I see your point there- they might be trying to compensate for broken MPI distributions that don't define the necessary IO routines in mpi.h.
Removing that include *might* fix it, but I think that the core issue is that Slurm has an 'mpi.h' that does very different things and appears to be getting loaded by the compiler in favor of the OpenMPI 'mpi.h' requested by the HDF5 headers. FWIW, I patched Makefile.am, removed 'src/common' from the includes, and it builds just fine: --- a/src/plugins/acct_gather_profile/hdf5/sh5util/libsh5util_old/Makefile.am +++ b/src/plugins/acct_gather_profile/hdf5/sh5util/libsh5util_old/Makefile.am @@ -4,7 +4,7 @@ AUTOMAKE_OPTIONS = foreign -AM_CPPFLAGS = -I$(top_srcdir) -I$(top_srcdir)/src/common -I. +AM_CPPFLAGS = -I$(top_srcdir) -I. noinst_LTLIBRARIES = libsh5util_old.la Just installing it now, will see how it runs. M On Sat, Sep 12, 2015 at 8:50 PM, Ralph Castain <[email protected]> wrote: > > I think what we are saying is that HDF5 appears to be violating the MPI > standard - they are not supposed to be directly including the mpio.h > header. Someone might want to discuss that with them :-) > > > > On Sep 12, 2015, at 8:13 PM, Christopher Samuel <[email protected]> > wrote: > > > > > > On 12/09/15 18:56, dani wrote: > > > >> You shouldn't need openmpi preinstalled with recent slurm. > >> Slurm build pmi2, and then when building openmpi you supply --with-pmi > >> and point it to the pmi dir created by slurm > > > > But Michael appears to want HDF5 support in Slurm and it is that HDF5 > > code that seems to need these MPI headers. > > > > All the best, > > Chris (finally in DC for the Slurm User Group) > > -- > > Christopher Samuel Senior Systems Administrator > > VLSCI - Victorian Life Sciences Computation Initiative > > Email: [email protected] Phone: +61 (0)3 903 55545 > > http://www.vlsci.org.au/ http://twitter.com/vlsci >
