Bernard, this has to be an environmental problem if it has always
failed to work for you. Take a look at your paging space. Is it twice
the real memory on your system? I would suspect that the ipmap
processes for your r/o users are dying, and insufficient paging
space can cause this. And you won't see the shortage with lsps -a
because it happens very fast. There should be messages in the
nv6000.log(s) and in nettl as well.

Cordially,

Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking