NVIDIA/cuda-repo-management

Question!

anaesco opened this issue · 2 comments

anaesco commented

Hi there, I am currently running a sequencing program called Master of Pores, and I am using the NVIDIA Docker/Cuda Container process to be able to run my pipeline using the GPU. However, I keep running to the error below and I was wondering if anyone could help me decipher what it is trying to say and/or if it is related to a failure in setting the development environment for CUDA... (i am a student in bioinformatics! so my R knowledge is still pretty novel) Thanks in advance!
Original command line:

Command executed:
nextflow run mop_preprocess.nf -with-docker > test.txt

export LD_LIBRARY_PATH="/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/.singularity.d/libs"

guppy_basecaller -x "cuda:0" --fast5_out --flowcell FLO-MIN106 --kit SQK-RNA002 -i ./ --save_path ./wt---1_out --gpu_runners_per_device 1 --cpu_threads_per_caller 1 --num_callers 1
cat wt---1_out/.fastq >> wt---1.fastq
rm wt---1_out/.fastq
gzip wt---1.fastq

kmittman commented

Hi @anaesco
This cuda-repo-management project is for RPM/Debian package repositories and related CDN issues.

Moving to nvidia-docker in case they can help.

elezar commented

Hi @anaesco. Please open an issue against either https://github.com/NVIDIA/libnvidia-container or https://github.com/NVIDIA/nvidia-container-toolkit and provide more information on your exact setup and the error message that you see.

@kmittman feel free to close this issue here.