smb og smbfs-error; Connection reset by peer

[Per Andreas Buer]

Hello,


I have a samba server serving som e shares to a heavily loaded web-
server. There a four shares fg 60Gb each. Approx. once a day, one of
the shares die (it tends to be the one under the harddest strain).


The server log says:

[2000/05/01 17:59:52, 0] smbd/service.c:make_connection(336)
  samba logged in as admin user (root privileges)
[2000/05/01 17:59:52, 1] smbd/service.c:make_connection(550)
  www (10.1.1.1) connect to service data3 as user samba (uid=0, gid=1001) (pid 29024)
[2000/05/01 17:59:52, 1] smbd/service.c:close_cnum(583)
  www (10.1.1.1) closed connection to service data3
[2000/05/01 17:59:52, 0] lib/util_sock.c:read_socket_data(477)
  read_socket_data: recv failure for 4. Error = Connection reset by peer
[2000/05/01 17:59:52, 1] smbd/service.c:close_cnum(583)
  www (10.1.1.1) closed connection to service data3



And on the client:

May  1 17:59:51 www kernel: smb_get_length: recv error = 512
May  1 17:59:51 www kernel: smb_request: result -512, setting invalid
May  1 17:59:51 www kernel: smb_retry: caught signal
May  1 17:59:51 www last message repeated 2 times


Now the share is dead. An mount/umount usually gets the share started
again. 




The socket options I use are
   socket options = IPTOS_LOWDELAY TCP_NODELAY SO_SNDBUF=4096 SO_RCVBUF=4096

Each share looks like this:
[data1]
   comment = Data 1
   browseable = yes
   read only = no
   writable = yes
   public = no
   path=/data/1

   create mask = 0775
   directory mask = 1775


I don't use syslog to log this. 


Any hint or help would be very much appreciated. 


-- 
Per Andreas Buer                          We've had enough edutainment.
                                                We need funderstanding.