[Ocfs2-tools-devel] [RFC] mkfs adaptable cluster size depending on device size

[Joel Becker]

On Fri, Jul 23, 2010 at 01:22:22PM -0500, Goldwyn Rodrigues wrote:
> On Fri, Jul 23, 2010 at 12:23 PM, Sunil Mushran
> <sunil.mushran at oracle.com> wrote:
> > mkfs.ocfs2 has a concept of fstype which allows us to choose the
> > appropriate cluster size. The calculation takes the size of the volume
> > into consideration.
> 
> Unfortunately, the users do not check the man page before creating a
> filesystem. If they do understand the architecture, it is best. Yes,
> users can be stupid.

	You're not being fair at all.  Most people don't want large
cluster sizes.  The minute I'm using 32K or 1M for each 1K email in my
home directory, I'm very very mad at the filesystem developers.
	Most people really want the efficiency of 4K clusters on every
one of their disks, no matter how big those disks are.  We have real
customers, today, that wish we could do >16TB with 4K clusters.  This is
a big feature request.
	I agree it makes fsck longer and harder.  Our job is to make
that better, not foist the pain on the users.

Joel

-- 

 One look at the From:
 understanding has blossomed
 .procmailrc grows
	- Alexander Viro

Joel Becker
Consulting Software Developer
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127