We run a 6-node OES2-Cluster for about 40 NSS-Shares and more tahn 1200 clients. More and more MAC-Clients are working with CIFS on these shares. No problem so far. It works here nearly perfect and also on several Standalone-Server out in the WAN. All Servers are in the same tree.

Our problem is, when the cluster resource goes offline or must migrate, most MAC-Clients cannot access these CIFS, although the novell-cifs-daemon is running and linux-clients can perfectly access these shares. On the clean Windows-Side (Clients without Novell-Client) cannot access the shares, too.

We are running a wins-server on Samba, a different Server under SLES10, works fine.

The last solution is, to offline the resource and online back again and so hopefully the cifs will work again. This cannot be last solution.

What I found out: (nothing special foung in the web... :-( )

when I use net view \\DNS-NAME: system error 53 or 67 (depends on the daily form)
when I use net view \\WINS-NAME: same.
when I use net view \\IP-ADDRESS: works Perfect!

I have tried lmhosts, nope. Have tried the Cluster-Node-Name: runs perfectly! But I cannot say: this resource is not on node 1, tomorrow on node 4 or so. Therefor is the IP of the resource, which is walking...

But this is only in Windows. I cannot test MAC-Clients this way, but a colleague with MAC says: nope, nothing is working.

I feel totally helpless. Always when I migrate the resource, I hear the MAC-User crying.

Always keep in mind, that smbclient from Linux works always perfectly!

Who can help? Networktraces always show: Port 445 is not working (and does not even try 139), when it works, it first uses 445, but then at once uses 139, on which novcifsd ist listening.

Frank Cordes
Netadmin and Novell-Admin