Running on clusters
There is a file in the templates
called cluster.yaml
. This can be used to modify ressources needed for your data. I generally recommand moving the file to the root of the folder so that it doesn't get replaced by updates.
Bellow is an example of running on a cluster using the template file cluster.yaml
on SLURM.
snakemake --cluster 'sbatch -n {cluster.n} -t {cluster.time} --clusters=CLUSTERNAME --output={cluster.output}' --jobs N --cluster-config cluster.yaml --use-conda --local-cores C
- N: is the number of jobs you are allowed to run at the same time
- C: is the local-cores of the host machine. A few simple rules are gonna be run locally (not sent to nodes) because they are not that heavy (mostly plotting)
- CLUSTERNAME: the name of the cluster you want to use
Note: The default path for cluster logs in the cluster.yaml is logs/cluster/
. If that folder doesn't exist, our cluster can't write and will crash without an error message.