I'm not certain if autoscaling will ever make sense for Dask clusters, but it would certainly be useful for the more generic clusters associated with other Snakemake tasks (e.g. normalizing phenotype data or converting bgen to zarr).
I never found a way to get this to work (particularly with multiple node pools). Switching to Cloud Provider instead since autoscaling is simpler that way.
I'm not certain if autoscaling will ever make sense for Dask clusters, but it would certainly be useful for the more generic clusters associated with other Snakemake tasks (e.g. normalizing phenotype data or converting bgen to zarr).