2011-04-09 21:58:53

by Santiago Garcia Mantinan

[permalink] [raw]
Subject: flex_bg with tune2fs posible?

Hi!

I had converted an ext3 filesystem to ext4 long ago but comparing it's
features with the ones found on a newly formatted ext4 filesystem I found
that I had these features missing:
flex_bg huge_file dir_nlink extra_isize


Current doc at ext4.wiki.kernel.org says: "There are additional ext4
features, such as flex_bg and > 16TB filesystem support that can only
be enabled at format time via mke2fs."

I did activate all these 4 without any problem (2.6.38 with 1.41.12
e2fsprogs) is the doc old and doesn't reflect that this is now possible?

Or have I done something I'll regret?

So far the filesystem seems ok (e2fsck doesn't detect any problems) and
files seem ok as well.

Regards...
--
Manty/BestiaTester -> http://manty.net


2011-04-09 23:01:49

by Andreas Dilger

[permalink] [raw]
Subject: Re: flex_bg with tune2fs posible?

On 2011-04-09, at 3:50 PM, Santiago Garcia Mantinan <[email protected]> wrote:
> I had converted an ext3 filesystem to ext4 long ago but comparing it's
> features with the ones found on a newly formatted ext4 filesystem I found
> that I had these features missing:
> flex_bg huge_file dir_nlink extra_isize

> Current doc at ext4.wiki.kernel.org says: "There are additional ext4
> features, such as flex_bg and > 16TB filesystem support that can only
> be enabled at format time via mke2fs."
>
> I did activate all these 4 without any problem (2.6.38 with 1.41.12
> e2fsprogs) is the doc old and doesn't reflect that this is now possible?
>
> Or have I done something I'll regret?

Technically, flex_bg can be activated after the filesystem is formatted, but it doesn't actually provide the benefit as if enabled at format time.

The dir_nlink and huge_file features can safely be enabled on an existing filesystem, allowing > 64000 subdirs and > 2 TB files, respectively.

The extra_isize feature will not itself do anything if the filesystem wasn't formatted with large inodes (>128 bytes), though it is possible this was done even for ext3, depending on the version of e2fsprogs used.

> So far the filesystem seems ok (e2fsck doesn't detect any problems) and
> files seem ok as well.