Are you using TCP as the MPI transport? If so, another thing to try is to limit the IP interfaces that MPI uses for its traffic to see if there's some kind of problem with specific networks.
For example: mpirun --mca btl_tcp_if_include eth0 ... If that works, then try adding in any/all other IP interfaces that you have on your machines. A sorta-wild guess: you have some IP interfaces that aren't working, or at least, don't work in the way that OMPI wants them to work. So the first barrier works because it flows across eth0 (or some other first network that, as far as OMPI is concerned, works just fine). But then the next barrier round-robin advances to the next IP interface, and it doesn't work for some reason. We've seen virtual machine bridge interfaces cause problems, for example. E.g., if a machine has a Xen virtual machine interface (vibr0, IIRC?), then OMPI will try to use it to communicate with peer MPI processes because it has a "compatible" IP address, and OMPI think it should be connected/reachable to peers. If this is the case, you might want to disable such interfaces and/or use btl_tcp_if_include or btl_tcp_if_exclude to select the interfaces that you want to use. Pro tip: if you use btl_tcp_if_exclude, remember to exclude the loopback interface, too. OMPI defaults to a btl_tcp_if_include="" (blank) and btl_tcp_if_exclude="lo0". So if you override btl_tcp_if_exclude, you need to be sure to *also* include lo0 in the new value. For example: mpirun --mca btl_tcp_if_exclude lo0,virb0 ... Also, if possible, try upgrading to Open MPI 1.8.1. On May 4, 2014, at 2:15 PM, Clay Kirkland <clay.kirkl...@versityinc.com> wrote: > I am configuring with all defaults. Just doing a ./configure and then > make and make install. I have used open mpi on several kinds of > unix systems this way and have had no trouble before. I believe I > last had success on a redhat version of linux. > > > On Sat, May 3, 2014 at 11:00 AM, <users-requ...@open-mpi.org> wrote: > Send users mailing list submissions to > us...@open-mpi.org > > To subscribe or unsubscribe via the World Wide Web, visit > http://www.open-mpi.org/mailman/listinfo.cgi/users > or, via email, send a message with subject or body 'help' to > users-requ...@open-mpi.org > > You can reach the person managing the list at > users-ow...@open-mpi.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of users digest..." > > > Today's Topics: > > 1. MPI_Barrier hangs on second attempt but only when multiple > hosts used. (Clay Kirkland) > 2. Re: MPI_Barrier hangs on second attempt but only when > multiple hosts used. (Ralph Castain) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Fri, 2 May 2014 16:24:04 -0500 > From: Clay Kirkland <clay.kirkl...@versityinc.com> > To: us...@open-mpi.org > Subject: [OMPI users] MPI_Barrier hangs on second attempt but only > when multiple hosts used. > Message-ID: > <CAJDnjA8Wi=FEjz6Vz+Bc34b+nFE=tf4b7g0bqgmbekg7h-p...@mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > I have been using MPI for many many years so I have very well debugged mpi > tests. I am > having trouble on either openmpi-1.4.5 or openmpi-1.6.5 versions though > with getting the > MPI_Barrier calls to work. It works fine when I run all processes on one > machine but when > I run with two or more hosts the second call to MPI_Barrier always hangs. > Not the first one, > but always the second one. I looked at FAQ's and such but found nothing > except for a comment > that MPI_Barrier problems were often problems with fire walls. Also > mentioned as a problem > was not having the same version of mpi on both machines. I turned > firewalls off and removed > and reinstalled the same version on both hosts but I still see the same > thing. I then installed > lam mpi on two of my machines and that works fine. I can call the > MPI_Barrier function when run on > one of two machines by itself many times with no hangs. Only hangs if two > or more hosts are involved. > These runs are all being done on CentOS release 6.4. Here is test program > I used. > > main (argc, argv) > int argc; > char **argv; > { > char message[20]; > char hoster[256]; > char nameis[256]; > int fd, i, j, jnp, iret, myrank, np, ranker, recker; > MPI_Comm comm; > MPI_Status status; > > MPI_Init( &argc, &argv ); > MPI_Comm_rank( MPI_COMM_WORLD, &myrank); > MPI_Comm_size( MPI_COMM_WORLD, &np); > > gethostname(hoster,256); > > printf(" In rank %d and host= %s Do Barrier call > 1.\n",myrank,hoster); > MPI_Barrier(MPI_COMM_WORLD); > printf(" In rank %d and host= %s Do Barrier call > 2.\n",myrank,hoster); > MPI_Barrier(MPI_COMM_WORLD); > printf(" In rank %d and host= %s Do Barrier call > 3.\n",myrank,hoster); > MPI_Barrier(MPI_COMM_WORLD); > MPI_Finalize(); > exit(0); > } > > Here are three runs of test program. First with two processes on one > host, then with > two processes on another host, and finally with one process on each of two > hosts. The > first two runs are fine but the last run hangs on the second MPI_Barrier. > > [root@centos MPI]# /usr/local/bin/mpirun -np 2 --host centos a.out > In rank 0 and host= centos Do Barrier call 1. > In rank 1 and host= centos Do Barrier call 1. > In rank 1 and host= centos Do Barrier call 2. > In rank 1 and host= centos Do Barrier call 3. > In rank 0 and host= centos Do Barrier call 2. > In rank 0 and host= centos Do Barrier call 3. > [root@centos MPI]# /usr/local/bin/mpirun -np 2 --host RAID a.out > /root/.bashrc: line 14: unalias: ls: not found > In rank 0 and host= RAID Do Barrier call 1. > In rank 0 and host= RAID Do Barrier call 2. > In rank 0 and host= RAID Do Barrier call 3. > In rank 1 and host= RAID Do Barrier call 1. > In rank 1 and host= RAID Do Barrier call 2. > In rank 1 and host= RAID Do Barrier call 3. > [root@centos MPI]# /usr/local/bin/mpirun -np 2 --host centos,RAID a.out > /root/.bashrc: line 14: unalias: ls: not found > In rank 0 and host= centos Do Barrier call 1. > In rank 0 and host= centos Do Barrier call 2. > In rank 1 and host= RAID Do Barrier call 1. > In rank 1 and host= RAID Do Barrier call 2. > > Since it is such a simple test and problem and such a widely used MPI > function, it must obviously > be an installation or configuration problem. A pstack for each of the > hung MPI_Barrier processes > on the two machines shows this: > > [root@centos ~]# pstack 31666 > #0 0x0000003baf0e8ee3 in __epoll_wait_nocancel () from /lib64/libc.so.6 > #1 0x00007f5de06125eb in epoll_dispatch () from /usr/local/lib/libmpi.so.1 > #2 0x00007f5de061475a in opal_event_base_loop () from > /usr/local/lib/libmpi.so.1 > #3 0x00007f5de0639229 in opal_progress () from /usr/local/lib/libmpi.so.1 > #4 0x00007f5de0586f75 in ompi_request_default_wait_all () from > /usr/local/lib/libmpi.so.1 > #5 0x00007f5ddc59565e in ompi_coll_tuned_sendrecv_actual () from > /usr/local/lib/openmpi/mca_coll_tuned.so > #6 0x00007f5ddc59d8ff in ompi_coll_tuned_barrier_intra_two_procs () from > /usr/local/lib/openmpi/mca_coll_tuned.so > #7 0x00007f5de05941c2 in PMPI_Barrier () from /usr/local/lib/libmpi.so.1 > #8 0x0000000000400a43 in main () > > [root@RAID openmpi-1.6.5]# pstack 22167 > #0 0x00000030302e8ee3 in __epoll_wait_nocancel () from /lib64/libc.so.6 > #1 0x00007f7ee46885eb in epoll_dispatch () from /usr/local/lib/libmpi.so.1 > #2 0x00007f7ee468a75a in opal_event_base_loop () from > /usr/local/lib/libmpi.so.1 > #3 0x00007f7ee46af229 in opal_progress () from /usr/local/lib/libmpi.so.1 > #4 0x00007f7ee45fcf75 in ompi_request_default_wait_all () from > /usr/local/lib/libmpi.so.1 > #5 0x00007f7ee060b65e in ompi_coll_tuned_sendrecv_actual () from > /usr/local/lib/openmpi/mca_coll_tuned.so > #6 0x00007f7ee06138ff in ompi_coll_tuned_barrier_intra_two_procs () from > /usr/local/lib/openmpi/mca_coll_tuned.so > #7 0x00007f7ee460a1c2 in PMPI_Barrier () from /usr/local/lib/libmpi.so.1 > #8 0x0000000000400a43 in main () > > Which looks exactly the same on each machine. Any thoughts or ideas would > be greatly appreciated as > I am stuck. > > Clay Kirkland > -------------- next part -------------- > HTML attachment scrubbed and removed > > ------------------------------ > > Message: 2 > Date: Sat, 3 May 2014 06:39:20 -0700 > From: Ralph Castain <r...@open-mpi.org> > To: Open MPI Users <us...@open-mpi.org> > Subject: Re: [OMPI users] MPI_Barrier hangs on second attempt but only > when multiple hosts used. > Message-ID: <3cf53d73-15d9-40bb-a2de-50ba3561a...@open-mpi.org> > Content-Type: text/plain; charset="us-ascii" > > Hmmm...just testing on my little cluster here on two nodes, it works just > fine with 1.8.2: > > [rhc@bend001 v1.8]$ mpirun -n 2 --map-by node ./a.out > In rank 0 and host= bend001 Do Barrier call 1. > In rank 0 and host= bend001 Do Barrier call 2. > In rank 0 and host= bend001 Do Barrier call 3. > In rank 1 and host= bend002 Do Barrier call 1. > In rank 1 and host= bend002 Do Barrier call 2. > In rank 1 and host= bend002 Do Barrier call 3. > [rhc@bend001 v1.8]$ > > > How are you configuring OMPI? > > > On May 2, 2014, at 2:24 PM, Clay Kirkland <clay.kirkl...@versityinc.com> > wrote: > > > I have been using MPI for many many years so I have very well debugged mpi > > tests. I am > > having trouble on either openmpi-1.4.5 or openmpi-1.6.5 versions though > > with getting the > > MPI_Barrier calls to work. It works fine when I run all processes on one > > machine but when > > I run with two or more hosts the second call to MPI_Barrier always hangs. > > Not the first one, > > but always the second one. I looked at FAQ's and such but found nothing > > except for a comment > > that MPI_Barrier problems were often problems with fire walls. Also > > mentioned as a problem > > was not having the same version of mpi on both machines. I turned > > firewalls off and removed > > and reinstalled the same version on both hosts but I still see the same > > thing. I then installed > > lam mpi on two of my machines and that works fine. I can call the > > MPI_Barrier function when run on > > one of two machines by itself many times with no hangs. Only hangs if two > > or more hosts are involved. > > These runs are all being done on CentOS release 6.4. Here is test program > > I used. > > > > main (argc, argv) > > int argc; > > char **argv; > > { > > char message[20]; > > char hoster[256]; > > char nameis[256]; > > int fd, i, j, jnp, iret, myrank, np, ranker, recker; > > MPI_Comm comm; > > MPI_Status status; > > > > MPI_Init( &argc, &argv ); > > MPI_Comm_rank( MPI_COMM_WORLD, &myrank); > > MPI_Comm_size( MPI_COMM_WORLD, &np); > > > > gethostname(hoster,256); > > > > printf(" In rank %d and host= %s Do Barrier call > > 1.\n",myrank,hoster); > > MPI_Barrier(MPI_COMM_WORLD); > > printf(" In rank %d and host= %s Do Barrier call > > 2.\n",myrank,hoster); > > MPI_Barrier(MPI_COMM_WORLD); > > printf(" In rank %d and host= %s Do Barrier call > > 3.\n",myrank,hoster); > > MPI_Barrier(MPI_COMM_WORLD); > > MPI_Finalize(); > > exit(0); > > } > > > > Here are three runs of test program. First with two processes on one > > host, then with > > two processes on another host, and finally with one process on each of two > > hosts. The > > first two runs are fine but the last run hangs on the second MPI_Barrier. > > > > [root@centos MPI]# /usr/local/bin/mpirun -np 2 --host centos a.out > > In rank 0 and host= centos Do Barrier call 1. > > In rank 1 and host= centos Do Barrier call 1. > > In rank 1 and host= centos Do Barrier call 2. > > In rank 1 and host= centos Do Barrier call 3. > > In rank 0 and host= centos Do Barrier call 2. > > In rank 0 and host= centos Do Barrier call 3. > > [root@centos MPI]# /usr/local/bin/mpirun -np 2 --host RAID a.out > > /root/.bashrc: line 14: unalias: ls: not found > > In rank 0 and host= RAID Do Barrier call 1. > > In rank 0 and host= RAID Do Barrier call 2. > > In rank 0 and host= RAID Do Barrier call 3. > > In rank 1 and host= RAID Do Barrier call 1. > > In rank 1 and host= RAID Do Barrier call 2. > > In rank 1 and host= RAID Do Barrier call 3. > > [root@centos MPI]# /usr/local/bin/mpirun -np 2 --host centos,RAID a.out > > /root/.bashrc: line 14: unalias: ls: not found > > In rank 0 and host= centos Do Barrier call 1. > > In rank 0 and host= centos Do Barrier call 2. > > In rank 1 and host= RAID Do Barrier call 1. > > In rank 1 and host= RAID Do Barrier call 2. > > > > Since it is such a simple test and problem and such a widely used MPI > > function, it must obviously > > be an installation or configuration problem. A pstack for each of the > > hung MPI_Barrier processes > > on the two machines shows this: > > > > [root@centos ~]# pstack 31666 > > #0 0x0000003baf0e8ee3 in __epoll_wait_nocancel () from /lib64/libc.so.6 > > #1 0x00007f5de06125eb in epoll_dispatch () from /usr/local/lib/libmpi.so.1 > > #2 0x00007f5de061475a in opal_event_base_loop () from > > /usr/local/lib/libmpi.so.1 > > #3 0x00007f5de0639229 in opal_progress () from /usr/local/lib/libmpi.so.1 > > #4 0x00007f5de0586f75 in ompi_request_default_wait_all () from > > /usr/local/lib/libmpi.so.1 > > #5 0x00007f5ddc59565e in ompi_coll_tuned_sendrecv_actual () from > > /usr/local/lib/openmpi/mca_coll_tuned.so > > #6 0x00007f5ddc59d8ff in ompi_coll_tuned_barrier_intra_two_procs () from > > /usr/local/lib/openmpi/mca_coll_tuned.so > > #7 0x00007f5de05941c2 in PMPI_Barrier () from /usr/local/lib/libmpi.so.1 > > #8 0x0000000000400a43 in main () > > > > [root@RAID openmpi-1.6.5]# pstack 22167 > > #0 0x00000030302e8ee3 in __epoll_wait_nocancel () from /lib64/libc.so.6 > > #1 0x00007f7ee46885eb in epoll_dispatch () from /usr/local/lib/libmpi.so.1 > > #2 0x00007f7ee468a75a in opal_event_base_loop () from > > /usr/local/lib/libmpi.so.1 > > #3 0x00007f7ee46af229 in opal_progress () from /usr/local/lib/libmpi.so.1 > > #4 0x00007f7ee45fcf75 in ompi_request_default_wait_all () from > > /usr/local/lib/libmpi.so.1 > > #5 0x00007f7ee060b65e in ompi_coll_tuned_sendrecv_actual () from > > /usr/local/lib/openmpi/mca_coll_tuned.so > > #6 0x00007f7ee06138ff in ompi_coll_tuned_barrier_intra_two_procs () from > > /usr/local/lib/openmpi/mca_coll_tuned.so > > #7 0x00007f7ee460a1c2 in PMPI_Barrier () from /usr/local/lib/libmpi.so.1 > > #8 0x0000000000400a43 in main () > > > > Which looks exactly the same on each machine. Any thoughts or ideas would > > be greatly appreciated as > > I am stuck. > > > > Clay Kirkland > > > > > > > > > > _______________________________________________ > > users mailing list > > us...@open-mpi.org > > http://www.open-mpi.org/mailman/listinfo.cgi/users > > -------------- next part -------------- > HTML attachment scrubbed and removed > > ------------------------------ > > Subject: Digest Footer > > _______________________________________________ > users mailing list > us...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/users > > ------------------------------ > > End of users Digest, Vol 2879, Issue 1 > ************************************** > > _______________________________________________ > users mailing list > us...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/users -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
