opp-repl-ssh-cluster

To run distributed simulations across multiple machines over SSH, you can use Dask to distribute parameter sweep tasks across SSH-accessible worker nodes with a cluster scheduler.

Setting up a Dask SSH cluster for parallel simulation tasks requires pre-authenticated SSH between nodes, identical Python and opp_repl versions on all machines, and the dask and distributed cluster extras installed.

You should use a Dask SSH cluster instead of a single workstation when executing large parameter sweeps that cannot finish in a reasonable timeframe on one machine and you have SSH-accessible workers available.

To dispatch simulation binaries to Dask worker nodes over SSH, you can use rsync for binary distribution and set scheduler="cluster" to dispatch tasks across the configured worker hosts.

Yes, you can monitor distributed simulation progress on a Dask SSH cluster by accessing the Dask dashboard, which provides visibility into task execution across your worker nodes.

Distributed simulation tasks may fail across SSH nodes if Python and opp_repl versions are not identical across all machines, or if SSH access is not properly pre-authenticated between the cluster nodes.