Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] strange error, seems inable to launch job
From: Mr Yann JOBIC (jobic_at_[hidden])
Date: 2009-02-12 04:21:40


Ralph Castain wrote:
> Never seen that one before, though we did put the test in there "just
> in case".
>
> Basically, this message indicates that the system returned an error
> when we attempted to dup2 the stdout/err file descriptors so we could
> setup a pipe by which we can forward that output to you.
>
> I have no idea why the system would refuse the request. You might
> check your fd limits to see if they are strangely low, or write a
> little test program to see if there is something strange about pipe
> permissions.
>
> Most likely thing I can think of, though, is that you possibly are
> picking up incompatible libraries, or something got crosswise during
> build.
>
> Ralph
Thanks for trying to help me.

I've got in this file : /usr/include/linux/limits.h
#define NR_OPEN 1024
Is it ok ?
Trixy03:~# cat /proc/sys/fs/file-max
781095
Seems fine no ?

How can i write a test code for the pipe permissions? May i've something
wrong in my /dev ? (right permissions ?)

For the library part, i don't know how to check that. On linux, every
libs are in /usr/lib.

Yann

>
> On Feb 11, 2009, at 3:10 AM, Mr Yann JOBIC wrote:
>
>> Hello all,
>>
>> I compiled ompi v1.3 (tarball) with the intel compiler on debian
>> etch. Everything went fine, thanks for the FAQ (quite complet)
>>
>> But, when i'm running a job, i've got this error :
>> Trixy03-jobic% mpirun --verbose --debug-daemons -np 4 ./exe
>> [Trixy03:15140] [[19525,0],0] orted_cmd: received add_local_procs
>> [Trixy03:15140] [[19525,0],0] node[0].name Trixy03 daemon 0 arch
>> ffc91200
>> --------------------------------------------------------------------------
>>
>> mpirun was unable to launch the specified application as it
>> encountered an error:
>>
>> Error: pipe function call failed when setting up I/O forwarding
>> subsystem
>> Node: Trixy03
>>
>> while attempting to start process rank 0.
>> --------------------------------------------------------------------------
>>
>>
>> I don't understand what's going on, and how to debug...
>> I compiled mpich1, and i can successfully launch a job.
>>
>> Have you got some ideas of what's going on ?
>>
>> Many thanks,
>>
>> Yann
>>
>> PS : Some possible interesting information :
>> Open MPI SVN revision: r20295
>> Open MPI release date: Jan 19, 2009
>> Open RTE: 1.3
>> Build CFLAGS: -DNDEBUG -mp1 -m64 -O3 -fno-alias -msse3 -static-intel
>> -finline-functions -fno-strict-aliasing
>> -restrict -fexceptions -pthread
>> -fvisibility=hidden
>> Build CXXFLAGS: -DNDEBUG -mp1 -m64 -O3 -fno-alias -msse3
>> -static-intel -finline-functions -fexceptions -pthread
>> Build FFLAGS: -mp1 -m64 -O3 -fno-alias -msse3 -static-intel -fexceptions
>> Build FCFLAGS: -mp1 -m64 -O3 -fno-alias -msse3 -static-intel
>> -fexceptions -fexceptions
>> Build LDFLAGS: -export-dynamic -fexceptions
>> Build LIBS: -lnsl -lutil Wrapper extra CFLAGS: -fexceptions -pthread
>> Wrapper extra CXXFLAGS: -fexceptions -pthread
>> Wrapper extra FFLAGS: -fexceptions
>> Wrapper extra FCFLAGS: -fexceptions
>> Wrapper extra LDFLAGS: Wrapper extra LIBS: -ldl
>> -Wl,--export-dynamic -lnsl -lutil
>>
>>

-- 
___________________________
Yann JOBIC
HPC engineer
Polytech Marseille DME
IUSTI-CNRS UMR 6595
Technopôle de Château Gombert
5 rue Enrico Fermi
13453 Marseille cedex 13
Tel : (33) 4 91 10 69 39
  ou  (33) 4 91 10 69 43
Fax : (33) 4 91 10 69 69