WorkDir empty for execution of .command.sh #4804
Description
Bug report
Hello,
I have been using nextflow 23.10.0 for a bioinformatics analysis using HPC. It happens that for example, I have a batch of 33 samples, and all of them are processed correctly until certain step of my pipeline, after a while the pipeline ends up processing only 28/33. When I look into the nextflow.log file, I tried to look into those samples that weren't processed (details like workdir, job id in for the HPC, etc). When I look into the workdir it appears to be empty only containing the .command.run and .command.sh, but not the files that are supposed to be in there for the execution of the .command.sh and therefore the jobs submition to the HPC queue. I have experienced this problem in two versions of nextflow, 23.10.0 and 23.04.3.
Expected behavior and actual behavior
What I would like to see is all the samples getting to the end through the pipeline. In the past, using older versions of nextflow never experienced this problem before (even with larger number of samples)
Please let me know if there are more details that can help me to understand how to solve this issue
Thanks
Alex