Skip to content

Latest commit

 

History

History
62 lines (42 loc) · 2.56 KB

O2_portal_errors.md

File metadata and controls

62 lines (42 loc) · 2.56 KB

O2 Portal - R

These are common errors found when running R on the O2 portal and ways to fix them.

How to launch Rstudio

  • Besides your private R library, we have now platform shared R library, add this to Shared R Personal Library section : "/n/data1/cores/bcbio/R/library/4.3.1" (RNAseq and scRNAseq)
  • Minimum modules to load for R4.3.*: cmake/3.22.2 gcc/9.2.0 R/4.3.1
  • Minimum modules to load for 4.2.1 single cell analyses (some might be specific to trajectory analysis): gcc/9.2.0 imageMagick/7.1.0 geos/3.10.2 cmake/3.22.2 R/4.2.1 fftw/3.3.10 gdal/3.1.4 udunits/2.2.28
  • Sometimes specific nodes work better: under "Slurm Custom Arguments": -x compute-f-17-[09-25]

Issues

Issue 1 - You can make a session and open Rstudio on O2 but cannot actually type.

Potential solution: Make a new session and put the following under "Slurm Custom Arguments":

-x compute-f-17-[09-25]

Issue 2 - Everything was fine but then you lost connection.

When you attempt to reload you see:

Potential solutions: Refresh your interactive sessions page first then refresh your R page. If that doesn't work close your R session and re-open from the interactive sessions page. If that doesn't work wait 5-10 min then repeat.

Issue 3 - You made a session but cannot connect

When you attempt to connect you see:

Potential solutions: This error indicates that either you did not load a gcc module or you loaded the incorrect one for the version of R you are running. Kill the current session and start a new one with the correct gcc loaded in the modules to be loaded tab.

Issue 4 - When you finally refresh your environment is gone (THE WORST)

What happened is you ran out of memory and R restarted itself behind the scenes. You will NOT get an error message for this of any kind. The best thing to do is quit your session and restart a new one with more memory.

Issue 5 - Crashing

Also, previous issues with O2portal RStudio crashing - “the compute-f architecture is not good enough and this part of the process fails because (maybe) it was built/installed on a newer node” . Solution: add the flag when you start the session to just exclude those nodes -x compute-f-17-[09-25]

Issue 6 - commands using cores fail

Error in `.rowNamesDF<-`(x, value = value) : invalid 'row.names' length
In addition: Warning message:
In mclapply(X, function(...) { :
  scheduled cores 1, 2 did not deliver results, all values of the jobs will be affected