odd problem with large ext4 file system

[gaf-mNDKBlG2WHs at public.gmane.org (Jerry Feldman)]

I hope that you are doing your fsck (assume fsck.ext4) on an unmounted
file system. If you allow fsck to fix a mounted file system, you are
asking for trouble. The ONLY time fsck should be run on a mounted file
system is on root, and when it is mounted readonly.

Basically, I think that the Fedora 12 version of ext4 has been well
tested (read some prior Jarod Wilson posts).

I also assume that you are using either LVM or RAID0 striping.


On 12/24/2009 01:11 PM, Stephen Adler wrote:
> Hello all (and merry holidays!)
>
> I'm playing around with a 6 terabyte partition I put together which has=
=20
> an ext4 file system mapped onto it. In order to make sure its working=20
> well, I do periodic forced file system checks. I assume that if a file =

> system is working well, and the underlying hardware, I should be able t=
o=20
> run fsck -f as many times as I want and it should pass each time. Well,=
=20
> with this particular file system, on the first fsck -f I performed on=20
> it, a few errors were found which I fixed. I then re-ran fsck -f and it=
=20
> went through with no errors. I then ran it again, and it found a bunch =

> of errors. I ran it again, and it found more errors. All the errors its=
=20
> found I've fixed, but I'm getting a bit worried that ext4 is not ready =

> for prime time. I'm running fedora 12, with all its patches. I know ext=
4=20
> made its debut in fedora 11.
>
> Should I bite the bullet and reformat the disk as ext3? Can ext3 handle=
=20
> a 6 terabyte file system (I think it can...) Are the advantages of ext4=
=20
> such that it makes sense to try and keep the file system at ext4?
>
>  =20

--=20
Jerry Feldman <gaf-mNDKBlG2WHs at public.gmane.org>
Boston Linux and Unix
PGP key id: 537C5846
PGP Key fingerprint: 3D1B 8377 A3C0 A5F2 ECBB  CA3B 4607 4319 537C 5846