Re: [Pytom-mail] Pytom-mail post from khannakanika11@gmail.com requires approval

[Dustin M. <dus...@gm...>]

You need to do this command in your terminal due to something in mpi4py. 

bourne-shell (bash, zsh, etc.)
export LD_PRELOAD='<path to libmpi.so>'

csh (csh, tcsh, etc.)
setenv LD_PRELOAD '<path to libmpi.so>'

where the path to the library is something like ‘/usr/lib/openmpi/lib/libmpi.so’ on my ubuntu 14.04 system.

If you forget again (I always do), it’s also mentioned in the pytom FAQ

—
Cheers,
Dustin

> On Feb 24, 2016, at 4:37 PM, Kanika Khanna <kk...@uc...> wrote:
> 
> Hello,
> 
> So I am trying localization with the tutorial data set itself. I did not use the hostfile as I am running on only one host. Hence, I specified the host and no. of CPUs as per the following command
> 
> mpirun --host 'HostName' -c 4 pytom /usr/local/pytom/bin/localization.py 'PathToJobFile'/job.xml 2 2 2
> 
> I get the following error:
> 
> It looks like orte_init failed for some reason; your parallel process is
> likely to abort.  There are many reasons that a parallel process can
> fail during orte_init; some of which are due to configuration or
> environment problems.  This failure appears to be an internal failure;
> here's some additional information (which may only be relevant to an
> Open MPI developer):
> 
>   opal_init failed
>   --> Returned value Error (-1) instead of ORTE_SUCCESS
> --------------------------------------------------------------------------
> *** 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)
> [pordoi:29128] Local abort before MPI_INIT completed successfully; not able to aggregate error messages, and not able to guarantee that all other processes were killed!
> 
> 
> On Tue, Feb 23, 2016 at 3:44 PM, Dustin Morado <dus...@gm...> wrote:
> Hi Kanika,
> 
> What was the command you ran exactly? It looks like there’s an error with your host file. Are you running this on a cluster, or just a workstation with openmpi?
> 
> If it’s the latter you can simply run it like this:
> mpirun --hostfile localhost -c 8 pytomlocalization.py job.xml 2 2 2
> 
> which will send one of each of the 8 chunks of the tomogram to a processor.
> 
> If you are on a cluster, you might want to ask the cluster admin if there’s a specific submit script you need to run to have access to the correct default hostfile.
> 
> —
> Cheers,
> Dustin
> 
> 
> > Thanks you for that. When I run the process with openmpi, i get the following
> > error message:
> > Open RTE detected a parse error in the hostfile: /home/kkhanna/Documents/Pytom/Bacillus_ribosome/job.xml It occured on line number 37 on token 1.
> > —————————————————————————————————————
> > [pordoi:25878] [[4037,0],0] ORTE_ERROR_LOG: Error in file base/rmaps_base_support_fns.c at line 176
> > [pordoi:25878] [[4037,0],0] ORTE_ERROR_LOG: Error in file rmaps_rr.c at line 121
> > [pordoi:25878] [[4037,0],0] ORTE_ERROR_LOG: Error in file base/rmaps_base_map_job.c at line 379
> > [pordoi:25878] 2 more processes have sent help message help-hostfile.txt / parse_error
> > [pordoi:25878] Set MCA parameter "orte_base_help_aggregate" to 0 to see all help / error messages
>