On Thu, Jan 9, 2014 at 2:08 PM, Paul Hargrove <phhargrove_at_[hidden]> wrote:
> However, only my Solaris (10/SPARC and 11/x86-64) systems have NFS-mounted
> filesystems. So, I don't have any means to ensure that the "newly active"
> code performs correctly on the BSD systems. In other words,
> opal_path_nfs() might continue to always return false on BSD systems and
> I'd not know the difference.
Nevermind that - I just recalled that test/util/opal_path_nfs.c is
Linux-specific. So, there won't be any testing coverage of the new code on
any of my Solaris or BSD systems. Nor will Mac OSX get any "real" testing.
Does anybody have suggestions (assuming I can mount an NFS dir on each of
my systems) as to how one might verify that the revised opal_path_nfs() is
*correct* on the BSDs and Solaris? If there is no validation possible,
then it might be safer for 1.7.4 to leave the no-op behavior. However, as
far as I can tell opal_path_nfs() is currently used ONLY for by shmem/mmap
to *warn* about mapping files on network filesystems. So, maybe this isn't
critical at all.
I *will* make a point of NOT
setting OMPI_MCA_shmem_mmap_enable_nfs_warning=0 as I normally do in my
testing (thought I can't remember why I added that to my testing scripts in
the first place). Then at least I might be able to tell if opal_path_nfs()
is returning false positives.
Paul H. Hargrove PHHargrove_at_[hidden]
Future Technologies Group
Computer and Data Sciences Department Tel: +1-510-495-2352
Lawrence Berkeley National Laboratory Fax: +1-510-486-6900