Aargh, that was the reason. I compiled openmpi 1.10.3,and now it works too at my side. Thanks a lot for the hint.
Now a hint from my side if somebody runs into the same problem: If you have compiled the program with mpicc-2.0.0, you'll propably now get an error like: ompi_mpi_init: ompi_rte_init failed --> Returned "(null)" (-43) instead of "Success" (0) -------------------------------------------------------------------------- *** An error occurred in MPI_Init *** on a NULL communicator *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort, *** and potentially your MPI job) [karun01:26952] Local abort before MPI_INIT completed completed successfully, but am not able to aggregate error messages, and not able to guarantee that all other processes were killed! Just recompile the program with mpicc-1.10.3 and it will work. Again thanks a lot for everybody who tried to help, especially Reuti, Hugh and William. With kind reagrds, ulrich On 08/11/2016 12:16 PM, Reuti wrote: > I just compiled openmpi-2.0.0 on my own and it looks like a regression to use > `ssh` although it's running under SGE. Also for `mpicc` it was necessary to > supply "-ldl" to succeed, this wasn't necessary in former versions. > > I'll look into it. > > For now I think it's best to stay with 1.10.3. > > Note that after 1.6.5 they do a core binding (bad in case if several Open MPI > jobs are running on one and the same node, as all will use core 0 upwards) > and check the network topology. If it's set up with dead routes/interfaces > (which normally won't matter), the startup of the parallel job may be delayed > by one to two minutes (until they face a timeout). > > -- Reuti > > >> Am 10.08.2016 um 21:15 schrieb Ulrich Hiller <hil...@mpia-hd.mpg.de>: >> >> Hello, >> >> My problem: How can i make gridengine not to use ssh? >> >> Installed: >> openmpi-2.0.0 - configured with sge support. >> gridengine (son of gridengine) 8.1.9-1 >> >> I have a simple openmpi program 'teste' which only gives "hello world" >> output. >> I start it with: >> qsub -pe orte 160 -V -j yes -cwd -S /bin/bash <<< "mpiexec -n 160 teste >>>> /home/ljohndoe/out.dat" >> on the master node. >> I get back the error: >> >> Host key verification failed. >> Host key verification failed. >> Permission denied, please try again. >> Permission denied, please try again. >> Received disconnect from 192.168.117.6: 2: Too many authentication >> failures for johndoe >> Permission denied, please try again. >> Permission denied, please try again. >> Received disconnect from 192.168.117.5: 2: Too many authentication >> failures for johndoe >> [...] >> >> When i configure a passwordless ssh login to the execute nodes >> (exchanging the ssh key from master with 'ssh-copy-id), it works like >> charm. So it obviuously uses ssh connection to the execute nodes. >> >> the output of 'qconf -sconf' contains: >> >> login_shells sh,bash,ksh,csh,tcsh >> qlogin_command builtin >> qlogin_daemon builtin >> rlogin_command builtin >> rlogin_daemon builtin >> rsh_command builtin >> rsh_daemon builtin >> >> (as far as i read this was the problem of a thread some time ago in this >> list. But i seem to have the correct values) >> >> So everything should be fine- or not? >> Also with >> qlogin -l 'h=exec01' >> and >> qrsh -l 'h=exec01' >> i can go without problems to the first node.(called exec01), and i can >> also login to all other execute nodes as well. >> >> Is there anywhere another 'switch' where i can let qsub run _not_ over ssh? >> >> If is is of interest, the output of 'qconf -sp orte' is: >> pe_name orte >> slots 9999999 >> user_lists NONE >> xuser_lists NONE >> start_proc_args NONE >> stop_proc_args NONE >> allocation_rule $round_robin >> control_slaves FALSE >> job_is_first_task TRUE >> urgency_slots min >> accounting_summary FALSE >> qsort_args NONE >> >> Also, i do not have any ssh lines in ~/.profile or ~/.bashrc >> >> >> Kind regards, ulrich >> >> >> >> >> >> >> _______________________________________________ >> users mailing list >> users@gridengine.org >> https://gridengine.org/mailman/listinfo/users > _______________________________________________ users mailing list users@gridengine.org https://gridengine.org/mailman/listinfo/users
