a scheduled job would restart its cluster if it is terminated.
AdminPM (Product Manager, Databricks) commented
It is supported from version 2.71 (May 17th, 2018). See more here: https://docs.databricks.com/release-notes/product/2018/may.html#cluster-autostart
if a dashboard is updated when its cluster is terminated the update fails.
my present workaround is to write a script using the api to restart the desired cluster just prior to when the dashboard update is scheduled.