collision(s): /usr/bin/mpiexec Searching all installed packages for file collisions... Press Ctrl-C to Stop sys-cluster/openmpi-1.2.8 /usr/bin/mpiexec Package 'sys-cluster/mpiexec-0.82' NOT merged due to file collisions. If necessary, refer to your elog messages for the whole content of the above message.
sys-cluster/torque installs /usr/bin/mpiexec as well. At least the one from torque is stupid what everybody on torque-users list knows and that is why everybody overwrites the mpiexec by the "OSC mpiexec" which is in sys-cluster-mpiexec. So, at least torque should depend on mpixec and maybe openmpi as well? I think compatibile packages are listed on the homepage of OSC mpiexec.
(In reply to comment #1) > sys-cluster/torque installs /usr/bin/mpiexec as well. At least the one from > torque is stupid what everybody on torque-users list knows and that is why > everybody overwrites the mpiexec by the "OSC mpiexec" which is in > sys-cluster-mpiexec. > > So, at least torque should depend on mpixec and maybe openmpi as well? > I think compatibile packages are listed on the homepage of OSC mpiexec. > Uhhh, which version of torque? I haven't seen openmpi and torque conflict since I started bumping them. Also, torque doesn't depend on any implementation of mpi or having mpiexec around. Tt's just a batch system, albeit one that is typically used in conjunction with mpi.
Oh, sorry, looks like that was fixed in the meantime since I found that. ;-) Anyway, some remnant exists: # equery files torque | grep mpi /usr/share/doc/torque-2.3.1/README.mpiexec.bz2 # equery belongs mpiexec [ Searching for file(s) mpiexec in *... ] sys-cluster/mpich2-1.0.6 (/usr/bin/mpiexec -> mpiexec.py) sys-cluster/mpiexec-0.82-r1 (/usr/bin/mpiexec) #
# Copyright 1999-2009 Gentoo Foundation # Distributed under the terms of the GNU General Public License v2 # $Header: /var/cvsroot/gentoo-x86/sys-cluster/mpich2/mpich2-1.0.8.ebuild,v 1.2 2009/01/19 00:27:59 mr_bones_ Exp $ EAPI=1 inherit python eutils fortran DESCRIPTION="MPICH2 - A portable MPI implementation" HOMEPAGE="http://www-unix.mcs.anl.gov/mpi/mpich2" SRC_URI="http://www.mcs.anl.gov/research/projects/mpich2/downloads/tarballs/${PV}/${P}.tar.gz" LICENSE="as-is" SLOT="0" KEYWORDS="~amd64 ~ppc ~ppc64 ~x86" IUSE="+cxx debug doc fortran pvfs2 threads romio mpi-threads" COMMON_DEPEND="dev-lang/perl >=dev-lang/python-2.3 romio? ( net-fs/nfs-utils ) pvfs2? ( >=sys-cluster/pvfs2-2.7.0 ) dev-libs/libaio !media-sound/mpd !sys-cluster/mpiexec Why is mpiexec blocker?
Please consider using osc-mpiexec-0.83. This package is functionally the same as mpiexec, but moves the binary and manpage to be prefixed with 'osc-' to avoid the blocking issues.
(In reply to comment #5) > Please consider using osc-mpiexec-0.83. This package is functionally the same > as mpiexec, but moves the binary and manpage to be prefixed with 'osc-' to > avoid the blocking issues. > Justin, I thought I already asked you once but I don't remember an answer, I am not going to install the renamed binary osc-mpiexec untill you tell me how is qsub from torque supposed to know that I want it to execute /usr/bin/osc-mpiexec instead of /usr/bin/mpiexec?
(In reply to comment #6) > (In reply to comment #5) > > Please consider using osc-mpiexec-0.83. This package is functionally the > > same as mpiexec, but moves the binary and manpage to be prefixed with > > 'osc-' to avoid the blocking issues. > > Justin, I thought I already asked you once but I don't remember an answer ... Indeed I did: bug #165671#c9 ;-)
sys-cluster/mpiexec was removed from the tree and replaced with sys-cluster/osc-mpiexec.
