Kevin Layer layer at known.net
Thu Jan 18 02:49:48 GMT 2007

An update to my problem.  It's still there.  However, I reinstalled XP
Pro on the box that wasn't working, since I suspected it was the
cause.  Apparently it was not.

On the FC6 side, I uninstalled (again) samba, samba-client and
system-config-samba, making sure that no files were left behind.
Well, same deal.

Recall that I have 2 XP Pro boxes, one that works and one that doesn't
work, *to the same samba server*.

The one that fails does this:

   c:\>net view \\relay
   The Server service is not started.

   More help is available by typing NET HELPMSG 2114.


Using Wireshark 0.9.44 I got a capture of the packets during from the
good and bad machines.

-------------- next part --------------

I would really appreciate it is someone could look at the pcap files
and help figure this out.



********* the original email:

  One day samba was working fine, the next it wasn't.  I didn't change
  anything on either the Fedora Core 6 machine (relay) nor the XP Pro
  machine (oob2).

  I'm running Samba 3.0.23c on a fully patched FC6 x86 box.

  I followed all the steps in "Procedure 38.1. Diagnosing Your Samba
  Server", and until I got to step 8 everything worked fine.  From the
  broken xp pro machine:

     oob2$ net view '\\relay'
     The Server service is not started.

     More help is available by typing NET HELPMSG 2114.


  (`relay' is the `fc6' machine.  `oob2' is the broken xp pro box.)

  The `server' service is running.  I stopped and started again
  countless times, and rebooted the xppro machine.

  I uninstalled samba from the fc6 machine (saving /etc/samba/smb.conf
  and restoring it afterwards... see below).  Same problem after

  I can "view" other machines (an xppro machine) on the broken
  xppro machine:

     oob2$ net view '\\dude'
     Shared resources at \\dude

     kevin's computer

     Share name  Type  Used as  Comment  

     c           Disk                    
     The command completed successfully.


  I downloaded wireshark 0.99.4 and found the point at which the good
  xppro (dude) and bad xppro (oob2) machines diverge:

     dude -> relay: NT Trans Request, NT CREATE
     relay -> dude: Nt Trans Response, NT CREATE, FID: 0x4000

     oob2 -> relay: NT Trans Request, NT CREATE
     relay -> oob2: Nt Trans Response, NT CREATE,

  So, I turned on debugging in smbd, and see this, which lines up
  perfectly with the wireshark output:

  [2006/12/27 14:01:05, 5] auth/auth_util.c:debug_unix_user_token(474)
    UNIX token of user 500
    Primary group is 100 and contains 1 supplementary groups
    Group[  0]: 100
  [2006/12/27 14:01:05, 5] smbd/uid.c:change_to_user(260)
    change_to_user uid=(500,500) gid=(0,100)
  [2006/12/27 14:01:05, 4] smbd/vfs.c:vfs_ChDir(741)
    vfs_ChDir to /tmp
  [2006/12/27 14:01:05, 5] smbd/nttrans.c:call_nt_transact_create(1117)
  [2006/12/27 14:01:05, 4] smbd/nttrans.c:nt_open_pipe(325)
    nt_open_pipe: Opening pipe srvsvc.
  [2006/12/27 14:01:05, 3] smbd/error.c:error_packet(146)
    error packet at smbd/nttrans.c(340) cmd=160 (SMBnttrans) NT_STATUS_OBJECT_NAME_NOT_FOUND

  Thinking it might have to do with cruft in /tmp, I removed everything
  that `lsof' said wasn't open by something (all but 2 files, unrelated
  to samba). 

  Here's my /etc/samba/smb.conf file:

     workgroup = FRANZ
     server string = Samba Server
     security = user
     log file = /var/log/samba/%m.log
     interfaces =
     allow hosts =
     bind interfaces only = yes
  ##### disable printing completely:
     load printers = no
     printing = bsd
     printcap name = /dev/null
     disable spoolss = yes

     comment = Home Directories
     browseable = no
     read only = no
     create mode = 0750

     comment = /tmp
     path = /tmp
     valid users = layer
     public = no
     writable = yes
     create mask = 0755


  Ideas?  I'm fresh out.  I have no idea how to debug this further.



More information about the samba mailing list