Snakemake version 8.27.1 on LSF cluster

Question

[John]

I'm trying to run a Snakemake workflow in a new lab - the Snakefile already exists. For context we are using LSF submission system.

If I run snakemake <options> at the command line, it all runs locally, despite the bsub arguments being provided in the Snakefile.

This is obviously an issue when using Kraken2 (or similar) since the databases all seem to get loaded locally and then cause RAM issues.

I do not want to use memory-map.

What is the proper way to do this in 8.27? The documentation online is very unclear and some of the "official" documentation doesn't even work (e.g., --executor lsf isn't available, only --executor <local,dryrun,touch>)

[Ram]

There is --executor lsf in the docs, only --executor cluster-generic. I'm looking at: https://snakemake.readthedocs.io/en/stable/getting_started/migration.html#id1

If you're looking at https://github.com/BEFH/snakemake-executor-plugin-lsf, you may have to add the plug-in (I've never done that, I don't know what it even means).

Answer 1

[DBScan]

First, you should check if the lsf plugin is installed. I think by default only the SLURM plugin is installed (at least in my case). You can find the instructions on how to install the lsf plugin here https://snakemake.github.io/snakemake-plugin-catalog/plugins/executor/lsf.html