jade-hpc-gpu / jade-hpc-gpu.github.io

Joint Academic Data Science Endeavour (JADE) is the largest GPU facility in the UK supporting world-leading research in machine learning (and this is the repo that powers its website)
http://www.jade.ac.uk/
Other
24 stars 7 forks source link

Apptainer is not available on the login nodes. #193

Open pinakinathc opened 2 months ago

pinakinathc commented 2 months ago

I think JADE moved from Singularity to Apptainer containers. On login, I execute the following command: module avail and get the following output:

-------------------------------------------------------------------------------------------------- /jmain02/apps/lmod/lmod/modulefiles/Core --------------------------------------------------------------------------------------------------
   StdEnv    lmod    settarg    use.dev    use.jade    use.old    use.own    use.std    use.viz

------------------------------------------------------------------------------------------------- /jmain02/apps/Modules/modulefiles/packages -------------------------------------------------------------------------------------------------
   alphafold/2.2.0        apptainer/1.3.0       firefox/123.0.1                gromacs/2020.5-ramd-2.0    gromacs/2021.5        mdtraj/1.9.5           opencv/4.2.0    pytorch/1.9.0         tensorflow/2.3.1
   amber/18               autodock-gpu/1.5.2    gromacs/2020.3                 gromacs/2021.1             gromacs/2022          namd/2.14              openmm/7.5.0    pytorch/1.12.1 (D)    tensorflow/2.7.0
   amber/20        (D)    ffmpeg/4.2.2          gromacs/2020.4-plumed-2.6.2    gromacs/2021.2             gromacs/2022.2 (D)    namd/3.0-alpha7 (D)    plumed/2.6.2    sox/14.4.2            tensorflow/2.9.1 (D)

-------------------------------------------------------------------------------------------------- /jmain02/apps/Modules/modulefiles/other ---------------------------------------------------------------------------------------------------
   cuda/10.1    cuda/11.1-gcc-9.1.0    cuda/12.4  (D)    gcc/9.1.0        google/cloud-sdk    nano/5.3        openmpi/4.1.5-gcc9    python/jupyter     python/3.8.6 (D)    utilities
   cuda/10.2    cuda/11.2              emacs/27.1        gnuplot/5.2.8    hdf5/1.10.7         netcdf/4.7.4    python/anaconda3      python/notebook    tmux/3.2a           vim/8.2

Next, I try to import module apptainer and get the error: Apptainer is not available on the login nodes, please use Slurm to submit to the compute nodes.

Can anyone please point me to some doc of how to use Apptainer/Singularity in JADE2?

sigmoid-bar commented 1 month ago

I was able to work with this by first requesting a interactive session on a compute node. srun -p devel --gres=gpu:2 --pty /bin/bash. Once I had a session I was able to run module load apptainer which worked. I did not test with running a container though.