Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
Guido, This error message is from MPICH and not Open MPI. Make sure your environment is correct and the shared filesystem is mounted on the compute nodes. Cheers, Gilles Sent from my iPod > On Dec 12, 2019, at 1:44, Guido granda muñoz via users > wrote: > > Hi, > after following the instructions of the error message, in other works running > like this: > > #!/bin/bash > #PBS -l nodes=1:ppn=32 > #PBS -N mc_cond_0_h3 > #PBS -o mc_cond_0_h3.o > #PBS -e mc_cond_0_h3.e > > PATH=$HOME/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/bin:$PATH > LD_LIBRARY_PATH=/share/apps/gcc-7.3.0/lib64:$LD_LIBRARY_PATH > cd $PBS_O_WORKDIR > mpirun --mca btl vader,self -np 32 ./flash4 > > I get the following error messages: > > [mpiexec@compute-0-34.local] match_arg (utils/args/args.c:159): unrecognized > argument mca > [mpiexec@compute-0-34.local] HYDU_parse_array (utils/args/args.c:174): > argument matching returned error > [mpiexec@compute-0-34.local] parse_args (ui/mpich/utils.c:1596): error > parsing input array > [mpiexec@compute-0-34.local] HYD_uii_mpx_get_parameters > (ui/mpich/utils.c:1648): unable to parse user arguments > [mpiexec@compute-0-34.local] main (ui/mpich/mpiexec.c:149): error parsing > parameters > > Am I running it incorrectly ? > Cheers, > > El mar., 10 dic. 2019 a las 15:40, Guido granda muñoz > () escribió: >> Hello, >> I compiled the application now using openmpi-4.0.2: >> >> linux-vdso.so.1 => (0x7fffb23ff000) >> libhdf5.so.103 => >> /home/guido/libraries/compiled_with_gcc-7.3.0/hdf5-1.10.5_serial/lib/libhdf5.so.103 >> (0x2b3cd188c000) >> libz.so.1 => /lib64/libz.so.1 (0x2b3cd1e74000) >> libmpi_usempif08.so.40 => >> /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_usempif08.so.40 >> (0x2b3cd208a000) >> libmpi_usempi_ignore_tkr.so.40 => >> /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_usempi_ignore_tkr.so.40 >> (0x2b3cd22c) >> libmpi_mpifh.so.40 => >> /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_mpifh.so.40 >> (0x2b3cd24c7000) >> libmpi.so.40 => >> /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi.so.40 >> (0x2b3cd2723000) >> libgfortran.so.4 => /share/apps/gcc-7.3.0/lib64/libgfortran.so.4 >> (0x2b3cd2a55000) >> libm.so.6 => /lib64/libm.so.6 (0x2b3cd2dc3000) >> libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x2b3cd3047000) >> libquadmath.so.0 => /share/apps/gcc-5.4.0/lib64/libquadmath.so.0 >> (0x2b3cd325e000) >> libpthread.so.0 => /lib64/libpthread.so.0 (0x2b3cd349c000) >> libc.so.6 => /lib64/libc.so.6 (0x2b3cd36b9000) >> librt.so.1 => /lib64/librt.so.1 (0x2b3cd3a4e000) >> libdl.so.2 => /lib64/libdl.so.2 (0x2b3cd3c56000) >> libopen-rte.so.40 => >> /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libopen-rte.so.40 >> (0x2b3cd3e5b000) >> libopen-pal.so.40 => >> /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libopen-pal.so.40 >> (0x2b3cd411) >> libudev.so.0 => /lib64/libudev.so.0 (0x2b3cd4425000) >> libutil.so.1 => /lib64/libutil.so.1 (0x2b3cd4634000) >> /lib64/ld-linux-x86-64.so.2 (0x2b3cd166a000) >> >> and ran it like this: >> >> #!/bin/bash >> #PBS -l nodes=1:ppn=32 >> #PBS -N mc_cond_0_h3 >> #PBS -o mc_cond_0_h3.o >> #PBS -e mc_cond_0_h3.e >> >> PATH=$HOME/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/bin:$PATH >> LD_LIBRARY_PATH=/share/apps/gcc-7.3.0/lib64:$LD_LIBRARY_PATH >> cd $PBS_O_WORKDIR >> mpirun -np 32 ./flash4 >> >> and now I'm getting this error messages: >> >> -- >> As of version 3.0.0, the "sm" BTL is no longer available in Open MPI. >> >> Efficient, high-speed same-node shared memory communication support in >> Open MPI is available in the "vader" BTL. To use the vader BTL, you >> can re-run your job with: >> >> mpirun --mca btl vader,self,... your_mpi_application >> -- >> -- >> A requested component was not found, or was unable to be opened. This >> means that this component is either not installed or is unable to be >> used on your system (e.g., sometimes this means that shared libraries >> that the component requires are unable to be found/loaded). Note that >> Open MPI stopped checking at the first component that it did not find. >> >> Host: compute-0-34.local >> Framework: btl >> Component: sm >> -- >> -- >> It looks like MPI_INIT failed for some reason; your parallel process is >> likely to abort. There are many reasons that a parallel process can >> fail during MPI_INIT; some of which are due to configuration or environment >> problems. This failure appears to be an
Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
Hi, after following the instructions of the error message, in other works running like this: #!/bin/bash #PBS -l nodes=1:ppn=32 #PBS -N mc_cond_0_h3 #PBS -o mc_cond_0_h3.o #PBS -e mc_cond_0_h3.e PATH=$HOME/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/bin:$PATH LD_LIBRARY_PATH=/share/apps/gcc-7.3.0/lib64:$LD_LIBRARY_PATH cd $PBS_O_WORKDIR mpirun --mca btl vader,self -np 32 ./flash4 I get the following error messages: [mpiexec@compute-0-34.local] match_arg (utils/args/args.c:159): unrecognized argument mca [mpiexec@compute-0-34.local] HYDU_parse_array (utils/args/args.c:174): argument matching returned error [mpiexec@compute-0-34.local] parse_args (ui/mpich/utils.c:1596): error parsing input array [mpiexec@compute-0-34.local] HYD_uii_mpx_get_parameters (ui/mpich/utils.c:1648): unable to parse user arguments [mpiexec@compute-0-34.local] main (ui/mpich/mpiexec.c:149): error parsing parameters Am I running it incorrectly ? Cheers, El mar., 10 dic. 2019 a las 15:40, Guido granda muñoz (< guidogra...@gmail.com>) escribió: > Hello, > I compiled the application now using openmpi-4.0.2: > > linux-vdso.so.1 => (0x7fffb23ff000) > libhdf5.so.103 => > /home/guido/libraries/compiled_with_gcc-7.3.0/hdf5-1.10.5_serial/lib/libhdf5.so.103 > (0x2b3cd188c000) > libz.so.1 => /lib64/libz.so.1 (0x2b3cd1e74000) > libmpi_usempif08.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_usempif08.so.40 > (0x2b3cd208a000) > libmpi_usempi_ignore_tkr.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_usempi_ignore_tkr.so.40 > (0x2b3cd22c) > libmpi_mpifh.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_mpifh.so.40 > (0x2b3cd24c7000) > libmpi.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi.so.40 > (0x2b3cd2723000) > libgfortran.so.4 => /share/apps/gcc-7.3.0/lib64/libgfortran.so.4 > (0x2b3cd2a55000) > libm.so.6 => /lib64/libm.so.6 (0x2b3cd2dc3000) > libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x2b3cd3047000) > libquadmath.so.0 => /share/apps/gcc-5.4.0/lib64/libquadmath.so.0 > (0x2b3cd325e000) > libpthread.so.0 => /lib64/libpthread.so.0 (0x2b3cd349c000) > libc.so.6 => /lib64/libc.so.6 (0x2b3cd36b9000) > librt.so.1 => /lib64/librt.so.1 (0x2b3cd3a4e000) > libdl.so.2 => /lib64/libdl.so.2 (0x2b3cd3c56000) > libopen-rte.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libopen-rte.so.40 > (0x2b3cd3e5b000) > libopen-pal.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libopen-pal.so.40 > (0x2b3cd411) > libudev.so.0 => /lib64/libudev.so.0 (0x2b3cd4425000) > libutil.so.1 => /lib64/libutil.so.1 (0x2b3cd4634000) > /lib64/ld-linux-x86-64.so.2 (0x2b3cd166a000) > > and ran it like this: > > #!/bin/bash > #PBS -l nodes=1:ppn=32 > #PBS -N mc_cond_0_h3 > #PBS -o mc_cond_0_h3.o > #PBS -e mc_cond_0_h3.e > > PATH=$HOME/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/bin:$PATH > LD_LIBRARY_PATH=/share/apps/gcc-7.3.0/lib64:$LD_LIBRARY_PATH > cd $PBS_O_WORKDIR > mpirun -np 32 ./flash4 > > and now I'm getting this error messages: > > -- > > As of version 3.0.0, the "sm" BTL is no longer available in Open MPI. > > > Efficient, high-speed same-node shared memory communication support in > > Open MPI is available in the "vader" BTL. To use the vader BTL, you > > can re-run your job with: > > > mpirun --mca btl vader,self,... your_mpi_application > > -- > > -- > > A requested component was not found, or was unable to be opened. This > > means that this component is either not installed or is unable to be > > used on your system (e.g., sometimes this means that shared libraries > > that the component requires are unable to be found/loaded). Note that > > Open MPI stopped checking at the first component that it did not find. > > > Host: compute-0-34.local > > Framework: btl > > Component: sm > > -- > > -- > > It looks like MPI_INIT failed for some reason; your parallel process is > > likely to abort. There are many reasons that a parallel process can > > fail during MPI_INIT; some of which are due to configuration or environment > > problems. This failure appears to be an internal failure; here's some > > additional information (which may only be relevant to an Open MPI > > developer): > > > mca_bml_base_open() failed > > --> Returned "Not found" (-13) instead of "Success" (0) > > -- > > [compute-0-34:16915] *** An error occurred in MPI_Init > > [compute-0-34:16915] ***
Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
Hi Guido Your PATH and LD_LIBRARY_PATH seem to be inconsistent with each other: PATH=$HOME/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/bin:$PATH LD_LIBRARY_PATH=/share/apps/gcc-7.3.0/lib64:$LD_LIBRARY_PATH Hence, you may be mixing different versions of Open MPI. It looks like you installer Open MPI 4.0.2 here: /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/ Have you tried this instead? LD_LIBRARY_PATH=$HOME/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib:$LD_LIBRARY_PATH I hope this helps, Gus Correa On Tue, Dec 10, 2019 at 4:40 PM Guido granda muñoz via users < users@lists.open-mpi.org> wrote: > Hello, > I compiled the application now using openmpi-4.0.2: > > linux-vdso.so.1 => (0x7fffb23ff000) > libhdf5.so.103 => > /home/guido/libraries/compiled_with_gcc-7.3.0/hdf5-1.10.5_serial/lib/libhdf5.so.103 > (0x2b3cd188c000) > libz.so.1 => /lib64/libz.so.1 (0x2b3cd1e74000) > libmpi_usempif08.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_usempif08.so.40 > (0x2b3cd208a000) > libmpi_usempi_ignore_tkr.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_usempi_ignore_tkr.so.40 > (0x2b3cd22c) > libmpi_mpifh.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_mpifh.so.40 > (0x2b3cd24c7000) > libmpi.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi.so.40 > (0x2b3cd2723000) > libgfortran.so.4 => /share/apps/gcc-7.3.0/lib64/libgfortran.so.4 > (0x2b3cd2a55000) > libm.so.6 => /lib64/libm.so.6 (0x2b3cd2dc3000) > libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x2b3cd3047000) > libquadmath.so.0 => /share/apps/gcc-5.4.0/lib64/libquadmath.so.0 > (0x2b3cd325e000) > libpthread.so.0 => /lib64/libpthread.so.0 (0x2b3cd349c000) > libc.so.6 => /lib64/libc.so.6 (0x2b3cd36b9000) > librt.so.1 => /lib64/librt.so.1 (0x2b3cd3a4e000) > libdl.so.2 => /lib64/libdl.so.2 (0x2b3cd3c56000) > libopen-rte.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libopen-rte.so.40 > (0x2b3cd3e5b000) > libopen-pal.so.40 => > /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libopen-pal.so.40 > (0x2b3cd411) > libudev.so.0 => /lib64/libudev.so.0 (0x2b3cd4425000) > libutil.so.1 => /lib64/libutil.so.1 (0x2b3cd4634000) > /lib64/ld-linux-x86-64.so.2 (0x2b3cd166a000) > > and ran it like this: > > #!/bin/bash > #PBS -l nodes=1:ppn=32 > #PBS -N mc_cond_0_h3 > #PBS -o mc_cond_0_h3.o > #PBS -e mc_cond_0_h3.e > > PATH=$HOME/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/bin:$PATH > LD_LIBRARY_PATH=/share/apps/gcc-7.3.0/lib64:$LD_LIBRARY_PATH > cd $PBS_O_WORKDIR > mpirun -np 32 ./flash4 > > and now I'm getting this error messages: > > -- > > As of version 3.0.0, the "sm" BTL is no longer available in Open MPI. > > > Efficient, high-speed same-node shared memory communication support in > > Open MPI is available in the "vader" BTL. To use the vader BTL, you > > can re-run your job with: > > > mpirun --mca btl vader,self,... your_mpi_application > > -- > > -- > > A requested component was not found, or was unable to be opened. This > > means that this component is either not installed or is unable to be > > used on your system (e.g., sometimes this means that shared libraries > > that the component requires are unable to be found/loaded). Note that > > Open MPI stopped checking at the first component that it did not find. > > > Host: compute-0-34.local > > Framework: btl > > Component: sm > > -- > > -- > > It looks like MPI_INIT failed for some reason; your parallel process is > > likely to abort. There are many reasons that a parallel process can > > fail during MPI_INIT; some of which are due to configuration or environment > > problems. This failure appears to be an internal failure; here's some > > additional information (which may only be relevant to an Open MPI > > developer): > > > mca_bml_base_open() failed > > --> Returned "Not found" (-13) instead of "Success" (0) > > -- > > [compute-0-34:16915] *** An error occurred in MPI_Init > > [compute-0-34:16915] *** reported by process [3776708609,5] > > [compute-0-34:16915] *** on a NULL communicator > > [compute-0-34:16915] *** Unknown error > > [compute-0-34:16915] *** MPI_ERRORS_ARE_FATAL (processes in this > communicator will now abort, > > [compute-0-34:16915] *** and potentially your MPI job) > > [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file > server/pmix_server.c at line 2147 > > [compute-0-34.local:16902]
Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
Hello, I compiled the application now using openmpi-4.0.2: linux-vdso.so.1 => (0x7fffb23ff000) libhdf5.so.103 => /home/guido/libraries/compiled_with_gcc-7.3.0/hdf5-1.10.5_serial/lib/libhdf5.so.103 (0x2b3cd188c000) libz.so.1 => /lib64/libz.so.1 (0x2b3cd1e74000) libmpi_usempif08.so.40 => /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_usempif08.so.40 (0x2b3cd208a000) libmpi_usempi_ignore_tkr.so.40 => /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_usempi_ignore_tkr.so.40 (0x2b3cd22c) libmpi_mpifh.so.40 => /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi_mpifh.so.40 (0x2b3cd24c7000) libmpi.so.40 => /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libmpi.so.40 (0x2b3cd2723000) libgfortran.so.4 => /share/apps/gcc-7.3.0/lib64/libgfortran.so.4 (0x2b3cd2a55000) libm.so.6 => /lib64/libm.so.6 (0x2b3cd2dc3000) libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x2b3cd3047000) libquadmath.so.0 => /share/apps/gcc-5.4.0/lib64/libquadmath.so.0 (0x2b3cd325e000) libpthread.so.0 => /lib64/libpthread.so.0 (0x2b3cd349c000) libc.so.6 => /lib64/libc.so.6 (0x2b3cd36b9000) librt.so.1 => /lib64/librt.so.1 (0x2b3cd3a4e000) libdl.so.2 => /lib64/libdl.so.2 (0x2b3cd3c56000) libopen-rte.so.40 => /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libopen-rte.so.40 (0x2b3cd3e5b000) libopen-pal.so.40 => /home/guido/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/lib/libopen-pal.so.40 (0x2b3cd411) libudev.so.0 => /lib64/libudev.so.0 (0x2b3cd4425000) libutil.so.1 => /lib64/libutil.so.1 (0x2b3cd4634000) /lib64/ld-linux-x86-64.so.2 (0x2b3cd166a000) and ran it like this: #!/bin/bash #PBS -l nodes=1:ppn=32 #PBS -N mc_cond_0_h3 #PBS -o mc_cond_0_h3.o #PBS -e mc_cond_0_h3.e PATH=$HOME/libraries/compiled_with_gcc-7.3.0/openmpi-4.0.2/bin:$PATH LD_LIBRARY_PATH=/share/apps/gcc-7.3.0/lib64:$LD_LIBRARY_PATH cd $PBS_O_WORKDIR mpirun -np 32 ./flash4 and now I'm getting this error messages: -- As of version 3.0.0, the "sm" BTL is no longer available in Open MPI. Efficient, high-speed same-node shared memory communication support in Open MPI is available in the "vader" BTL. To use the vader BTL, you can re-run your job with: mpirun --mca btl vader,self,... your_mpi_application -- -- A requested component was not found, or was unable to be opened. This means that this component is either not installed or is unable to be used on your system (e.g., sometimes this means that shared libraries that the component requires are unable to be found/loaded). Note that Open MPI stopped checking at the first component that it did not find. Host: compute-0-34.local Framework: btl Component: sm -- -- It looks like MPI_INIT failed for some reason; your parallel process is likely to abort. There are many reasons that a parallel process can fail during MPI_INIT; some of which are due to configuration or environment problems. This failure appears to be an internal failure; here's some additional information (which may only be relevant to an Open MPI developer): mca_bml_base_open() failed --> Returned "Not found" (-13) instead of "Success" (0) -- [compute-0-34:16915] *** An error occurred in MPI_Init [compute-0-34:16915] *** reported by process [3776708609,5] [compute-0-34:16915] *** on a NULL communicator [compute-0-34:16915] *** Unknown error [compute-0-34:16915] *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort, [compute-0-34:16915] *** and potentially your MPI job) [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902] PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2147 [compute-0-34.local:16902]
Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
On Dec 6, 2019, at 1:03 PM, Jeff Squyres (jsquyres) via users wrote: > >> I get the same error when running in a single node. I will try to use the >> last version. Is there way to check if different versions of open mpi were >> used in different nodes? > > mpirun -np 2 ompi_info | head > > Or something like that. With 1.10, I don't know/remember the mpirun CLI > option to make one process per node (when ppn>1); you may have to check that. > Or just "mpirun -np 33 ompi_info | head" and examine the output carefully to > find the 33rd output and see if it's different. Poor quoting on my part. The intent was to see just the first few lines from running `ompi_info` on each node. So maybe something like: -- $ cat foo.sh #!/bin/sh ompi_info | head $ mpirun -np 2 foo.sh -- Or "mprun -np 33 foo.sh", etc. -- Jeff Squyres jsquy...@cisco.com
Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
On Dec 6, 2019, at 12:40 PM, Guido granda muñoz wrote: > > I get the same error when running in a single node. I will try to use the > last version. Is there way to check if different versions of open mpi were > used in different nodes? mpirun -np 2 ompi_info | head Or something like that. With 1.10, I don't know/remember the mpirun CLI option to make one process per node (when ppn>1); you may have to check that. Or just "mpirun -np 33 ompi_info | head" and examine the output carefully to find the 33rd output and see if it's different. That being said, thinking a little deeper on this: with your ld output, it's probably kinda unlikely that you have mismatched versions (because you're linking against very specific OMPI library versions) -- particularly if you're getting the same error when running on a single node. Open MPI is 100% user space code -- you can just install Open MPI v4.0.2 under your $HOME and run it there. You might want to see how your system-level Open MPI is installed (e.g., use the right ./configure option to get the PBS/TM integration, if you have a high-speed/HPC-class network support API library, ...etc.), and mimic all of that in your own personal Open MPI install. -- Jeff Squyres jsquy...@cisco.com
Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
Hello, I get the same error when running in a single node. I will try to use the last version. Is there way to check if different versions of open mpi were used in different nodes? Cheers, El jue., 5 dic. 2019 a las 19:10, Jeff Squyres (jsquyres) (< jsquy...@cisco.com>) escribió: > Are you able to run on a single node? > > Is there any chance you can upgrade your Open MPI? 1.10 is ancient and > isn't really supported any more. 4.0.2 is the current version. > > > On Dec 5, 2019, at 7:15 PM, Guido granda muñoz > wrote: > > Hello Jeff, > > Thank you for replying. I ran it using PBS like this: > > #!/bin/bash > #PBS -l nodes=2:ppn=32 > #PBS -N cond_0_h3 > #PBS -o cond_0_h3.o > #PBS -e cond_0_h3.e > > > PATH=$PATH:/usr/mpi/intel/openmpi-1.10.3/bin > > LD_LIBRARY_PATH=/share/apps/composerxe-2011.2.137/lib/intel64:$LD_LIBRARY_PATH > cd $PBS_O_WORKDIR > mpirun -np 64 ./flash4 > > Besides that, my .bashrc has the following lines: > # Source global definitions > if [ -f /etc/bashrc ]; then > . /etc/bashrc > fi > #irya.guido.intel > export PATH=$PATH:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/bin > export > INCLUDE=$INCLUDE:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/include > export > CPATH=$CPATH:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/include > export > LIBRARY_PATH=$LIBRARY_PATH:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib > export > LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib > ###intel open-mpi system### > export PATH=$PATH:/usr/mpi/intel/openmpi-1.10.3/bin > export INCLUDE=$INCLUDE:/usr/mpi/intel/openmpi-1.10.3/include > export CPATH=$CPATH:/usr/mpi/intel/openmpi-1.10.3/include > export LIBRARY_PATH=$LIBRARY_PATH:/usr/mpi/intel/openmpi-1.10.3/lib > export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/mpi/intel/openmpi-1.10.3/lib > ##anaconda 3 ## > export PATH="$PATH:/home/guido/anaconda3/bin" # commented out by conda > initialize > # Intel > export > PATH=$PATH:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ > export > INCLUDE=$INCLUDE:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ > export > CPATH=$CPATH:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ > export > LIBRARY_PATH=$LIBRARY_PATH:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ > export > LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ > > How Can I check that what you suggested was the reason for this error? > Cheers, > > El jue., 5 dic. 2019 a las 18:02, Jeff Squyres (jsquyres) (< > jsquy...@cisco.com>) escribió: > >> How did you try to execute your application? >> >> An error message like this can mean that you accidentally mixed versions >> of Open MPI within your run (e.g., used Open MPI va.b.c on node A but used >> Open MPI vx.y.z on node B). >> >> >> > On Dec 5, 2019, at 5:28 PM, Guido granda muñoz via users < >> users@lists.open-mpi.org> wrote: >> > >> > Hello open-mpi users, >> > I'm getting some problem while using openmpi-1.10.3. The executable was >> compiled using : (ldd output) >> > >> > linux-vdso.so.1 => (0x7fffd9e8b000) >> > libhdf5.so.10 => >> /home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib/libhdf5.so.10 >> (0x2ac4313c4000) >> > libhdf5_fortran.so.10 => >> /home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib/libhdf5_fortran.so.10 >> (0x2ac4319ca000) >> > libz.so.1 => /lib64/libz.so.1 (0x2ac431c2d000) >> > libmpi_usempif08.so.11 => >> /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_usempif08.so.11 >> (0x2ac431e44000) >> > libmpi_usempi_ignore_tkr.so.6 => >> /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_usempi_ignore_tkr.so.6 >> (0x2ac432077000) >> > libmpi_mpifh.so.12 => >> /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_mpifh.so.12 (0x2ac43228) >> > libmpi.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libmpi.so.12 >> (0x2ac4324df000) >> > libm.so.6 => /lib64/libm.so.6 (0x2ac4327e7000) >> > libpthread.so.0 => /lib64/libpthread.so.0 (0x2ac432a6b000) >> > libc.so.6 => /lib64/libc.so.6 (0x2ac432c88000) >> > libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x2ac43301d000) >> > libdl.so.2 => /lib64/libdl.so.2 (0x2ac433233000) >> > libimf.so => >> /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libimf.so >> (0x2ac433437000) >> > libsvml.so => >> /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libsvml.so >> (0x2ac43381b000) >> > libintlc.so.5 => >> /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libintlc.so.5 >> (0x2ac433ec3000) >> > libifport.so.5 => >> /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifport.so.5 >> (0x2ac434013000) >> > libifcore.so.5 => >> /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifcore.so.5 >> (0x2ac43414c000) >> > libopen-rte.so.12 => >>
Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
Hello Jeff, Thank you for replying. I ran it using PBS like this: #!/bin/bash #PBS -l nodes=2:ppn=32 #PBS -N cond_0_h3 #PBS -o cond_0_h3.o #PBS -e cond_0_h3.e PATH=$PATH:/usr/mpi/intel/openmpi-1.10.3/bin LD_LIBRARY_PATH=/share/apps/composerxe-2011.2.137/lib/intel64:$LD_LIBRARY_PATH cd $PBS_O_WORKDIR mpirun -np 64 ./flash4 Besides that, my .bashrc has the following lines: # Source global definitions if [ -f /etc/bashrc ]; then . /etc/bashrc fi #irya.guido.intel export PATH=$PATH:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/bin export INCLUDE=$INCLUDE:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/include export CPATH=$CPATH:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/include export LIBRARY_PATH=$LIBRARY_PATH:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib ###intel open-mpi system### export PATH=$PATH:/usr/mpi/intel/openmpi-1.10.3/bin export INCLUDE=$INCLUDE:/usr/mpi/intel/openmpi-1.10.3/include export CPATH=$CPATH:/usr/mpi/intel/openmpi-1.10.3/include export LIBRARY_PATH=$LIBRARY_PATH:/usr/mpi/intel/openmpi-1.10.3/lib export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/mpi/intel/openmpi-1.10.3/lib ##anaconda 3 ## export PATH="$PATH:/home/guido/anaconda3/bin" # commented out by conda initialize # Intel export PATH=$PATH:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ export INCLUDE=$INCLUDE:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ export CPATH=$CPATH:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ export LIBRARY_PATH=$LIBRARY_PATH:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/share/apps/composerxe-2011.2.137/composerxe-2011.2.137/bin/intel64/ How Can I check that what you suggested was the reason for this error? Cheers, El jue., 5 dic. 2019 a las 18:02, Jeff Squyres (jsquyres) (< jsquy...@cisco.com>) escribió: > How did you try to execute your application? > > An error message like this can mean that you accidentally mixed versions > of Open MPI within your run (e.g., used Open MPI va.b.c on node A but used > Open MPI vx.y.z on node B). > > > > On Dec 5, 2019, at 5:28 PM, Guido granda muñoz via users < > users@lists.open-mpi.org> wrote: > > > > Hello open-mpi users, > > I'm getting some problem while using openmpi-1.10.3. The executable was > compiled using : (ldd output) > > > > linux-vdso.so.1 => (0x7fffd9e8b000) > > libhdf5.so.10 => > /home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib/libhdf5.so.10 > (0x2ac4313c4000) > > libhdf5_fortran.so.10 => > /home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib/libhdf5_fortran.so.10 > (0x2ac4319ca000) > > libz.so.1 => /lib64/libz.so.1 (0x2ac431c2d000) > > libmpi_usempif08.so.11 => > /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_usempif08.so.11 > (0x2ac431e44000) > > libmpi_usempi_ignore_tkr.so.6 => > /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_usempi_ignore_tkr.so.6 > (0x2ac432077000) > > libmpi_mpifh.so.12 => > /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_mpifh.so.12 (0x2ac43228) > > libmpi.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libmpi.so.12 > (0x2ac4324df000) > > libm.so.6 => /lib64/libm.so.6 (0x2ac4327e7000) > > libpthread.so.0 => /lib64/libpthread.so.0 (0x2ac432a6b000) > > libc.so.6 => /lib64/libc.so.6 (0x2ac432c88000) > > libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x2ac43301d000) > > libdl.so.2 => /lib64/libdl.so.2 (0x2ac433233000) > > libimf.so => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libimf.so > (0x2ac433437000) > > libsvml.so => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libsvml.so > (0x2ac43381b000) > > libintlc.so.5 => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libintlc.so.5 > (0x2ac433ec3000) > > libifport.so.5 => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifport.so.5 > (0x2ac434013000) > > libifcore.so.5 => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifcore.so.5 > (0x2ac43414c000) > > libopen-rte.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libopen-rte.so.12 > (0x2ac4343ad000) > > libopen-pal.so.13 => /usr/mpi/intel/openmpi-1.10.3/lib/libopen-pal.so.13 > (0x2ac43464e000) > > librt.so.1 => /lib64/librt.so.1 (0x2ac43496a000) > > libutil.so.1 => /lib64/libutil.so.1 (0x2ac434b72000) > > libifcoremt.so.5 => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifcoremt.so.5 > (0x2ac434d76000) > > /lib64/ld-linux-x86-64.so.2 (0x2ac4311a2000) > > > > When I run it, I get the following message: > > > > [compute-0-34.local:17553] [[5279,0],0] mca_oob_tcp_recv_handler: > invalid message type: 15 > > > -- > >
Re: [OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
How did you try to execute your application? An error message like this can mean that you accidentally mixed versions of Open MPI within your run (e.g., used Open MPI va.b.c on node A but used Open MPI vx.y.z on node B). > On Dec 5, 2019, at 5:28 PM, Guido granda muñoz via users > wrote: > > Hello open-mpi users, > I'm getting some problem while using openmpi-1.10.3. The executable was > compiled using : (ldd output) > > linux-vdso.so.1 => (0x7fffd9e8b000) > libhdf5.so.10 => > /home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib/libhdf5.so.10 > (0x2ac4313c4000) > libhdf5_fortran.so.10 => > /home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib/libhdf5_fortran.so.10 > (0x2ac4319ca000) > libz.so.1 => /lib64/libz.so.1 (0x2ac431c2d000) > libmpi_usempif08.so.11 => > /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_usempif08.so.11 (0x2ac431e44000) > libmpi_usempi_ignore_tkr.so.6 => > /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_usempi_ignore_tkr.so.6 > (0x2ac432077000) > libmpi_mpifh.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_mpifh.so.12 > (0x2ac43228) > libmpi.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libmpi.so.12 > (0x2ac4324df000) > libm.so.6 => /lib64/libm.so.6 (0x2ac4327e7000) > libpthread.so.0 => /lib64/libpthread.so.0 (0x2ac432a6b000) > libc.so.6 => /lib64/libc.so.6 (0x2ac432c88000) > libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x2ac43301d000) > libdl.so.2 => /lib64/libdl.so.2 (0x2ac433233000) > libimf.so => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libimf.so > (0x2ac433437000) > libsvml.so => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libsvml.so > (0x2ac43381b000) > libintlc.so.5 => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libintlc.so.5 > (0x2ac433ec3000) > libifport.so.5 => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifport.so.5 > (0x2ac434013000) > libifcore.so.5 => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifcore.so.5 > (0x2ac43414c000) > libopen-rte.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libopen-rte.so.12 > (0x2ac4343ad000) > libopen-pal.so.13 => /usr/mpi/intel/openmpi-1.10.3/lib/libopen-pal.so.13 > (0x2ac43464e000) > librt.so.1 => /lib64/librt.so.1 (0x2ac43496a000) > libutil.so.1 => /lib64/libutil.so.1 (0x2ac434b72000) > libifcoremt.so.5 => > /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifcoremt.so.5 > (0x2ac434d76000) > /lib64/ld-linux-x86-64.so.2 (0x2ac4311a2000) > > When I run it, I get the following message: > > [compute-0-34.local:17553] [[5279,0],0] mca_oob_tcp_recv_handler: invalid > message type: 15 > -- > mpirun noticed that the job aborted, but has no info as to the process > that caused that situation. > > The executable was also compiled using hdf5-1.8.20 > I really don't know wht this error means, could please help me? > Cheers, > -- > Guido -- Jeff Squyres jsquy...@cisco.com
[OMPI users] mca_oob_tcp_recv_handler: invalid message type: 15
Hello open-mpi users, I'm getting some problem while using openmpi-1.10.3. The executable was compiled using : (ldd output) linux-vdso.so.1 => (0x7fffd9e8b000) libhdf5.so.10 => /home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib/libhdf5.so.10 (0x2ac4313c4000) libhdf5_fortran.so.10 => /home/guido/libraries/compiled_with_intel/hdf5-1.8.20/lib/libhdf5_fortran.so.10 (0x2ac4319ca000) libz.so.1 => /lib64/libz.so.1 (0x2ac431c2d000) libmpi_usempif08.so.11 => /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_usempif08.so.11 (0x2ac431e44000) libmpi_usempi_ignore_tkr.so.6 => /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_usempi_ignore_tkr.so.6 (0x2ac432077000) libmpi_mpifh.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libmpi_mpifh.so.12 (0x2ac43228) libmpi.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libmpi.so.12 (0x2ac4324df000) libm.so.6 => /lib64/libm.so.6 (0x2ac4327e7000) libpthread.so.0 => /lib64/libpthread.so.0 (0x2ac432a6b000) libc.so.6 => /lib64/libc.so.6 (0x2ac432c88000) libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x2ac43301d000) libdl.so.2 => /lib64/libdl.so.2 (0x2ac433233000) libimf.so => /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libimf.so (0x2ac433437000) libsvml.so => /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libsvml.so (0x2ac43381b000) libintlc.so.5 => /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libintlc.so.5 (0x2ac433ec3000) libifport.so.5 => /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifport.so.5 (0x2ac434013000) libifcore.so.5 => /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifcore.so.5 (0x2ac43414c000) libopen-rte.so.12 => /usr/mpi/intel/openmpi-1.10.3/lib/libopen-rte.so.12 (0x2ac4343ad000) libopen-pal.so.13 => /usr/mpi/intel/openmpi-1.10.3/lib/libopen-pal.so.13 (0x2ac43464e000) librt.so.1 => /lib64/librt.so.1 (0x2ac43496a000) libutil.so.1 => /lib64/libutil.so.1 (0x2ac434b72000) libifcoremt.so.5 => /share/apps/composerxe-2011.2.137/composerxe-2011.2.137/compiler/lib/intel64/libifcoremt.so.5 (0x2ac434d76000) /lib64/ld-linux-x86-64.so.2 (0x2ac4311a2000) When I run it, I get the following message: [compute-0-34.local:17553] [[5279,0],0] mca_oob_tcp_recv_handler: invalid message type: 15 -- mpirun noticed that the job aborted, but has no info as to the process that caused that situation. The executable was also compiled using hdf5-1.8.20 I really don't know wht this error means, could please help me? Cheers, -- Guido