Entering edit mode
I encountered this odd problem while debugging an issue with derfinder (see the original thread here: derfinder: analyzeChr() breaks on chromosomes > 22. This code causes my machine to hang and never return the R prompt:
library(BiocParallel) xx <- bplapply(seq_len(1), function(x) sample(1:10, 1), BPPARAM = SnowParam())
However this code works as expected:
library(BiocParallel) xx <- bplapply(seq_len(1), function(x) sample(1:10, 1), BPPARAM = MulticoreParam())
as does this:
library(BiocParallel) xx <- bplapply(seq_len(1), function(x) sample(1:10, 1))
I've run that first snippet successfully on my Mac and on another Linux machine. The machine which won't run it is a 32 processor Fedora machine.
Thanks so much for any insights into what's going on!
> devtools::session_info() Session info------------------------------------------------------------------- setting value version R version 3.1.1 (2014-07-10) system x86_64, linux-gnu ui X11 language (EN) collate en_US.UTF-8 tz America/Chicago Packages----------------------------------------------------------------------- package * version date source base64enc 0.1.2 2014-06-26 CRAN (R 3.1.1) BatchJobs 1.5 2014-10-30 CRAN (R 3.1.1) BBmisc 1.8 2014-10-30 CRAN (R 3.1.1) BiocGenerics 0.12.0 2014-10-15 Bioconductor BiocParallel * 1.0.0 2014-11-11 Bioconductor brew 1.0.6 2011-04-13 CRAN (R 3.1.1) checkmate 1.5.0 2014-10-19 CRAN (R 3.1.1) codetools 0.2.9 2014-08-21 CRAN (R 3.1.1) DBI 0.3.1 2014-09-24 CRAN (R 3.1.1) devtools 1.6.1 2014-10-07 CRAN (R 3.1.1) digest 0.6.4 2013-12-03 CRAN (R 3.0.2) fail 1.2 2013-09-19 CRAN (R 3.1.1) foreach 1.4.2 2014-04-11 CRAN (R 3.1.1) iterators 1.0.7 2014-04-11 CRAN (R 3.1.1) RSQLite 1.0.0 2014-10-25 CRAN (R 3.1.1) rstudioapi 0.1 2014-03-27 CRAN (R 3.1.1) sendmailR 1.2.1 2014-09-21 CRAN (R 3.1.1) stringr 0.6.2 2012-12-06 CRAN (R 3.0.2)
Jessica
The cluster is being opened with
where
makeClusteris from the parallel package (apparently the package is not included with devtools' sessionInfo(), which is too bad) andx$.clusterargsisSnowParam()$.clusterags. So for me I haveand can create the cluster with
It sounds like this will fail for you, and then yes, it's something about available ports on your system. I believe you can create your SnowParam with the
portargument, for instance, as documented on?parallel::makeCluster.If you're running this is as a parallel job on a single Linux / mac computer, then it's almost always better to use the (default) MulticoreParam.
The parallel package is v3.1.1.
Correct, that line of code hangs on my machine. I'll talk with the admin and see if port access was changed recently in some way. I'll also ask the derfinder maintainer if he can use MulticoreParam instead of SnowParam.
Thank you!
Jessica
Regarding `devtools::session_info()` not printing the `base` package version, I just created an issue about this: https://github.com/hadley/devtools/issues/646
utils::sessionInfo() does not print base package versions either. Arguably it should.
Martin edited his earlier comment and wanted the version of the `parallel` package. However, see https://github.com/hadley/devtools/issues/646#issuecomment-62804289 Like I said in https://github.com/hadley/devtools/issues/646#issuecomment-62837880 I can't use `install.packages('parallel')` on R 3.1.2 or 3.2, so maybe this was an issue in the past but it's no longer one.
If you have a case where the versions of R and `parallel` could be un-linked, please share it. Otherwise I'll just close the github issue.
Thanks!
What is the output of just `SnowParam()`?