I've got 5 IBM x345 hyperthreading uniproc servers using acpidrv.psm
netware6sp3 w. nss3c, tcp607k and nwmem3 patches, running in a

Things work well for a while (between 2 days and 14 days) then
suddenly cpu0
peaks constantly at 100 %, causing alot of "work to do response time
server is in suspect state". Apart from that everything works as
but the cpu0 keeps peaking until the server is restartet.

In the remote manager I see alot (10-50) server processes with a
thread state.
Clicking the process shows that the process is assigned to cpu0 and
that the
work owner is cifs.nlm.

Trying to unload the cifs.nlm halts the server, which then need to becoldbooted or "debugged" down.

Any ideas?