Differences between "df -m" and summarized "du -sm *" in FC4
Todd Zullinger
tmz at pobox.com
Wed Jun 14 01:52:56 UTC 2006
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Cameron Simpson wrote:
> On 13Jun2006 11:41, Tony Nelson <tonynelson at georgeanelson.com> wrote:
> | At 9:05 AM +0100 6/13/06, Paul Howarth wrote:
> |>On Tue, 2006-06-13 at 08:13 +0200, Joachim Backes wrote:
> |>> having an FC4 server and problems with /tmp:
> |>> df -m says, that /tmp 100% full. But if going to /tmp, calling "du -sm
> |>>*" and summarizing all sizes
> |>> gives a huge difference against the df command (the sum is about 10% of
> |>>the df-value).
> |>>
> |>> Remark: there are no .-files in /tmp so "du -sm *" fits all files in /tmp.
> |>
> |>You probably have a process running that has a a large open file in /tmp
> |>that has been deleted. The disk space is not recovered for use by other
> |>processes until the file is closed.
> |>
> |>Deleting an open temp file is a fairly common technique - it prevents
> |>other processes getting any access to the file.
> |
> | Could 10% be the root reservation?
>
> Almost certainly.
I believe that Joachim is saying that the size reported by du is only
10% of the size reported by df. This would mean the reserved blocks
percentage would have to be 90%. That doesn't seem likely. Though
stranger things do happen. :)
- --
Todd OpenPGP -> KeyID: 0xD654075A | URL: www.pobox.com/~tmz/pgp
======================================================================
Go the extra mile. It makes your boss look like an incompetent
slacker.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: When crypto is outlawed bayl bhgynjf jvyy unir cevinpl.
iG0EARECAC0FAkSPa/cmGGh0dHA6Ly93d3cucG9ib3guY29tL350bXovcGdwL3Rt
ei5hc2MACgkQuv+09NZUB1pu3gCeL9BakZurebokf8tu7j/tkmcUVAEAnjY0uSgD
3NeaCBn16s+PfAyp+Q8z
=it9M
-----END PGP SIGNATURE-----
More information about the fedora-list
mailing list