Home | Docs | Forums | Lists | Bugs | Planet | Store | GMN | Get Gentoo!
View Bug Activity | Format For Printing | XML | Clone This Bug
We'd like to make openmpi the default implementation in virtual/mpi, the sooner the better as mpich is old enough to just cause problems. Ebuild has a test suite. Thanks.
With USE="fortran threads" /usr/bin/install -c -m 644 openmpi-mca-params.conf /var/tmp/portage/sys-cluster/openmpi-1.2.6/image//etc/openmpi/openmpi-mca-params.conf /usr/bin/install: cannot create regular file `/var/tmp/portage/sys-cluster/openmpi-1.2.6/image//etc/openmpi/openmpi-mca-params.conf': No such file or directory make[3]: *** [install-data-local] Error 1 make[3]: *** Waiting for unfinished jobs....
I cannot confirm this behavior on ppc64. waiting for another person to confirm before marking stable.
(In reply to comment #1) > With USE="fortran threads" > > > /usr/bin/install -c -m 644 openmpi-mca-params.conf > /var/tmp/portage/sys-cluster/openmpi-1.2.6/image//etc/openmpi/openmpi-mca-params.conf > /usr/bin/install: cannot create regular file > `/var/tmp/portage/sys-cluster/openmpi-1.2.6/image//etc/openmpi/openmpi-mca-params.conf': > No such file or directory > make[3]: *** [install-data-local] Error 1 > make[3]: *** Waiting for unfinished jobs.... > I'm also unable to replicate this, can you provide any more information?
Could be a parallel install problem, what's MAKEOPTS?
(In reply to comment #4) > Could be a parallel install problem, what's MAKEOPTS? I am not able to reproduce this anymore.
cool. ppc64 stable then
I'm getting these test failures on sparc(USE="-threads"): --> Testing atomic_cmpset atomic_cmpset: atomic_cmpset.c:198: main: Assertion `volptr == newptr' failed. ./run_tests: line 8: 21517 Aborted $* $threads - 1 threads: Failed atomic_cmpset: atomic_cmpset.c:198: main: Assertion `volptr == newptr' failed. ./run_tests: line 8: 21518 Aborted $* $threads - 2 threads: Failed atomic_cmpset: atomic_cmpset.c:198: main: Assertion `volptr == newptr' failed. ./run_tests: line 8: 21519 Aborted $* $threads - 4 threads: Failed atomic_cmpset: atomic_cmpset.c:198: main: Assertion `volptr == newptr' failed. ./run_tests: line 8: 21520 Aborted $* $threads - 5 threads: Failed atomic_cmpset: atomic_cmpset.c:198: main: Assertion `volptr == newptr' failed. ./run_tests: line 8: 21521 Aborted $* $threads - 8 threads: Failed FAIL: atomic_cmpset --> Testing atomic_cmpset_noinline - 1 threads: Passed - 2 threads: Passed - 4 threads: Passed - 5 threads: Passed - 8 threads: Passed PASS: atomic_cmpset_noinline ======================================================== 1 of 8 tests failed Please report to http://www.open-mpi.org/community/help/ ======================================================== If I ignore them, non-mpi and mpi programms seem to work fine... # emerge --info Portage 2.1.4.4 (default-linux/sparc/sparc64/2007.0/server, gcc-4.1.2, glibc-2.6.1-r0, 2.6.24-gentoo-r8 sparc64) ================================================================= System uname: 2.6.24-gentoo-r8 sparc64 sun4u Timestamp of tree: Sat, 31 May 2008 07:34:01 +0000 app-shells/bash: 3.2_p33 dev-lang/python: 2.4.4-r13 dev-python/pycrypto: 2.0.1-r6 sys-apps/baselayout: 1.12.11.1 sys-apps/sandbox: 1.2.18.1-r2 sys-devel/autoconf: 2.13, 2.61-r1 sys-devel/automake: 1.10.1 sys-devel/binutils: 2.18-r1 sys-devel/gcc-config: 1.4.0-r4 sys-devel/libtool: 1.5.26 virtual/os-headers: 2.6.24 ACCEPT_KEYWORDS="sparc" CBUILD="sparc-unknown-linux-gnu" CFLAGS="-mcpu=ultrasparc -mtune=ultrasparc -mvis -Wa,-Av8plusa -frename-registers -O2 -pipe" CHOST="sparc-unknown-linux-gnu" CONFIG_PROTECT="/etc" CONFIG_PROTECT_MASK="/etc/env.d /etc/gconf /etc/revdep-rebuild /etc/terminfo /etc/udev/rules.d" CXXFLAGS="-mcpu=ultrasparc -mtune=ultrasparc -mvis -Wa,-Av8plusa -frename-registers -O2 -pipe" DISTDIR="/tmp/distfiles" FEATURES="collision-protect distlocks metadata-transfer parallel-fetch sandbox strict test unmerge-orphans userfetch userpriv usersandbox" GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/linux/distributions/gentoo" LANG="de_DE.UTF-8" LDFLAGS="-Wl,-O1" LINGUAS="en de" MAKEOPTS="-j17" PKGDIR="/usr/portage/packages" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="caps cli cracklib cups dri extensions fortran gdbm gpm hpn iconv ipv6 isdnlog ldap logrotate mailwrapper midi mudflap nls nothreads nptl nptlonly openmp pcre ppds pppd qos reflection server session snmp sparc spl ssl symlink tftp truetype unicode userlocales vim xml xorg" ELIBC="glibc" INPUT_DEVICES="keyboard mouse evdev" KERNEL="linux" LINGUAS="en de" USERLAND="GNU" Unset: CPPFLAGS, CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LC_ALL, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS
x86 stable
ppc stable
alpha stable
Sparc stable, all tests pass for me. Also marking sys-cluster/torque-2.3.0-r1 stable on sparc because torque is a dependency for openmpi. I don't see the failures mentioned in Comment 7; I wonder if this is USE flags related? (I build with USE="fortran heterogeneous pbs smp")
(In reply to comment #11) > I don't see the failures mentioned in Comment 7; I wonder if this is USE flags > related? (I build with USE="fortran heterogeneous pbs smp") More testing seems to show that my test failures were caused by the -ggdb CFLAG. Can someone confirm this and change the ebuild to filter out that flag if appropriate?
(In reply to comment #12) > More testing seems to show that my test failures were caused by the -ggdb > CFLAG. Can someone confirm this and change the ebuild to filter out that flag > if appropriate? All tests still pass for me with USE="fortran pbs romio smp" and -ggdb in CFLAGS. This is on amd64 though.
Created an attachment (id=158295) [edit] build.log-amd64-test-fail Tests fails on amd64 with USE="fortran" and the rest of use flags unset.
I haven't digged too deep, but I'd say the problem in amd64 is: https://svn.open-mpi.org/trac/ompi/ticket/1350 https://svn.open-mpi.org/trac/ompi/ticket/1351
The ebuild uses '$(use_enable romio romio-io)' but it's 'io-romio'.
(In reply to comment #14) > Created an attachment (id=158295) [edit] > build.log-amd64-test-fail > > Tests fails on amd64 with USE="fortran" and the rest of use flags unset. > Just a guess, but were you on a SMP machine? These tests will fail if you don't have the locking built in, which is configured by the smp useflag. If that's the case, I'm going to suggest that I should probably just remove the option as I doubt that many people are really using openmpi on single processor machines and really concerned about locking. Also, thanks for spotting the romio issue, it's fixed in cvs.
(In reply to comment #17) > > Just a guess, but were you on a SMP machine? These tests will fail if you > don't have the locking built in, which is configured by the smp useflag. > Yes. That's it ;-)
Justin, any news?
(In reply to comment #19) > Justin, any news? Sorry for the delay, I was moving over the past week. Anyways, as the test fails due to an SMP machine not adding the smp USE flag, is that really blocking stablization? I agree that the flag should be removed in the next version, but I don't want to revbump and force everyone to stablize again. You've been around a lot longer than me, so I'll follow your advice on this. Thanks.
I've just discussed with other amd64 members and amd64 will wait until a revbump is done addressing this. SMP is the rule these days for amd64 users and USE="smp" isn't even in default profiles, so this is a bug for us. Maybe this is less important for other arches, so I don't care that much if they stabilize the new revision inmediately or after the usual 30 days. But anyway, it shouldn't be a problem for them to stabilize too since it's a minor change. Thanks ;)
(In reply to comment #21) Fair enough, expect another stabilization bug shortly :) amd64 is the last arch, so I'm just going to close this, thanks everyone for your time.
