I experiment a very strange slow performance issue with my cluster configuration.

I am currently running a 3 clustered-node Linux Oes2Sp2 servers.

Clustered services :
- iPrint
- Nss files
- Dhcp

Non clustered services running on each of the 3 nodes :
- slp
- dns

The issue is specificaly with Nss volumes.

I got 3 Nss reserves configured with a NCP server name (--ncpservername in the cluster script) different from the real object
reserve name :
Reserve03_Server (ncpservername : NSS01)
Reserve04_Server (ncpservername : NSS02)
Reserve05_Server (ncpservername : NSS03)

Please consider that NSS01 and NSS02 is hold by the same physical server (for example Srv-Oes01). So That,
I am able to see the same volume if I connect using \\NSS01 or \\NSS02

Here is the behavior I experiment :

Using a test winXP workstation with Novell Client installed (4.91 SP5)

If I try to copy a small file (~10MB) to :

- \\NSS01\Directory, then it is very slow to copy (windows copy screen shows more 15 min...)
- \\NSS02\Directory, then it is "normal" (less than 1 min...)
- \\NSS03\AnotherDirectory, then it is also very slow

The most amazing thing is that the behavior is not the same on all of my workstation !!!!

I can perform the same test on 2 different workstation : 1 will work perfectly, the other one won't -_-

I have tested to apply same configuration to both Novell client etc... and nothing : behavior is the same 1 will work, the other won't.

And last amazing thing :

If I try this time to copy this same file to the same volume BUT using server name instead of NCP server name (e.g \\Srv-Oes01\Directory instead of \\NSS01\Directory)
from the workstation where the copy was very slow, then it also works perfectly !

Any Ideas would be appreciated...