Closed tobigithub closed 8 years ago
I recently closed a similar issue. After re-install, I thought all was good - turns out I was wrong. I have the same issue in that I can get the optimisation to work once only, followed by "Error in makeCluster(nSlaves, type = "PSOCK") : unknown cluster type".
OS: Windows 7 R version 3.2.2 Latest IPO builds + latest depencies.
Hi,
Thanks Tobias and M-R-JONES. I think I now understand better where the problem comes from:
The snow clusters are started by xcmsSet when calculating the xcmsSet-object with the best parameter settings found within the last DoE. But IPO uses "PSOCK"-clusters.
There are two similar but different makeCluster-functions. One from the package 'snow' and one from 'parallel'. Using snow::makeCluster(1, type="PSOCK") results in an error.
Therefore i preceded the functions 'makeCluster', 'clusterExport', 'parSapply' and 'stopCluster' with 'parallel::' and uploaded a new version.
Could you try it and tell me if the changes helped?
Cheers, Gunnar
Hi, I used the lipidomics benchmark from the WIKI "myIPOSet2 <- function ()" and got the same output,
starting new DoE with:
min_peakwidth: c(12, 28)
max_peakwidth: c(35, 65)
ppm: c(17, 32)
mzdiff: c(-0.001, 0.01)
snthresh: 10
noise: 0
prefilter: 3
value_of_prefilter: 100
mzCenterFun: wMean
integrate: 1
fitgauss: FALSE
verbose.columns: FALSE
nSlaves: 1
Error in mpi.spawn.Rslaves(nslaves = nSlaves, needlog = FALSE) :
Spawning is not implemented. Please use mpiexec with Rprofile.
In addition: Warning messages:
1: package ‘Rcpp’ was built under R version 3.1.3
2: In fun(libname, pkgname) :
mzR has been built against a different Rcpp version (0.11.3)
than is installed on your system (0.12.0). This might lead to errors
when loading mzR. If you encounter such issues, please send a report,
including the output of sessionInfo() to the Bioc mailing list at
http://www.bioconductor.org/help/mailing-list. For details see also
https://github.com/sneumann/mzR/wiki/mzR-Rcpp-compiler-linker-issue.
3: package ‘rsm’ was built under R version 3.1.3
4: package ‘igraph’ was built under R version 3.1.3
5: replacing previous import by ‘igraph::groups’ when loading ‘CAMERA’
Timing stopped at: 3.85 0.36 1098.55
>
>
> ### [1] "Finished. Thank you for using IPO."
> ### user system elapsed
> ### 4577.88 2253.56 16310.44
> ### 1h:16 min 37 min 4h:32 min
>
> ### ***********************************************************************
> ### function finished
> ### ***********************************************************************
I guess its just that Rmpi function. I also get GUI window pop-ups if mpich is not installed, but that is an "unrelated" xcms issue under windows. Cheers Tobias
Looks like the parallelization problem within IPO is fixed now, therefore I'm closing this issue.
Thanks for your help
Hi, thanks for the fix, still need to redo the examples. Tobias
Hi, under WINDOWS7 I uninstalled Rmpi and only use snow now. IPO fires up, but after the first optimization round it breaks. BTW this was a tested and working version. Central Carbon Metabolism was the script I used. SO first round is OK and this is the second DoE round, I guess everything is fine in the first round, then in the second round something happens.
Why would IPO built use makeCluster(nSlaves, type = "PSOCK") in the second round, but not in the first round?
T