015 04:42:08 +0100
Christoph Anton Mitterer <calestyo@scientia.net> wrote:

> On Mon, 2015-12-28 at 16:43 -0500, Sanidhya Solanki wrote:
> May get even much more complicated, if reshaping (i.e. conversion from
> one chunk size to another) should get implemented as well...

That sounds like an absolutely ghastly idea. Lots of potential for
mistakes and potential data loss. I take up the offer to implement
such a feature. 
Only question is should it be in-place replacement or replace out to
another disk or storage type. Will wait for comments on that question
before implementing. 

> Sure :)
> Being able to set the RAID chunk size makes AFAIU sense for every RAID
> level, depending which IO patterns one has (i.e. even for RAID1).

I do not understand this part. Is the stripe size not independent of
the RAID type? To me it would seem that the filesystem just treats the
RAID type as a means of duplicating (if in the case of a redundant
RAID) or parallelizing data (in case of "R"AID 0), as opposed to the
Stripe length, which is a measure of each logical block for the
filesystem.

Thanks.