WinXP and Win2K see wrong free space on Samba 2.2.4+

Andrew Esh aesh at tricord.com
Wed May 8 16:32:04 GMT 2002


I am running tip SAMBA_2_2 code, sharing a local hard disk. When I connect to
the share with an XP client, the free space is always reported as the size of
the disk, even when the disk is not empty.

Here's the Linux free space report showing 5G out of 6G free:

[root at pluto hdd1]# df .
Filesystem           1k-blocks      Used Available Use% Mounted on
/dev/hdd1              6245396   1024708   5220688  17% /mnt/hdd1

Here's a DIR of the share from XP, showing 6G free, not 5G:

05/08/2002  05:55 PM    <DIR>          .
04/01/2002  04:56 PM    <DIR>          ..
04/03/2002  09:48 AM                 0 unix.txt
04/04/2001  12:29 PM            26,816 aaaa.sys
05/01/2002  10:39 AM            32,256 Phone1.xls
05/01/2002  10:41 AM            11,776 Copy of t100.xls
05/01/2002  10:43 AM            13,824 Copy (4) of t100.xls
05/01/2002  10:41 AM            13,824 t100.xls
05/01/2002  10:42 AM            13,824 Copy (2) of t100.xls
05/01/2002  10:53 AM            13,824 Copy (5) of t100.xls
05/08/2002  05:50 PM    <DIR>          a
05/08/2002  05:50 PM    <DIR>          b
05/08/2002  05:51 PM    <DIR>          c
05/08/2002  05:51 PM    <DIR>          d
05/08/2002  05:52 PM    <DIR>          e
05/08/2002  05:52 PM    <DIR>          f
05/08/2002  05:53 PM    <DIR>          g
05/08/2002  05:53 PM    <DIR>          h
05/08/2002  05:54 PM    <DIR>          i
05/08/2002  05:55 PM    <DIR>          j
               8 File(s)        126,144 bytes
              12 Dir(s)   6,395,265,024 bytes free

Z:\>

Here's my smb.conf:

[global]
   include = /usr/local/tsms/etc/application.appliance.smb.conf
   workgroup = LABWEB
   server string = NAS powered by Illumina(TM)
   username level = 20
   interfaces = eth0
   os level = 1 
   writable = yes
   map archive = no
   map hidden = no
   map system = no
   include = /usr/local/tsms/etc/unix_masks_and_modes
   encrypt passwords = yes
   nt acl support = yes
   change notify timeout = 7
   log file = /var/log/samba/log.%m
   max log size = 16384
   include = /usr/local/tsms/etc/netbios.aliases
   include = /usr/local/tsms/etc/replication.readonly
   security = domain
   socket options = TCP_NODELAY SO_RCVBUF=32768 SO_SNDBUF=16384
   comment = Share powered by Illumina(TM)
   fstype = NTFS
   browseable = yes
   winbind separator = +
   winbind cache time = 30
   template shell = /bin/false
   template homedir = /tmp
   winbind enum users = false
   winbind enum groups = true
   winbind uid = 1000-65000
   winbind gid = 1000-65000
   force group = friends
   ; preexec = "/etc/tfs/bin/tcpclient -p 2503 -m HEALTH_COND_ALERT -t"
   oplocks = yes
   fake oplocks = no
   kernel oplocks = no
   level2 oplocks = yes
   large readwrite = yes
   tricord attrib = yes
   debug pid = yes
   debug uid = yes
   admin users = LABWEB+Administrator

   debug level = 100
   debug timestamp = yes
   panic action = "gnome-terminal --command 'gdb /proc/%d/exe %d'"

#   username map = /etc/samba/username.map.tfs

   wins server = 64.240.27.29
   password server = IIS40

[test]
    comment = Tricord Illumina Share
    browseable = yes
    comment = Test Share
    path = /mnt/hdd1
    writable = yes
    read list = 
    valid users = "LABWEB+AndrewE"
    write list = "LABWEB+AndrewE"
;end_of_section -- Tag line do not remove

I have logs, and can repro this at will, so I can go back and get packet traces.
Since I think it's pretty easy to reproduce, I won't spam you with the 2M log
I captured.

I'll continue to characterize this. I just wanted to get a quick report out so
others can tell me what's wrong if this has been seen before.

-- 
Andrew C. Esh              mail:Andrew.Esh at tricord.com
Tricord Systems, Inc.
2905 Northwest Blvd., Suite 20   763-557-9005 (main)
Plymouth, MN 55441-2644 USA      763-551-6418 (direct)
http://www.tricord.com - Tricord Home Page




More information about the samba-technical mailing list