Autostop and Autodown

The autostop (or autodown) feature automatically stops (or tears down) a cluster after it becomes idle.

With autostop, users can simply submit jobs and leave their laptops, while ensuring no unnecessary spending occurs. After jobs have finished, the clusters used will be automatically stopped, which can be restarted later.

With autodown, the clusters used will be automatically torn down, i.e., terminated.

Note

The autostop/autodown logic is executed by the remote cluster. Your local machine does not need to stay up for them to take effect.

Setting autostop

To schedule autostop for a cluster, set autostop in the SkyPilot YAML:

resources:
  autostop: true  # Stop after default idle minutes (5).

  # Or:
  autostop: 10  # Stop after this many idle minutes.

Alternatively, use sky autostop or sky launch -i <idle minutes>:

# Launch a cluster with logging detached (the -d flag)
sky launch -d -c mycluster cluster.yaml

# Autostop the cluster after 10 minutes of idleness
sky autostop mycluster -i 10

# Use the default, 5 minutes of idleness
# sky autostop mycluster

# (Equivalent to the above) Use the -i flag:
sky launch -d -c mycluster cluster.yaml -i 10

Setting autodown

To schedule autodown for a cluster, set autodown in the SkyPilot YAML:

resources:
  autostop:
    idle_minutes: 10
    down: true  # Use autodown.

Alternatively, pass the --down flag to either sky autostop or sky launch:

# Add the --down flag to schedule autodown instead of autostop.

# This means the cluster will be torn down after 10 minutes of idleness.
sky launch -d -c mycluster2 cluster.yaml -i 10 --down

# Or:
sky autostop mycluster2 -i 10 --down

Canceling autostop/autodown

To cancel any scheduled autostop/autodown on the cluster:

sky autostop mycluster --cancel

Viewing autostop status

To view the status of the cluster, use sky dashboard or sky status:

$ sky status
NAME         INFRA           RESOURCES                     STATUS   AUTOSTOP       LAUNCHED
mycluster    AWS (us-east-1) 2x(cpus=8, m4.2xlarge, ...)   UP       10 min         1 min ago
mycluster2   AWS (us-east-1) 2x(cpus=8, m4.2xlarge, ...)   UP       10 min(down)   1 min ago

Cluster that are autostopped/autodowned are automatically removed from the status table.