[R-sig-hpc] Rmpi does only finds only master

[Rainer M Krug]

Just an additional info: I was just told to use qsub to submit the
job, and it was suggested to use the following script:


#!/bin/bash
#$ -pe openmpi_rr 4
#$ -cwd -j y
orterun -n $NSLOTS -h $TMPDIR/machines TheCode.r

Rainer


On Thu, Nov 27, 2008 at 11:20 AM, Rainer M Krug <r.m.krug at gmail.com> wrote:
> On Wed, Nov 26, 2008 at 6:26 PM, Markus Schmidberger
> <schmidb at ibe.med.uni-muenchen.de> wrote:
>> Hi Rainer,
>>
>> how do you start R?
>
> Hi Hao, Markus
>
> Both of you were spot on - the problem was that I have to orterun or
> mpirun to start the job.
> But I still have a problem: on the cluster, I have to use
>
> module add openmpi_gnu
>
> to adjust the paths to make the openmpi implementation visible (and usable).
> But this is only done on the head, and not on the nodes. Is it somhow
> possible to execute that command on each node on which snow creates a
> node?
>
> below the code I use and the output:
>
> When I try to execute code in R, I get the following error:
>
> rkrug at head001:~/Documents/examples/R> orterun -n 1 --hostfile ~/MPIhostfile R.r
> bash: orted: command not found
> [head001:28028] [0,0,0] ORTE_ERROR_LOG: Timeout in file
> base/pls_base_orted_cmds.c at line 275
> [head001:28028] [0,0,0] ORTE_ERROR_LOG: Timeout in file
> pls_rsh_module.c at line 1166
> [head001:28028] [0,0,0] ORTE_ERROR_LOG: Timeout in file errmgr_hnp.c at line 90
> [head001:28028] ERROR: A daemon on node comp001 failed to start as expected.
> [head001:28028] ERROR: There may be more information available from
> [head001:28028] ERROR: the remote shell (see above).
> [head001:28028] ERROR: The daemon exited unexpectedly with status 127.
> [head001:28028] [0,0,0] ORTE_ERROR_LOG: Timeout in file
> base/pls_base_orted_cmds.c at line 188
> [head001:28028] [0,0,0] ORTE_ERROR_LOG: Timeout in file
> pls_rsh_module.c at line 1198
> --------------------------------------------------------------------------
> orterun was unable to cleanly terminate the daemons for this job.
> Returned value Timeout instead of ORTE_SUCCESS.
> --------------------------------------------------------------------------
> rkrug at head001:~/Documents/examples/R>
>
>
> The file R.r:
> #################
> ## R.r
> #!/export/home/rkrug/bin/r
> library(Rmpi)
> library(snow)
>
>
>
>
>
>                                                 cl <-
> makeMPIcluster(4)
> res <- clusterCall(
>        cl,
>        function() {
>                Sys.info()["nodename"]
>        }
>        )
> print(do.call(rbind,res))
>
>
>
>
> My idea was to
> "add module openmpi_gnu
>
> for each node which the file R.r is using, but as the file R.sub is
> only executed on one node and then, within R.r, starts the processes
> on the other nodes, this does obviously not work.
>
>
> Thanks
>
> Rainer
>
>
>>
>> Try this:
>> orterun -n 1 --hostfile YOUR_HOSTFILE R --no-save
>>
>> And have a look to Dirk's tutorial of the UseR2008:
>> http://www.statistik.uni-dortmund.de/useR-2008/tutorials/useR2008introhighperfR.pdf
>>
>> Best
>> Markus
>>
>> Rainer M Krug wrote:
>>> Sorry - forgot the version and platform info:
>>>
>>>> version
>>>                _
>>> platform       x86_64-unknown-linux-gnu
>>> arch           x86_64
>>> os             linux-gnu
>>> system         x86_64, linux-gnu
>>> status
>>> major          2
>>> minor          7.2
>>> year           2008
>>> month          08
>>> day            25
>>> svn rev        46428
>>> language       R
>>> version.string R version 2.7.2 (2008-08-25)
>>>
>>> Open MPI 1.2.5
>>>
>>> On Wed, Nov 26, 2008 at 10:19 AM, Rainer M Krug <r.m.krug at gmail.com> wrote:
>>>> Hi
>>>>
>>>> I am trying to use Rmpi (and I managed to install it against openMPI
>>>> and it is loading), but it sees only the master, not the nodes of the
>>>> cluster.
>>>> MPI is working, as a c program successfully communicates with all nodes.
>>>>
>>>> Thanks for suggestions,
>>>>
>>>> Rainer
>>>>
>>>> --
>>>> Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation
>>>> Biology, UCT), Dipl. Phys. (Germany)
>>>>
>>>> Centre of Excellence for Invasion Biology
>>>> Faculty of Science
>>>> Natural Sciences Building
>>>> Private Bag X1
>>>> University of Stellenbosch
>>>> Matieland 7602
>>>> South Africa
>>>>
>>>
>>>
>>>
>>
>>
>> --
>> Dipl.-Tech. Math. Markus Schmidberger
>>
>> Ludwig-Maximilians-Universität München
>> IBE - Institut für medizinische Informationsverarbeitung,
>> Biometrie und Epidemiologie
>> Marchioninistr. 15, D-81377 Muenchen
>> URL: http://www.ibe.med.uni-muenchen.de
>> Mail: Markus.Schmidberger [at] ibe.med.uni-muenchen.de
>> Tel: +49 (089) 7095 - 4599
>>
>
>
>
> --
> Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation
> Biology, UCT), Dipl. Phys. (Germany)
>
> Centre of Excellence for Invasion Biology
> Faculty of Science
> Natural Sciences Building
> Private Bag X1
> University of Stellenbosch
> Matieland 7602
> South Africa
>



-- 
Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation
Biology, UCT), Dipl. Phys. (Germany)

Centre of Excellence for Invasion Biology
Faculty of Science
Natural Sciences Building
Private Bag X1
University of Stellenbosch
Matieland 7602
South Africa