status on 'tune2fs -I 256' ?

status on 'tune2fs -I 256' ?

[Doug Hunley]

I recently converted my '/' filesytem to ext4 from ext3 using:
tune2fs  -O extents,uninit_bg,dir_index /dev/md3

I did *not* use '-I 256' as I'd read several reports of this causing
corruption. However, I've just checked the ext4.txt as shipped with
2.6.29 and it quite clearly states:
If the filesystem was created with 128 byte inodes, it can be
    converted to use 256 byte for greater efficiency via:

        # tune2fs -I 256 /dev/hda1

Is this now safe to do? Or should the documentation be updated to
reflect the current corruption issue? Would I be ok to run 'tune2fs -I
256 /dev/md3' (followed by a forced fsck)?

Thanks

-- 
Douglas J Hunley - doug.hunley@gmail.com
http://douglasjhunley.com
Twitter: @hunleyd

Re: status on 'tune2fs -I 256' ?

[Eric Sandeen]

Doug Hunley wrote:
> I recently converted my '/' filesytem to ext4 from ext3 using:
> tune2fs  -O extents,uninit_bg,dir_index /dev/md3
> 
> I did *not* use '-I 256' as I'd read several reports of this causing
> corruption. However, I've just checked the ext4.txt as shipped with
> 2.6.29 and it quite clearly states:
> If the filesystem was created with 128 byte inodes, it can be
>     converted to use 256 byte for greater efficiency via:
> 
>         # tune2fs -I 256 /dev/hda1
> 
> Is this now safe to do? Or should the documentation be updated to
> reflect the current corruption issue? Would I be ok to run 'tune2fs -I
> 256 /dev/md3' (followed by a forced fsck)?

AFAIK it still has dangerous corners... We should probably update the
ext4.txt, and TBH I'd rather disable the functionality in e2fsprogs
until it's safe.

(the case I ran into was when there was not actually enough space to
double the size of all the inodes - it did not know this ahead of time
and it did not fail gracefully).

-Eric