Extremely large file on small disk

[L. Pehrsson]

Hi All

I have a bit of a file-corruption. On a 18 GB disk, I apparently have an
282 GB file :-)

I really need the file, but can't open it from notes because it is to large.

Here is ls:
-r--r--r-- 1 user user 282755043350 feb 3 16:16 database.nsf

and du:
509364 database.nsf

and df:
/dev/md5 16176232 2586692 12767824 17% /usr

The file should not be more than around 500 MB.

Is there any way I can get linux to accept that the file is 500 MB in
size and not 282 GB? I have run e3fsck but that didn't fix it.

yours sincerely

Pehrsson

[Dances With Crows]

On Wed, 04 Feb 2004 22:14:38 +0100, L. Pehrsson staggered into the Black
Sun and said:
> I have a bit of a file-corruption. On a 18 GB disk, I apparently have
> an 282 GB file :-) I really need the file, but can't open it from
> notes because it is to large.
>

> -r--r--r-- 1 user user 282755043350 feb 3 16:16 database.nsf

> The file should not be more than around 500 MB. Is there any way I
> can get linux to accept that the file is 500 MB in size and not 282
> GB? I have run e3fsck but that didn't fix it.

e3fsck doesn't exist AFAICT. Did you mean e2fsck? If the filesystem
has been marked as "clean", you'll need to do e2fsck -f . e2fsck will
almost certainly notice something's wrong here, and take various actions
to try and fix it. Be prepared to lose the data in this file (where are
your backups?) since e2fsck may not recover everything (or anything).
Another option is to read the man page for debugfs about 3 or 4 times,
then run debugfs on the filesystem and see what you can recover.

--
Matt G|There is no Darkness in Eternity/But only Light too dim for us to see
Brainbench MVP for Linux Admin / mail: TRAP + SPAN don't belong
http://www.brainbench.com / Hire me!
-----------------------------/ http://crow202.dyndns.org/~mhgraham/resume

[Mikael Pettersson]

In article <bvrncj$2h4r$1...@news.cybercity.dk>,

disk usage < file size if the file is sparse, i.e., contains
large unwritten chunks. This can easily happen if the file is
written with random access (lseek + write or mmap), as it
typical for databases.
--
Mikael Pettersson (mi...@csd.uu.se)
Computing Science Department, Uppsala University

[Trevor Hemsley]

On Wed, 4 Feb 2004 21:14:38 UTC in comp.os.linux.hardware, "L.
Pehrsson" <use...@pehrsson.info> wrote:

> I have a bit of a file-corruption. On a 18 GB disk, I apparently have an
> 282 GB file :-)
>
> I really need the file, but can't open it from notes because it is to large.

Did you try running fixup against it? Been a long time since I worked
with Notes and I hated it even then but I think you need to start the
server then do a 'load fixup database.nsf'. Fixup has been known to
fuxup so if you _can_ cp the file elsewhere as a backup first it might
help - you'll "only" need 18GB since you should hit EOF around about
then!

--
Trevor Hemsley, Brighton, UK.
Trevor-...@dial.pipex.com

[L. Pehrsson]

Trevor Hemsley wrote:

>>I have a bit of a file-corruption. On a 18 GB disk, I apparently have an
>>282 GB file :-)
>>
>>I really need the file, but can't open it from notes because it is to large.
>
>
> Did you try running fixup against it? Been a long time since I worked
> with Notes and I hated it even then but I think you need to start the
> server then do a 'load fixup database.nsf'. Fixup has been known to
> fuxup so if you _can_ cp the file elsewhere as a backup first it might
> help - you'll "only" need 18GB since you should hit EOF around about
> then!
>

Tried compact, replicate and fixup. Can't run compact as the database is
to large. Can't run replicate as the database need a fixup. Can't run
fixup as the database grown to large :-(

[Norman Elliott]

Don't call myself an expert but I wonder if you could use dd to grab a
chunk of the data and write it out to a new file.
Maybe someone could suggest the needed options and then maybe how to
write the correct EOF marker.
Or is this just my usual lateral dizziness?
Norm