File opening error while running Rscript as cronjob

Question

I am trying to run an R script as a cronjob with a conda environment. Running the script from terminal works well with following script.sh

/opt/anaconda/anaconda3/envs/{env-name}/bin/Rscript '/ABSOLUTE_PATH/script.R'  >> '/ABSOLUTE_PATH/script.log' 2>&1

However, when defining the job in crontab -e as

30 14 * * * /ABSOLUTE_PATH/script.sh

Results in Error: package or namespace load failed for 'tidyverse' in dyn.load(file, DLLpath = DLLpath, ...): unable to load shared object '/opt/anaconda/anaconda3/envs/admenv/lib/R/library/stringi/libs/stringi.so': libicui18n.so.58: cannot open shared object file: No such file or directory Execution halted

I have tried to run it as

/opt/anaconda/anaconda3/envs/{env-name}/bin/R '/ABSOLUTE_PATH/script.R' >> '/ABSOLUTE_PATH/script.log' 2>&1

source /opt/anaconda/anaconda3/bin/activate {env-name}; Rscript '/ABSOLUTE_PATH/script.R' >> '/ABSOLUTE_PATH/script.log' 2>&1

which all resulted in the same error as a cronjob. Reinstalling packages in conda conda remove r-tidyverse; conda install -cr r-tidyverse conda remove r-tidyverse; conda install -cr r-tidyverse did not help.

Used R version is 3.6.2 and Anaconda 4.8.2. How could I solve this problem?

Answer 1

You are most likely using the environment wrongly. Conda cannot be directly called in a bash script, as per it wants to use conda init bash to be used in shell.

I managed to make this work with following steps:

# Create the enviroment to be used
conda create -n r-env r-base r-essentials <other packages, for example r-tidyverse r-dbi>

Next create the script.sh :

conda activate r-env
Rscript <ABSOLUTE_PATH>/script.R 

Then call the cronjob as an interactive (the -i option) shell which allows conda environment to be activated.

# You can use obliviously any time you like
* * * * * bash -i <ABSOLUTE_PATH>/script.sh <any error streams>

Tested with R script

library(tidyverse)
library(DBI)
library(ggplot2)
print("Hello world!")

which was piped to external file.

Answer 2

Thanks vahvero for the excellent reply. The problem seemed to be in my conda environment, so just recreating the environment with your instructions solved the problem!