April 2014 - discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] credentials for mpiexec -f machinefile
by Jan Balewski 23 Apr '14

23 Apr '14

On Apr 23, 2014, at 5:10 PM, "Antonio J. Peña" <apenya(a)mcs.anl.gov> wrote: > No, what I meant was doing "ssh localhost", but now I look at your logs again, it may be a problem of your DNS configuration. > mpi-966f395e-bbb1-4a20-8cbc-c10081e91244 seems to be unable to resolve the IP address of mpi-be6bebee-55e3-4901-a5bb-637395ba46f6. Take a look at your /etc/hosts files in both hosts. > Right, I just figured that out as well. When VM's are launched the get assigned this funny names, not matching the DNS names. So I have manually adjusted the content of this 3 files (I run on SL6.5) vi /etc/hostname vi /etc/sysconfig/network vi /etc/hosts so the DNS name , public IP, and hostname --fqd return the same info, e.g. [cosy11@oswrk210 mpich-3.1]$ hostname --fqd oswrk210.lns.mit.edu Now the command below does not crashes but hangs for many minutes, eventually time-outing. So looks like I need to open port 60568 , right? After I shut down fire wall on IP=210 it complain abut port 45895 Next I shut down firewall in IP=207 and it complained abut new port 35167 Finally I shut down firewall between VM (on the openStack controller) and … IT WORKED: [cosy11@oswrk210 mpich-3.1]$ mpiexec -f machinefile2 -n 4 ./examples/cpi Process 0 of 4 is on oswrk210.lns.mit.edu Process 3 of 4 is on oswrk207.lns.mit.edu Process 1 of 4 is on oswrk210.lns.mit.edu Process 2 of 4 is on oswrk207.lns.mit.edu pi is approximately 3.1415926535899028, Error is 0.0000000000001097 wall clock time = 0.003602 QUESTION: Can you tell me what ports I need to have open on the master & worker VMs for mpiexec to distribute work on them? Jan

1 0

Re: [mpich-discuss] credentials for mpiexec -f machinefile
by "Antonio J. Peña" 23 Apr '14

23 Apr '14

No, what I meant was doing "ssh localhost", but now I look at your logs again, it may be a problem of your DNS configuration. mpi-966f395e-bbb1-4a20-8cbc-c10081e91244 seems to be unable to resolve the IP address of mpi-be6bebee-55e3-4901-a5bb-637395ba46f6. Take a look at your /etc/hosts files in both hosts. On 04/23/2014 03:57 PM, Jan Balewski wrote: > yes, from IP=212 I can do: > $ ssh cosy11(a)198.125.163.207 > Last login: Wed Apr 23 16:41:08 2014 from oswrk212.lns.mit.edu > > Jan > > > On Apr 23, 2014, at 4:54 PM, "Antonio J. Peï¿½a" <apenya(a)mcs.anl.gov> wrote: > >> You need to be able to ssh to the local host password-less as well. Is that working for you? >> >> >> On 04/23/2014 03:50 PM, Jan Balewski wrote: >>> On Apr 23, 2014, at 4:41 PM, "Antonio J. Peï¿½a" <apenya(a)mcs.anl.gov> >>> wrote: >>> >>> >>>> We use to encourage users to set up password-less ssh access by means of private/public key access. There are plenty of web pages with tutorials for this setup. This is just a random one: http://stackoverflow.com/questions/7260/how-do-i-setup-public-key-authentic… >>>> >>>> >>>> Antonio >>>> >>> Thanks Antonio. >>> I just did that. >>> >>> Now I'm able to use this setup: >>> $ cat machinefile >>> 198.125.163.212:2 >>> >>> to run on my login VM with IP=212: >>> $ mpiexec -f machinefile -n 2 ./examples/cpi >>> Process 0 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 >>> Process 1 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 >>> pi is approximately 3.1415926535899388, Error is 0.0000000000001457 >>> wall clock time = 0.003566 >>> >>> But if I add the remote host to the machine list: >>> $ cat machinefile2 >>> 198.125.163.212:2 >>> 198.125.163.207:2 >>> >>> Then some socket is not connected (see attached dump) >>> Do I need to enable some specific ports so IP2012 talks to IP=207? >>> >>> Is it just port 42548 ? TPC? UDP? a range of ports? >>> Do they need to be open on both the master & the worker? >>> Jan >>> >>> ----- >>> [cosy11@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 mpich-3.1]$ mpiexec -f machinefile2 -n 4 ./examples/cpi >>> [proxy:0:1@mpi-966f395e-bbb1-4a20-8cbc-c10081e91244] HYDU_sock_connect (utils/sock/sock.c:138): unable to get host address for mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 (1) >>> [proxy:0:1@mpi-966f395e-bbb1-4a20-8cbc-c10081e91244] main (pm/pmiserv/pmip.c:189): unable to connect to server mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 at port 42548 (check for firewalls!) >>> ^C[mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Sending Ctrl-C to processes as requested >>> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Press Ctrl-C again to force abort >>> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDU_sock_write (utils/sock/sock.c:286): write error (Bad file descriptor) >>> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmcd_pmiserv_send_signal (pm/pmiserv/pmiserv_cb.c:169): unable to write data to proxy >>> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] ui_cmd_cb (pm/pmiserv/pmiserv_pmci.c:79): unable to send signal downstream >>> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDT_dmxu_poll_wait_for_event (tools/demux/demux_poll.c:76): callback returned error status >>> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmci_wait_for_completion (pm/pmiserv/pmiserv_pmci.c:198): error waiting for event >>> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] main (ui/mpich/mpiexec.c:336): process manager error waiting for completion >>> >>> >>> >>> >>> >>> >>> >>> > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Antonio J. Peï¿½a Postdoctoral Appointee Mathematics and Computer Science Division Argonne National Laboratory 9700 South Cass Avenue, Bldg. 240, Of. 3148 Argonne, IL 60439-4847 apenya(a)mcs.anl.gov www.mcs.anl.gov/~apenya

1 0

Re: [mpich-discuss] credentials for mpiexec -f machinefile
by Jan Balewski 23 Apr '14

23 Apr '14

yes, from IP=212 I can do: $ ssh cosy11(a)198.125.163.207 Last login: Wed Apr 23 16:41:08 2014 from oswrk212.lns.mit.edu Jan On Apr 23, 2014, at 4:54 PM, "Antonio J. Peña" <apenya(a)mcs.anl.gov> wrote: > > You need to be able to ssh to the local host password-less as well. Is that working for you? > > > On 04/23/2014 03:50 PM, Jan Balewski wrote: >> On Apr 23, 2014, at 4:41 PM, "Antonio J. Peña" <apenya(a)mcs.anl.gov> >> wrote: >> >> >>> We use to encourage users to set up password-less ssh access by means of private/public key access. There are plenty of web pages with tutorials for this setup. This is just a random one: http://stackoverflow.com/questions/7260/how-do-i-setup-public-key-authentic… >>> >>> >>> Antonio >>> >> Thanks Antonio. >> I just did that. >> >> Now I'm able to use this setup: >> $ cat machinefile >> 198.125.163.212:2 >> >> to run on my login VM with IP=212: >> $ mpiexec -f machinefile -n 2 ./examples/cpi >> Process 0 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 >> Process 1 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 >> pi is approximately 3.1415926535899388, Error is 0.0000000000001457 >> wall clock time = 0.003566 >> >> But if I add the remote host to the machine list: >> $ cat machinefile2 >> 198.125.163.212:2 >> 198.125.163.207:2 >> >> Then some socket is not connected (see attached dump) >> Do I need to enable some specific ports so IP2012 talks to IP=207? >> >> Is it just port 42548 ? TPC? UDP? a range of ports? >> Do they need to be open on both the master & the worker? >> Jan >> >> ----- >> [cosy11@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 mpich-3.1]$ mpiexec -f machinefile2 -n 4 ./examples/cpi >> [proxy:0:1@mpi-966f395e-bbb1-4a20-8cbc-c10081e91244] HYDU_sock_connect (utils/sock/sock.c:138): unable to get host address for mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 (1) >> [proxy:0:1@mpi-966f395e-bbb1-4a20-8cbc-c10081e91244] main (pm/pmiserv/pmip.c:189): unable to connect to server mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 at port 42548 (check for firewalls!) >> ^C[mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Sending Ctrl-C to processes as requested >> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Press Ctrl-C again to force abort >> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDU_sock_write (utils/sock/sock.c:286): write error (Bad file descriptor) >> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmcd_pmiserv_send_signal (pm/pmiserv/pmiserv_cb.c:169): unable to write data to proxy >> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] ui_cmd_cb (pm/pmiserv/pmiserv_pmci.c:79): unable to send signal downstream >> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDT_dmxu_poll_wait_for_event (tools/demux/demux_poll.c:76): callback returned error status >> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmci_wait_for_completion (pm/pmiserv/pmiserv_pmci.c:198): error waiting for event >> [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] main (ui/mpich/mpiexec.c:336): process manager error waiting for completion >> >> >> >> >> >> >> >>

1 0

Re: [mpich-discuss] credentials for mpiexec -f machinefile
by "Antonio J. Peña" 23 Apr '14

23 Apr '14

You need to be able to ssh to the local host password-less as well. Is that working for you? On 04/23/2014 03:50 PM, Jan Balewski wrote: > On Apr 23, 2014, at 4:41 PM, "Antonio J. Peï¿½a" <apenya(a)mcs.anl.gov> wrote: > >> We use to encourage users to set up password-less ssh access by means of private/public key access. There are plenty of web pages with tutorials for this setup. This is just a random one: http://stackoverflow.com/questions/7260/how-do-i-setup-public-key-authentic… >> >> Antonio > Thanks Antonio. > I just did that. > > Now I'm able to use this setup: > $ cat machinefile > 198.125.163.212:2 > > to run on my login VM with IP=212: > $ mpiexec -f machinefile -n 2 ./examples/cpi > Process 0 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 > Process 1 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 > pi is approximately 3.1415926535899388, Error is 0.0000000000001457 > wall clock time = 0.003566 > > But if I add the remote host to the machine list: > $ cat machinefile2 > 198.125.163.212:2 > 198.125.163.207:2 > > Then some socket is not connected (see attached dump) > Do I need to enable some specific ports so IP2012 talks to IP=207? > > Is it just port 42548 ? TPC? UDP? a range of ports? > Do they need to be open on both the master & the worker? > Jan > > ----- > [cosy11@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 mpich-3.1]$ mpiexec -f machinefile2 -n 4 ./examples/cpi > [proxy:0:1@mpi-966f395e-bbb1-4a20-8cbc-c10081e91244] HYDU_sock_connect (utils/sock/sock.c:138): unable to get host address for mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 (1) > [proxy:0:1@mpi-966f395e-bbb1-4a20-8cbc-c10081e91244] main (pm/pmiserv/pmip.c:189): unable to connect to server mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 at port 42548 (check for firewalls!) > ^C[mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Sending Ctrl-C to processes as requested > [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Press Ctrl-C again to force abort > [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDU_sock_write (utils/sock/sock.c:286): write error (Bad file descriptor) > [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmcd_pmiserv_send_signal (pm/pmiserv/pmiserv_cb.c:169): unable to write data to proxy > [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] ui_cmd_cb (pm/pmiserv/pmiserv_pmci.c:79): unable to send signal downstream > [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDT_dmxu_poll_wait_for_event (tools/demux/demux_poll.c:76): callback returned error status > [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmci_wait_for_completion (pm/pmiserv/pmiserv_pmci.c:198): error waiting for event > [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] main (ui/mpich/mpiexec.c:336): process manager error waiting for completion > > > > > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Antonio J. Peï¿½a Postdoctoral Appointee Mathematics and Computer Science Division Argonne National Laboratory 9700 South Cass Avenue, Bldg. 240, Of. 3148 Argonne, IL 60439-4847 apenya(a)mcs.anl.gov www.mcs.anl.gov/~apenya

1 0

Re: [mpich-discuss] credentials for mpiexec -f machinefile
by Jan Balewski 23 Apr '14

23 Apr '14

On Apr 23, 2014, at 4:41 PM, "Antonio J. Peña" <apenya(a)mcs.anl.gov> wrote: > We use to encourage users to set up password-less ssh access by means of private/public key access. There are plenty of web pages with tutorials for this setup. This is just a random one: http://stackoverflow.com/questions/7260/how-do-i-setup-public-key-authentic… > > Antonio Thanks Antonio. I just did that. Now I'm able to use this setup: $ cat machinefile 198.125.163.212:2 to run on my login VM with IP=212: $ mpiexec -f machinefile -n 2 ./examples/cpi Process 0 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 Process 1 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 pi is approximately 3.1415926535899388, Error is 0.0000000000001457 wall clock time = 0.003566 But if I add the remote host to the machine list: $ cat machinefile2 198.125.163.212:2 198.125.163.207:2 Then some socket is not connected (see attached dump) Do I need to enable some specific ports so IP2012 talks to IP=207? Is it just port 42548 ? TPC? UDP? a range of ports? Do they need to be open on both the master & the worker? Jan ----- [cosy11@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 mpich-3.1]$ mpiexec -f machinefile2 -n 4 ./examples/cpi [proxy:0:1@mpi-966f395e-bbb1-4a20-8cbc-c10081e91244] HYDU_sock_connect (utils/sock/sock.c:138): unable to get host address for mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 (1) [proxy:0:1@mpi-966f395e-bbb1-4a20-8cbc-c10081e91244] main (pm/pmiserv/pmip.c:189): unable to connect to server mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 at port 42548 (check for firewalls!) ^C[mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Sending Ctrl-C to processes as requested [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Press Ctrl-C again to force abort [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDU_sock_write (utils/sock/sock.c:286): write error (Bad file descriptor) [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmcd_pmiserv_send_signal (pm/pmiserv/pmiserv_cb.c:169): unable to write data to proxy [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] ui_cmd_cb (pm/pmiserv/pmiserv_pmci.c:79): unable to send signal downstream [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDT_dmxu_poll_wait_for_event (tools/demux/demux_poll.c:76): callback returned error status [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmci_wait_for_completion (pm/pmiserv/pmiserv_pmci.c:198): error waiting for event [mpiexec@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] main (ui/mpich/mpiexec.c:336): process manager error waiting for completion

1 0

Re: [mpich-discuss] credentials for mpiexec -f machinefile
by "Antonio J. Peña" 23 Apr '14

23 Apr '14

We use to encourage users to set up password-less ssh access by means of private/public key access. There are plenty of web pages with tutorials for this setup. This is just a random one: http://stackoverflow.com/questions/7260/how-do-i-setup-public-key-authentic… Antonio On 04/23/2014 03:28 PM, Jan Balewski wrote: > Hi, > I was able to compile & install MPICH on a master virtual machine for > a user cosy11. > There are identical 2 VM running with IPs 198.125.163.212, 198.125.163.207 > > I picked IP=212 as my master node. > On this node I can run pi computation on two processes: > > cosy11@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 mpich-3.1]$ mpiexec -n > 2 ./examples/cpi > Process 1 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 > Process 0 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 > pi is approximately 3.1415926544231318, Error is 0.0000000008333387 > wall clock time = 0.000310 > > The next step of instruction (h) > http://www.mpich.org/static/downloads/3.1/mpich-3.1-README.txt > > tells me to define machinefile , below is my definition: > $ cat machinefile > 198.125.163.207:2 > 198.125.163.212:2 > > Note, I verified I can ssh to both machines from IP=212, but I must > type in the password > $ ssh cosy11(a)198.125.163.212 <mailto:[email protected]> > $ ssh cosy11(a)198.125.163.207 <mailto:[email protected]> > > Now, I try to compute pi on both VMs and mpiexec get stuck asking for > the password (see the screen shot) > mpiexec -f machinefile -n 4 ./examples/cpi > > I must somehow pass credentials for user=cosy11 to both VMs - how? > Have I missed this step of the instruction? > > Do I need to use ssh-key added to my ssh-add list on IP=212 to > make mpiexec to work with remote hosts? > Is there other solution you prefer? > Can you help me how to overcome that? > > Thanks > Jan > > > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Antonio J. Peï¿½a Postdoctoral Appointee Mathematics and Computer Science Division Argonne National Laboratory 9700 South Cass Avenue, Bldg. 240, Of. 3148 Argonne, IL 60439-4847 apenya(a)mcs.anl.gov www.mcs.anl.gov/~apenya

1 0

credentials for mpiexec -f machinefile
by Jan Balewski 23 Apr '14

23 Apr '14

Hi, I was able to compile & install MPICH on a master virtual machine for a user cosy11. There are identical 2 VM running with IPs 198.125.163.212, 198.125.163.207 I picked IP=212 as my master node. On this node I can run pi computation on two processes: cosy11@mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 mpich-3.1]$ mpiexec -n 2 ./examples/cpi Process 1 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 Process 0 of 2 is on mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 pi is approximately 3.1415926544231318, Error is 0.0000000008333387 wall clock time = 0.000310 The next step of instruction (h) http://www.mpich.org/static/downloads/3.1/mpich-3.1-README.txt tells me to define machinefile , below is my definition: $ cat machinefile 198.125.163.207:2 198.125.163.212:2 Note, I verified I can ssh to both machines from IP=212, but I must type in the password $ ssh cosy11(a)198.125.163.212 $ ssh cosy11(a)198.125.163.207 Now, I try to compute pi on both VMs and mpiexec get stuck asking for the password (see the screen shot) mpiexec -f machinefile -n 4 ./examples/cpi I must somehow pass credentials for user=cosy11 to both VMs - how? Have I missed this step of the instruction? Do I need to use ssh-key added to my ssh-add list on IP=212 to make mpiexec to work with remote hosts? Is there other solution you prefer? Can you help me how to overcome that? Thanks Jan

1 0

Re: [mpich-discuss] MPICH-3.1 make error: gcc: /lib/hwloc: No such file or directory
by Lu, Huiwei 22 Apr '14

22 Apr '14

Looks like it should be ‘configure --prefix’ instead of ‘configure -prefix’. — Huiwei Lu Postdoc Appointee Mathematics and Computer Science Division, Argonne National Laboratory http://www.mcs.anl.gov/~huiweilu/ On Apr 22, 2014, at 12:19 PM, Jan Balewski <balewski(a)MIT.EDU> wrote: > (try 6) > Hi, > I'm trying to install MPICH on SL6.5 for the first time and make crashed with the message: > ….. > make[3]: Nothing to be done for `all'. > make[3]: Leaving directory `/tmp/mpich-3.1/src/mpl' > Making all in tools/topo/hwloc/hwloc > make[3]: Entering directory `/tmp/mpich-3.1/src/pm/hydra/tools/topo/hwloc/hwloc' > Making all in src > make[4]: Entering directory `/tmp/mpich-3.1/src/pm/hydra/tools/topo/hwloc/hwloc/src' > CC topology.lo > gcc: /lib/hwloc": No such file or directory > <command-line>: warning: missing terminating " character > make[4]: *** [topology.lo] Error 1 > make[4]: Leaving directory `/tmp/mpich-3.1/src/pm/hydra/tools/topo/hwloc/hwloc/src' > make[3]: *** [all-recursive] Error 1 > make[3]: Leaving directory `/tmp/mpich-3.1/src/pm/hydra/tools/topo/hwloc/hwloc' > make[2]: *** [all-recursive] Error 1 > make[2]: Leaving directory `/tmp/mpich-3.1/src/pm/hydra' > make[1]: *** [all-recursive] Error 1 > make[1]: Leaving directory `/tmp/mpich-3.1' > make: *** [all] Error 2 > ….. > > My system is: > $ uname -a > Linux localhost.localdomain.localdomain 2.6.32-431.11.2.el6.x86_64 #1 SMP Tue Mar 25 11:15:18 CDT 2014 x86_64 x86_64 x86_64 GNU/Linux > > I compile MPICH as regular user. > > Attached are c.txt & m.txt > > Please advice how to proceed > Thanks in advance > Jan > <aaa.tar.gz>_______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss

1 0

MPICH-3.1 make error: gcc: /lib/hwloc: No such file or directory
by Jan Balewski 22 Apr '14

22 Apr '14

(try 6) Hi, I'm trying to install MPICH on SL6.5 for the first time and make crashed with the message: ….. make[3]: Nothing to be done for `all'. make[3]: Leaving directory `/tmp/mpich-3.1/src/mpl' Making all in tools/topo/hwloc/hwloc make[3]: Entering directory `/tmp/mpich-3.1/src/pm/hydra/tools/topo/hwloc/hwloc' Making all in src make[4]: Entering directory `/tmp/mpich-3.1/src/pm/hydra/tools/topo/hwloc/hwloc/src' CC topology.lo gcc: /lib/hwloc": No such file or directory <command-line>: warning: missing terminating " character make[4]: *** [topology.lo] Error 1 make[4]: Leaving directory `/tmp/mpich-3.1/src/pm/hydra/tools/topo/hwloc/hwloc/src' make[3]: *** [all-recursive] Error 1 make[3]: Leaving directory `/tmp/mpich-3.1/src/pm/hydra/tools/topo/hwloc/hwloc' make[2]: *** [all-recursive] Error 1 make[2]: Leaving directory `/tmp/mpich-3.1/src/pm/hydra' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/tmp/mpich-3.1' make: *** [all] Error 2 ….. My system is: $ uname -a Linux localhost.localdomain.localdomain 2.6.32-431.11.2.el6.x86_64 #1 SMP Tue Mar 25 11:15:18 CDT 2014 x86_64 x86_64 x86_64 GNU/Linux I compile MPICH as regular user. Attached are c.txt & m.txt Please advice how to proceed Thanks in advance Jan

1 0

Re: [mpich-discuss] Running mpi from bash script
by Francisco Pastor 22 Apr '14

22 Apr '14

Hi The application is RAMS meteorological model (http://www.atmet.com) mostly build in Fortran and some C. Compile options for RAMS are |MAKE=/usr/bin/make # libraries. LIBUTILS=./utils/libutils-$(UTILS_VERSION)-opt.a # Activate appropriate parts below, comment out others. #--------------------------------------------------------------- # NCAR Graphics. #--------------------------------------------------------------- #--------------------------------------------------------------- # If you are using a real distribution of NCAR Graphics... NCARG_DIR=/usr/local/ncarg-4.4.2/lib LIBNCARG=-L$(NCARG_DIR) -lncarg -lncarg_gks -lncarg_c -L/usr/local/ncarg-4.4.2/ncarg2d/src/libncarg_gks \ -L/usr/lib/i386-linux-gnu -lpthread -ldl -lc -lgcc_eh -lX11 #--------------------------------------------------------------- #--------------------------------------------------------------- # HDF libraries #--------------------------------------------------------------- HDF5_LIBS=-L/usr/local/hdf5/lib -lhdf5 -lz -lm HDF5_INCS=/usr/local/hdf5/include HDF5_DEFS=-DH5_USE_16_API #--------------------------------------------------------------- # Machine-dependent options. #--------------------------------------------------------------- #----------------- LINUX Portland Group pgf90/gcc --------------- CMACH=PC_LINUX1 F_COMP=/opt/intel/bin/ifort F_OPTS=-O3 -FR -fpp -assume byterecl -DRAMS_SUPPORT C_COMP=/opt/intel/bin/icc C_OPTS=-O3 -DUNDERSCORE -DLITTLE LOADER=/opt/intel/bin/ifort LOADER_OPTS= LIBS= ARCHIVE=ar rs| MPI_PATH=/usr/bin PAR_INCS=-I/usr/local/include PAR_LIBS=-L/usr/local/lib/ -lmpich PAR_DEFS=-DRAMS_MPI Regarding mpich2 path I installed from Ubuntu repositories with apt-get. Previously I just removed everything I found from mpi but can't be sure there was not anything left. When running in parallel from batch I get some processes running on the master node meteo 8288 7559 0 13:03 ? 00:00:00 time /usr/bin/mpirun -n 40 -f machinefile ./rams60 -f RAMSIN.operatiu meteo 8289 8288 0 13:03 ? 00:00:00 /usr/bin/mpirun -n 40 -f machinefile ./rams60 -f RAMSIN.operatiu meteo 8295 8290 0 13:03 ? 00:00:00 ./rams60 -f RAMSIN.operatiu meteo 8296 8290 0 13:03 ? 00:00:00 ./rams60 -f RAMSIN.operatiu meteo 8297 8290 0 13:03 ? 00:00:00 ./rams60 -f RAMSIN.operatiu meteo 8298 8290 0 13:03 ? 00:00:00 ./rams60 -f RAMSIN.operatiu meteo 8299 8290 0 13:03 ? 00:00:00 ./rams60 -f RAMSIN.operatiu meteo 8300 8290 0 13:03 ? 00:00:00 ./rams60 -f RAMSIN.operatiu meteo 8301 8290 0 13:03 ? 00:00:00 ./rams60 -f RAMSIN.operatiu meteo 8302 8290 0 13:03 ? 00:00:00 ./rams60 -f RAMSIN.operatiu When running mpirun from command line processes start in all nodes. There is still another problem that could or not be related to mpich, met model returns NaN where there should be numbers (values for meteorological variables) but maybe this has a completely different origin as initial data,... Thanks again El 22/04/2014 16:41, Reuti escribiï¿½: > Am 22.04.2014 um 09:18 schrieb Francisco Pastor: > >> Hi and sorry for the delay in answering. I was out on holidays since Friday. >> >> I recompiled the appication, just changing mpich2 path. > Fine. > > >> MPI_PATH=/usr/bin >> PAR_INCS=-I/usr/local/include >> PAR_LIBS=-L/usr/local/lib/ -lmpich > Is it a C application, or are you using the wrapper during compliation to add also the Fortran interface? > > >> PAR_DEFS=-DRAMS_MPI > Do you have more than one MPI library at this location? Often it's better to have a dedicated directory where you can check the used path to it instantly. > > Is there any output whether the application is now running in parallel, or executing a serial run N times only? > > -- Reuti > > >> Thanks >> >> El 17/04/2014 17:58, Reuti escribiï¿½: >>> Hi, >>> >>> Am 17.04.2014 um 11:35 schrieb Francisco Pastor: >>> >>>> The point is that I am using a program, meteorological model RAMS, that was running fine with this mpich2 version. And it runs fine from the command line. >>>> >>>> If the system works, should I upgrade? Will it work with my program? Can I install two different versions at the same time or should I delete the older one? Is there any chance to run >>> Well, using a new version usually means to recompile the application to gain advantage of it. Did you compile it on your own, or is this a binary you downloaded somewhere? >>> >>> -- Reuti >>> >>> >>>> Thanks >>>> >>>> El 17/04/2014 11:06, Reuti escribiï¿½: >>>>> Hi, >>>>> >>>>> Am 17.04.2014 um 10:04 schrieb Francisco Pastor: >>>>> >>>>>> I need to daily run an mpi job from a bash script. I have prepared the script and run from the command line, this works fine. But when I try to execute at a later time with at (at -f my_script now +1 minute) the parallel processes do not start. >>>>>> >>>>>> These are the significant parts of my script: >>>>>> >>>>>> # path to mpi executables >>>>>> ruta_mpi=/usr/local/mpich2-1.0.5p4/bin >>>>> This version of MPICH is quite old. Please use a more recent one, where the necessity for starting an MPD beforehand does no longer exist. >>>>> >>>>> -- Reuti >>>>> >>>>> >>>>>> # Function that checks if mpd processes are ok (checks mpdtrace output, five nodes should be up) >>>>>> Comprueba_MPD() >>>>>> { >>>>>> mpdtrace > log.mpd >>>>>> wc -l log.mpd > check.mpd >>>>>> check1=`awk '{ print $1 }' check.mpd` >>>>>> if [ $check1 -eq 5 ]; >>>>>> then >>>>>> echo "Salida de mpdtrace correcta, hosts=$check1" >>>>>> else >>>>>> echo "Salida de mpdtrace incorrecta, hosts=$check1" >>>>>> $ruta_mpi/mpdboot -n 5 & >>>>>> sleep 10 >>>>>> mpdtrace >>>>>> fi >>>>>> } >>>>>> >>>>>> >>>>>> # Run rams60 program with mpirun >>>>>> time $ruta_mpi/mpirun -n 21 ./rams60 -f RAMSIN.operatiu >>>>>> >>>>>> I am sure missing something but I can't find out what's going on. >>>>>> >>>>>> Thanks in advance >>>>>> >>>>>> -- >>>>>> ----------- >>>>>> >>>>>> -- ----------- Dr. Francisco Pastor Meteorology department, Instituto Universitario CEAM-UMH http://www.ceam.es ----------- Mendeley profile: http://www.mendeley.com/profiles/francisco-pastor1/ Google Scholar: http://scholar.google.com/citations?user=V3mmCdkAAAAJ&hl=es Researcher ID: http://www.researcherid.com/rid/B-8331-2008 Cosis profile: http://www.cosis.net/profile/francisco.pastor ----------- mail: paco(a)ceam.es skype: paco.pastor.guzman ----------- Parque Tecnologico, C/ Charles R. Darwin, 14 46980 PATERNA (Valencia), Spain Tlf. 96 131 82 27 - Fax. 96 131 81 90 --------------------------------------------------------------------- Este mensaje y los ficheros anexos son confidenciales. Los mismos contienen informaciï¿½n reservada de la empresa que no puede ser difundida. Si usted ha recibido este correo por error, tenga la amabilidad de eliminarlo de su sistema y avisar al remitente mediante reenvï¿½o a su direcciï¿½n electrï¿½nica; no deberï¿½ copiar el mensaje ni divulgar su contenido a ninguna persona. Su direcciï¿½n de correo electrï¿½nico junto a sus datos personales forman parte de un fichero titularidad de la Fundaciï¿½n de la Comunidad Valenciana Centro de Estudios Ambientales del Mediterrï¿½neo - CEAM, con CIF: G-46957213, cuya finalidad es la de mantener el contacto con Ud. De acuerdo con la Ley Orgï¿½nica 15/1999, usted puede ejercitar sus derechos de acceso, rectificaciï¿½n, cancelaciï¿½n y, en su caso, oposiciï¿½n enviando una solicitud por escrito, acompaï¿½ada de una fotocopia de su DNI dirigida a: Fundaciï¿½n de la Comunidad Valenciana Centro de Estudios Ambientales del Mediterrï¿½neo - CEAM. C/ Charles R. Darwin, 14. Parque Tecnolï¿½gico.46980 PATERNA (Valencia). This message and the attached files are confidential. They contain reserved information belonging to our centre and are not to be broadcast. If you have received this email by mistake, please delete it from your system and alert the sender by returning it to his/her email address. You must not copy or divulge the contents of the message to anyone. Your email address and personal data are included in a file belonging to the Fundaciï¿½n de la Comunidad Valenciana Centro de Estudios Ambientales del Mediterrï¿½neo - CEAM, con CIF: G-46957213. The purpose of this file is to allow us to keep in contact with you. In accordance with Organic Law 15/1999, you are permitted to access, rectify, cancel or oppose the contents of this file by submitting a written request, accompanied by a photocopy of your DNI, to: Fundaciï¿½n de la Comunidad Valenciana Centro de Estudios Ambientales del Mediterrï¿½neo - CEAM. C/ Charles R. Darwin, 14. Parque Tecnolï¿½gico.46980 PATERNA (Valencia).

1 0