on the internal side of the core it is running dhcp which the mds get there address from and any orbiters but the storage boxes have the ip address assigned by me to 192.168.80.2 and 192.168.80.3 for the second box this should not be a problem just because dhcp is running dosnt mean you cant hard code an ip address to a computer, After a lease is up on an ip address sometimes a new address can be assigned this is eliminated by me manualy setting the ip
...
Actually configuring ip-addresses manually that are within the DHCP range is a huge problem.
From /etc/dhcp3/dhcpd.conf:
subnet 192.168.80.0 netmask 255.255.255.0 {
next-server 192.168.80.1;
filename "/tftpboot/pxelinux.0";
option pxelinux.reboottime = 30;
default-lease-time 86400;
max-lease-time 604800;
pool {
allow unknown-clients;
range 192.168.80.129 192.168.80.253;
}
}
The lease pool is .129-.253, so it's ok to have static IP machines in .2-.128. I've had no problems having a static IP server (.20) with a NFS share. Simply follow the instructions on the wiki to manually add the servers and shares.
One note in favour of using DHCP is that once LinuxMCE assigns an IP to a file server, it binds the MAC address to a fixed IP address (see the group section in dhcpd.conf), so for all practicable purposes the server's address is static (at least until you change NIC cards).
I suspect that Collin is on the right track suspecting that dual homing the servers is the culprit. It might be time to dust off wireshark to see exactly why the connections to the CIFS shares are being rejected.
HTH,
Roy