[mpich-discuss] credentials for mpiexec -f machinefile

"Antonio J. Peña" apenya at mcs.anl.gov
Wed Apr 23 15:54:46 CDT 2014


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 at 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-authentication
>>
>>    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 at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6 mpich-3.1]$ mpiexec -f machinefile2 -n 4 ./examples/cpi
> [proxy:0:1 at 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 at 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 at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Sending Ctrl-C to processes as requested
> [mpiexec at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] Press Ctrl-C again to force abort
> [mpiexec at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDU_sock_write (utils/sock/sock.c:286): write error (Bad file descriptor)
> [mpiexec at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmcd_pmiserv_send_signal (pm/pmiserv/pmiserv_cb.c:169): unable to write data to proxy
> [mpiexec at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] ui_cmd_cb (pm/pmiserv/pmiserv_pmci.c:79): unable to send signal downstream
> [mpiexec at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYDT_dmxu_poll_wait_for_event (tools/demux/demux_poll.c:76): callback returned error status
> [mpiexec at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] HYD_pmci_wait_for_completion (pm/pmiserv/pmiserv_pmci.c:198): error waiting for event
> [mpiexec at mpi-be6bebee-55e3-4901-a5bb-637395ba46f6] main (ui/mpich/mpiexec.c:336): process manager error waiting for completion
>
>
>
>
>
>
> _______________________________________________
> discuss mailing list     discuss at mpich.org
> To manage subscription options or unsubscribe:
> https://lists.mpich.org/mailman/listinfo/discuss


-- 
Antonio J. Peña
Postdoctoral Appointee
Mathematics and Computer Science Division
Argonne National Laboratory
9700 South Cass Avenue, Bldg. 240, Of. 3148
Argonne, IL 60439-4847
apenya at mcs.anl.gov
www.mcs.anl.gov/~apenya

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpich.org/pipermail/discuss/attachments/20140423/3530365a/attachment.html>


More information about the discuss mailing list