[R-sig-hpc] Rmpi: mpi.close.Rslaves() 'hangs'

Marius Hofert marius.hofert at uwaterloo.ca
Tue Sep 26 11:56:56 CEST 2017


Hi,

The example on ?mpi.close.Rslaves seems to 'hang' for me:

library(Rmpi)
mpi.spawn.Rslaves(nslaves=2)
tailslave.log()
mpi.remote.exec(rnorm(10))
mpi.close.Rslaves() # => hangs

... with output (hanging at the last line, not returning to the prompt):

> library(Rmpi)
> mpi.spawn.Rslaves(nslaves=2)
tailslave.log()
2 slaves are spawned successfully. 0 failed.
master (rank 0, comm 1) of size 3 is running on: sklar
slave1 (rank 1, comm 1) of size 3 is running on: sklar
slave2 (rank 2, comm 1) of size 3 is running on: sklar
>
==> sklar.51249+1.51251.log <==
Host: sklar Rank(ID): 1 of Size: 3 on comm 1

==> sklar.51249+1.51252.log <==
Host: sklar Rank(ID): 2 of Size: 3 on comm 1
> mpi.remote.exec(rnorm(10))
           X1         X2
1   0.7266248 -1.3922235
2   0.7212728  1.6761796
3   0.4735390 -1.8894523
4   1.3799295 -1.0452228
5  -2.1399312 -0.3452672
6   0.0909944  0.3565657
7  -0.6401649  0.8735195
8  -0.4515465 -0.8951751
9  -0.2142557  0.1436790
10 -1.3349072 -0.7224436
> mpi.close.Rslaves()

Below is my sessionInfo(). OpenMPI was installed correctly:

mhofert at sklar:~$ mpirun --version
mpirun (Open MPI) 2.1.1

Report bugs to http://www.open-mpi.org/community/help/
mhofert at sklar:~$

More information:
1) The above behavior is also mentioned in the last comment on
https://github.com/Burke-Lauenroth-Lab/rSFSW2/issues/43
2) We use(d) similar code in 'simsalapar' some time/years ago and it
never failed. I haven't used the package myself in a while, so I just
noticed (while trying to check the package, it 'hung'/never
terminated). mpi.close.Rslaves() didn't seem to change in that time
either. The only thing that changed is my hardware (no MacBook Pro
"13, 2017; from MacBook Pro "15, 2015). I suspect it's related to that
(?) but not sure.
3) Since mpi.close.Rslaves() 'hangs', I see all workers still active
(htop output).
4) I reported this to both R-SIG-Mac and maintainer("Rmpi") but didn't
get a reply.

Thanks & cheers,
Marius


> sessionInfo()
R version 3.4.1 (2017-06-30)
Platform: x86_64-apple-darwin16.7.0 (64-bit)
Running under: macOS Sierra 10.12.6

Matrix products: default
BLAS: /usr/local/R/R-3.4.1_build/lib/libRblas.dylib
LAPACK: /usr/local/R/R-3.4.1_build/lib/libRlapack.dylib

locale:
[1] en_CA.UTF-8/en_CA.UTF-8/en_CA.UTF-8/C/en_CA.UTF-8/en_CA.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

other attached packages:
[1] Rmpi_0.6-6

loaded via a namespace (and not attached):
[1] compiler_3.4.1 parallel_3.4.1



More information about the R-sig-hpc mailing list