I have a oes2 cluster (linux) running with two nss volumes (1,2TB and 1,8TB). The cluster nodes are 64-bit installations with each node of 4GB RAM. The systems have a patch-level with all oes Patches installed from
May 2008. The problem is now, that sometimes folders and subfolders dissapear from the nss volumes from the view within the windows explorer (using Windows2000 and Vista Clients with latest NCP Clients...).

It seems to be a nss directory caching problem, because browsing within a ssh terminal session to the folders under /media/nss/VOLUMENAME/... still works without problems. Browsing with NRM to these folders causes the NRM to freeze and i have to restart the httpskd deamon. If I make a new directory (windows Explorer) within one of the last, highest reachable folder, the new folder disappears at that time I hit the enter button! (but the new folder is reachable via ssh...)

A umount and mount of the volume solves the problem. The effected folders are not always the same, but I don't know exactly how many folders are effected.

Turning the "LOG_LEVEL" to DEBUG in the Server Parameter Information (NRM -> NCP Manage Server) doesn't log something "interresting" to ncpserv.log.

Have anybody experience with tuning nss parameters in such a configuration described above? Or have anybody else such a problem?
Running TM's ServerProtect for Linux, folders "._NETWARE" from the volumes are excluded...nss parameters are set to "default", only OPLOCK_SUPPORT_LEVEL is set to 0....

best regards and thanks for your support