Reuse#

You can name clusters so that you can share them between Python sessions using the name= keyword. For example:

import coiled

cluster = coiled.Cluster(
    n_workers=100,
    name="production",
    shutdown_on_close=False,
)

Then any Python session connected to this account can connect to this same cluster using this same command.

This is useful in a few situations:

  • In interactive Python sessions (like Jupyter notebooks) where users frequently restart a notebook, but want rapid access to the same cluster

  • In production environments, where a web service (or possibly many replicas of the same service) offload work to the cluster

The first time someone calls coiled.Cluster(name="production") it will create a new cluster. Subsequent calls will connect to the same cluster.