[R-sig-hpc] snow: makeMPIcluster hanging
Florian Umlauf (CRIE)
umlauf at uni-bremen.de
Mon Aug 12 13:28:41 CEST 2013
Hi @all,
I have some trouble while using mpi.spawn.Rslaves from the snow package.
I have successfully installed mpich2 and its working fine with the Rmpi
package:
> library(Rmpi)
> mpi.spawn.Rslaves()
4 slaves are spawned successfully. 0 failed.
master (rank 0, comm 1) of size 5 is running on: nb-wimiforn2
slave1 (rank 1, comm 1) of size 5 is running on: nb-wimiforn2
slave2 (rank 2, comm 1) of size 5 is running on: nb-wimiforn2
slave3 (rank 3, comm 1) of size 5 is running on: nb-wimiforn2
slave4 (rank 4, comm 1) of size 5 is running on: nb-wimiforn2
> mpi.remote.exec(rnorm(10))
X1 X2 X3 X4
1 -0.826003439 1.1755321 -1.12667720 -0.7356504
2 -0.446206608 -1.0365283 0.23759725 -0.5533241
3 -0.675614170 -2.8628252 0.61035157 -0.5376180
4 1.471801215 -1.2432659 -0.09245669 -0.4177295
5 0.511399298 0.1925932 -0.21761067 0.8573787
6 0.499592894 -0.1596297 0.04846257 -0.2703588
7 -2.202159413 0.9161975 -1.39892893 0.3161000
8 -0.880671433 -0.0576492 0.84399188 0.4670704
9 -0.466700498 -0.2201990 -1.49148264 0.7604177
10 0.007915527 0.3047164 -0.54815462 0.4506002
> mpi.close.Rslaves()
[1] 1
But every time I run the makeMPIcluster procedure from the snow-package
on just one single machine it hangs up (right at the moment when
mpi.comm.spawn is loaded). I tried the code from the makeMPIcluster
procedure step by step. R freezes at this very moment and one cpu takes
100%. When I tried the same for mpi.spawn.Rslaves from Rmpi package
mpi.comm.spawn works fine. I tried to switching of the firewall, I tried
to run R in admin-mode and I give accessrights to everyone for my
R-Folder. Snow and Rmpi are both installed in the same directory. I also
looked for the parameters:
Mpitask : "C:/PROGRA~1/R/R-30~1.1/bin/Rscript.exe"
Args: c("\"C:/Program Files/R/R-3.0.1/library/snow/RMPInode.R\"",
"SNOWLIB=C:/Program Files/R/R-3.0.1/library", "OUT=/dev/null")
Count: 2
Intercomm: 2
> version _
platform x86_64-w64-mingw32
arch x86_64
os mingw32
system x86_64, mingw32
status
major 3
minor 0.1
year 2013
month 05
day 16
svn rev 62743
language R
version.string R version 3.0.1 (2013-05-16)
nickname Good Sport
Do you have any suggestions what’s going wrong or what to do?
Thanks a lot everyone
Best Florian
--
Florian Umlauf
Volkswirt, M.Sc.
Universität Bremen
Centre for Regional and Innovation Economics
Wilhelm-Herbst-Str. 12
28359 Bremen
Telefon: +(49) 421 218 66537
Telefax: +(49) 421 218 66832
Email: fumlauf at uni-bremen.de
More information about the R-sig-hpc
mailing list