spack stack for encfs
Stack must be built on a compute node (mandatory when rocm or cuda is used)
salloc -t 400 -p cpu
srun --pty bash
cd /dev/shm
git clone https://github.com/eth-cscs/stackinator.git
cd stackinator
./bootstraph.sh
bin/stack-config --develop -b /dev/shm//build-bwrap -s ~/stacks/alps-cluster-config/hohgant/ -r ~/encfs-stackinator -c /scratch/e1000/$(whoami)/cache-config.yaml
Then run the commands printed by stack-config.
Note:
alps-cluster-config
: https://github.com/eth-cscs/alps-cluster-configbin/stack-config --develop
:--develop
is only necessary when usingcommit: develop
inconfig.yaml
Wipe /dev/shm
before releasing the slurm allocation