Samba and Quota (Linux)

Andreas Abach Abach at
Thu Oct 14 07:15:44 GMT 1999

On Thu, 14 Oct 1999 02:27:16 +1000, Frank van Maarseveen wrote:

>On Thu, Oct 14, 1999 at 02:02:23AM +1000, Celso Kopp Webber wrote:
>> Whenever a Windows client tries to
>> write a file that goes beyond his/her hard limit quota, the file seems to be
>> created with the correct size, but it gets truncated on the server's
>> filesystem.
>Unless there's no error code this is not a bug. When a file is opened, seeked

I think this is a bug. If the Window-user gets an error-message, but
can see a file written to disk with the correct size(beyond his/her
hard limit quota). What shall he think ?
I know what he will think. He won't care about the error-message and
will be happy about his file. So he will collect damaged files till he
opens them again.
So my suggestion would be to erase the file again(by samba), if it cant
be written completely to disk.

>far beyond EOF and a few bytes are written then the file appears to be big
>according to ls -l but in reality contains so called phantom blocks: it
>is a sparse file. So, size may disagree with the actual number of disk blocks
>> I have also another situation here concerning quotas: every client maps its
>> home directory to a single server. But some users have their home dir on an
>> NFS mounted directory. Although a "quota -u login" reports his quota
>> correctly on the NFS server, Samba does report the whole NFS mounted file
>> system size and user space.
>So "quota -u login" on the NFS client reports the wrong number when NFS
>mounted fs are involved? then the bug is either in quota or in rpc.quotad
>on the server.

Dipl.-Ing. Andreas Abach *KT Lehrstuhl fuer Kunststofftechnik Uni Erlangen
 Tel. +49-9131-85-297-06 * Fax.: -09 * email: Abach at
Am Weiselgarten 9 * D-91058 Erlangen * url:
PGPID 0x5672B1F8 F-Print FD50 85C0 7580 5467 65E7 3BE8 58BC 65F4 5672 B1F8

More information about the samba-technical mailing list