Samba v2.2.0 strange behavior

[Andre Luis]

Dear samba admins,

    I'm experiencing some strange behavior from Samba 2.2.0. When the
clients start aplications from the share, the clients operate in a
intercalled manner - only one at a time. Also, the performance degrade
drasticaly client by client.

The snoop shows this when something is wrong:

>>>  10.200.1.35 -> solaris8     SMB C Code=0x24 Name=SMBlockingX
FileID=0x23a0 #Unlocks=0x1 #Locks=0x0 Error=0
>>>  solaris8 -> 10.200.1.35  SMB R Code=0x24 Name=SMBlockingX Error=0
>>>  10.200.1.35 -> solaris8     SMB C Code=0x2d Name=SMBopenX
FileName=??????L????????????????????????????????????????????????????????????
?????????????????????????????????????????????????????????? Error=0

and

10.200.1.35 -> solaris8     NBT C port=1025
 10.200.1.35 -> solaris8     SMB C Code=0x1a Name=SMBreadbraw FileID=0x23a6
FileOffset=0x5000 Error=0
    solaris8 -> 10.200.1.35  NBT Type=SESSION MESSAGE Length=1456
    solaris8 -> 10.200.1.35  NBT Type=Unknown Length=844
 10.200.1.35 -> solaris8     NBT C port=1025
 10.200.1.35 -> solaris8     SMB C Code=0x2d Name=SMBopenX
FileName=??????P????????????????????????????????????????????????????????????
????????????????????????????? Error=0
    solaris8 -> 10.200.1.35  SMB R Code=0x2d Name=SMBopenX FileID=0x7
Error=2
 10.200.1.35 -> solaris8     SMB C Code=0x2d Name=SMBopenX
FileName=??????N????????????????????????????????????????????????????????????
????????????????????????????? Error=0
    solaris8 -> 10.200.1.35  SMB R Code=0x2d Name=SMBopenX FileID=0x23a8
Size=0x2bd0 Error=0
 10.200.1.35 -> solaris8     SMB C Code=0x2d Name=SMBopenX
FileName=????????X??????????????????????????????????????????????????????????
????????????????????????????? Error=0
    solaris8 -> 10.200.1.35  SMB C Code=0x24 Name=SMBlockingX FileID=0x266d
#Unlocks=0x0 #Locks=0x0 Error=0
 10.200.1.35 -> solaris8     NBT C port=1028


    then the aplication locks and the file is corrupted.  Also, the log.smbd
shows this:

[2001/05/09 18:41:43, 0] smbd/service.c:make_connection(244)
  andre (200.250.91.179) couldn't find service $$orion$dados.cv
[2001/05/09 19:18:00, 0] lib/util_sock.c:set_socket_options(165)
  Failed to set socket option SO_KEEPALIVE (Error Bad file number)
[2001/05/09 19:18:00, 0] lib/util_sock.c:set_socket_options(165)
  Failed to set socket option TCP_NODELAY (Error Socket operation on
non-socket)
[[2001/05/09 19:42:052001/05/09 19:42:05, 0, 0]
lib/util_sock.c:read_socket_data(479)
  read_socket_data: recv failure for 4. Error = Connection reset by peer
] lib/util_sock.c:read_socket_data(479)
  read_socket_data: recv failure for 4. Error = Connection reset by peer
[2001/05/09 19:42:40, 0] lib/util_sock.c:set_socket_options(165)
  Failed to set socket option SO_KEEPALIVE (Error Bad file number)
[2001/05/09 19:42:40, 0] lib/util_sock.c:set_socket_options(165)
  Failed to set socket option TCP_NODELAY (Error Socket operation on
non-socket)

--------------
The plain smb.conf is:

[global]
 workgroup = TESTEPDC
 netbios name = ORION
 interfaces = hme0 10.200.1.15/255.255.255.0
 encrypt passwords = Yes
 time server = Yes
 load printers = No
 domain admin group = @adm
 add user script = useradd -g machines -d /dev/null -s /bin/false %m$
 logon script = startup.bat
 logon path =
 logon drive = U:
 logon home = \\orion\%U
 domain logons = Yes
 os level = 65
 preferred master = True
 domain master = True
 dns proxy = No
 guest account = tmp
 create mask = 0777
 directory mask = 0777

[homes]
 read only = No
 create mask = 0700
 directory mask = 0700
 browseable = No

[netlogon]
 path = /usr/local/samba/netlogon
 read only = No

[dados]
 path = /raid/export/samba/data
 read only = No
 write cache size = 262144
 share modes = No

-------------------------

This occurs in different servers and with different clients - I tested with
Solaris 8 and 7, on Sparc and X86.


    Any directions will be apreciated.

Thanks in advance.

andre at constel.com.br