Hi Prentice, The hwloc developers sent out a patch earlier this morning:
http://www.open-mpi.org/community/lists/hwloc-devel/2011/12/2576.php I will wait for the next hwloc patch release and refresh OGS with that version, and I also need to include a minor change in aimk (configure hwloc with "--disable-libnuma"). As a temp. workaround, the hwloc distribution in the OGS source tarball on sourceforge will be updated with the patch. We did not encounter this issue because our systems don't have libnuma installed - is libnuma required for your applications? Rayson On Thu, Dec 8, 2011 at 8:59 AM, Prentice Bisbal <[email protected]> wrote: > Rayson, > > Thanks for the information. I can use the pre-compiled binaries for now. > Not as much fun, but they'll do. :) > > Prentice > > > On 12/07/2011 04:35 PM, Rayson Ho wrote: >> Prentice, >> >> The problem similar to this one encountered by the Open MPI build (Re: >> [OMPI users] EXTERNAL: Re: Unresolved reference 'mbind' and >> 'get_mempolicy'): >> >> http://www.open-mpi.org/community/lists/users/2011/09/17479.php >> >> In Grid Engine 2011.11, we don't use the NUMA functions (as users can >> do it easily in the job scripts by using the numactl (8) command). So >> we don't need the NUMA library. In Open MPI, there is the >> "--without-libnuma" option, which is actually not available in hwloc's >> configure script. >> >> So there are 3 options: >> >> 1) Use the precompiled binary: >> http://dl.dropbox.com/u/47200624/ge2011.11-x64.tar.gz >> >> 2) Modify configure in hwloc-1.2.2 to force linking against "-lnuma" to fail. >> >> 3) Or wait 1-2 days and I will check in a fix that disables using the >> NUMA library in hwloc. >> >> I will check with the hwloc guys and see if they will add a configure >> switch for hwloc in their future releases. >> >> Rayson >> >> >> >> On Wed, Dec 7, 2011 at 4:09 PM, Prentice Bisbal <[email protected]> wrote: >>> Rayson, >>> >>> Thanks. That was the problem. Now I'm getting another problem. It looks >>> like it can't find libnuma, even though I have the numactl and >>> numactl-devel packages installed. See error messages at bottom. >>> >>> I've verified that I have the numa RPMs installed >>> >>> $ rpm -qa | grep numa >>> numactl-devel-0.9.8-12.el5_6 >>> numactl-0.9.8-12.el5_6 >>> numactl-0.9.8-12.el5_6 >>> numactl-devel-0.9.8-12.el5_6 >>> >>> They're listed twice because I have both the i386 and x86_64 versions >>> installed. In the past, I've had problems with the linker can't find the >>> correct version (32-bit vs. 64-bit version) of the library. I noticed >>> that there's a "-lhwloc" in the command, but no "-lnuma". Is that correct? >>> >>> >>> gcc -o sge_execd -L/build/berkeleydb/lib/ -L. -rdynamic >>> -Wl,-rpath,\$ORIGIN/../../lib/linux-x64 -L/build/openssl-install/lib >>> execd.o sge_load_sensor.o dispatcher.o exec_job.o execd_ck_to_do.o >>> execd_get_new_conf.o execd_job_exec.o execd_kill_execd.o >>> execd_signal_queue.o execd_ticket.o get_path.o job_report_execd.o >>> load_avg.o pdc.o procfs.o ptf.o reaper_execd.o sge_report_execd.o >>> setup_execd.o tmpdir.o usage.o sge_options.o admin_mail.o config_file.o >>> mail.o read_write_job.o sge_dirent.o execution_states.o sig_handlers.o >>> startprog.o shutdown.o sge_mt_init.o -lgdi -lsgeobj -lsgeobjd -lcull >>> -lcomm -lcommlists -luti -llck -lrmon -ldl >>> -L/work/prentice/GE2011.11/source/3rdparty/hwloc/hwloc-1.2.2/LINUXX64/src/.libs/ >>> -lhwloc -ljemalloc -lm -lpthread >>> /work/prentice/GE2011.11/source/3rdparty/hwloc/hwloc-1.2.2/LINUXX64/src/.libs//libhwloc.a(topology-linux.o): >>> In function `hwloc_linux_set_area_membind': >>> topology-linux.c:(.text+0x331): undefined reference to `mbind' >>> topology-linux.c:(.text+0x3ac): undefined reference to `mbind' >>> /work/prentice/GE2011.11/source/3rdparty/hwloc/hwloc-1.2.2/LINUXX64/src/.libs//libhwloc.a(topology-linux.o): >>> In function `hwloc_linux_get_thisthread_membind': >>> topology-linux.c:(.text+0x4d5): undefined reference to `get_mempolicy' >>> topology-linux.c:(.text+0x5eb): undefined reference to `get_mempolicy' >>> /work/prentice/GE2011.11/source/3rdparty/hwloc/hwloc-1.2.2/LINUXX64/src/.libs//libhwloc.a(topology-linux.o): >>> In function `hwloc_linux_set_thisthread_membind': >>> topology-linux.c:(.text+0x68e): undefined reference to `set_mempolicy' >>> topology-linux.c:(.text+0x711): undefined reference to `migrate_pages' >>> topology-linux.c:(.text+0x734): undefined reference to `set_mempolicy' >>> >>> -- >>> Prentice >>> >>> >>> On 12/07/2011 03:29 PM, Rayson Ho wrote: >>>> Likely due to your $JAVA_HOME. >>>> >>>> If you don't have Java installed, then compile with aimk flags: -no-java >>>> -no-jni >>>> >>>> Otherwise, make sure it is set correctly: >>>> >>>> 1) Find the java real directory: >>>> $ readlink -f `which java` >>>> /usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre/bin/java >>>> >>>> 2) set the JAVA_HOME env. var: >>>> $ setenv JAVA_HOME /usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/ >>>> >>>> ("export" instead of "setenv" for bash/sh/ksh) >>>> >>>> The $JAVA_HOME/include has to contain "jni.h" or else you won't be >>>> able to compile the JNI related code in Grid Engine. >>>> >>>> Rayson >>>> >>>> >>>> >>>> On Wed, Dec 7, 2011 at 3:12 PM, Prentice Bisbal <[email protected]> wrote: >>>>> I'm trying to compile GE2011.11. I downloaded the tarball from >>>>> sourceforge. And I'm getting this error: >>>>> >>>>> In file included from /usr/include/_G_config.h:44, >>>>> from /usr/include/libio.h:32, >>>>> from /usr/include/stdio.h:72, >>>>> from ../libs/sgeobj/config.c:33: >>>>> /usr/include/gconv.h:72: error: expected declaration specifiers or ‘...’ >>>>> before ‘size_t’ >>>>> /usr/include/gconv.h:88: error: expected declaration specifiers or ‘...’ >>>>> before ‘size_t’ >>>>> /usr/include/gconv.h:97: error: expected declaration specifiers or ‘...’ >>>>> before ‘size_t’ >>>>> /usr/include/gconv.h:174: error: expected specifier-qualifier-list >>>>> before ‘size_t’ >>>>> In file included from /usr/include/stdio.h:72, >>>>> >>>>> I'm sure I'm missing some include file(s) somewhere, but I can't figure >>>>> out which one. Of, course when I google for these errors, I get >>>>> a bazillion hits for other errors include the string "size_t", so it >>>>> hasn't been very fruitful. >>>>> >>>>> Any assistance will be greatly appreciated. I'm sure it's something >>>>> minor. I'm building on a RHEL 5.7 system, using gcc >>>>> >>>>> -- >>>>> Prentice >>>>> >>>>> _______________________________________________ >>>>> users mailing list >>>>> [email protected] >>>>> https://gridengine.org/mailman/listinfo/users >>> _______________________________________________ >>> users mailing list >>> [email protected] >>> https://gridengine.org/mailman/listinfo/users > _______________________________________________ > users mailing list > [email protected] > https://gridengine.org/mailman/listinfo/users _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
