[mpich-discuss] Working with TotalView & mpiexec.hydra

Pavan Balaji balaji at mcs.anl.gov
Tue Feb 28 09:54:48 CST 2012


Hmm.. This seems weird.  I think I might have seen such as error with 
csh earlier, which was unresolved.  Are you using csh/tcsh?

Can you try the following few things to help narrow down the problem?

1. Try a different debugger instead of totalview.  Replace tv with ddd 
for example.

2. Try setting the environment variable HYDRA_LAUNCHER_AUTOFORK=0 before 
mpiexec.

Thanks,

  -- Pavan

On 02/28/2012 08:51 AM, Anatoly G wrote:
> Dear MPICH2.
> I want to debug my MPI program. (files attached).
>
> *Execution line without TotalView is (works fine):*
> /space/local/bin/mpiexec.hydra -genvall -disable-auto-cleanup -f
> MpiConfigMachines.txt -n 1 -launcher=rsh mpi_rcv_waitany 100 1000000 90
> 1000 1 logs/res1_ : -n 1 mpi_rcv_waitany 100 1000000 90 1000 1
> logs/res2_ : -n 1 mpi_rcv_waitany 100 1000000 90 1000 1 logs/res3_
>
> *Execution line withTotalView is (fails):*
> /space/local/bin/mpiexec.hydra -genvall -disable-auto-cleanup -f
> MpiConfigMachines.txt -n 1 -launcher=rsh tv mpi_rcv_waitany -a 100
> 1000000 90 1000 1 logs/res1_ : -n 1 mpi_rcv_waitany 100 1000000 90 1000
> 1 logs/res2_ : -n 1 mpi_rcv_waitany 100 1000000 90 1000 1 logs/res3_
>
> TotalView is executed. Shows text of program. I start debugging using
> "Next" commands until I arrive to line:
> MPI::Init(argc, argv);
>
> Executing this line cause failure of program, but totalview is still up.
> _In shell I see an error_:
>
> [cli_0]: write_line error; fd=6 buf=:cmd=init pmi_version=1 pmi_subversion=1
> :
> system msg for write_line failure : Bad file descriptor
> [cli_0]: Unable to write to PMI_fd
> [cli_0]: write_line error; fd=6 buf=:cmd=get_appnum
> :
> system msg for write_line failure : Bad file descriptor
> Fatal error in MPI_Init: Other MPI error, error stack:
> MPIR_Init_thread(411):
> MPID_Init(138).......: channel initialization failed
> MPID_Init(423).......: PMI_Get_appnum returned -1
>
>
>
> Can you please help me with this problem.
>
> Anatoly.
>
>
> _______________________________________________
> mpich-discuss mailing list     mpich-discuss at mcs.anl.gov
> To manage subscription options or unsubscribe:
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss

-- 
Pavan Balaji
http://www.mcs.anl.gov/~balaji


More information about the mpich-discuss mailing list