Thanks Skip, That's the part I was missing. I thought that the current namespace is preserved after cpu command on the remote. But then I realize that it would be difficult to remap the bin namespaces to a different cpu architecture.
I suppose that I can always bind over what I want from /mnt/term or customize the profile script to do it automatically in cpu case. Chris On Oct 6, 2016, at 12:14 AM, Skip Tavakkolian <9...@9netics.com> wrote: >> I was under the impression that the namespace should come from the client. > > perhaps it's a confusion over cwd when you cpu to another machine? > > supermic% pwd > /usr/fst > supermic% cpu -h rpi > rpi% pwd > /usr/fst > rpi% > > since typically everything is served by fs, it is all the same > content. the local namespace is exported by the local cpu and the > far-end cpu mounts it on /mnt/term. > > e.g. > > rpi% devsysname='/dev/sysname' for (i in `{seq 3}) { > echo $devsysname ' = ' `{cat $devsysname} > devsysname='/mnt/term'^$devsysname > } > /dev/sysname = rpi > /mnt/term/dev/sysname = supermic > /mnt/term/mnt/term/dev/sysname = dell > rpi% > >