Hello,
I am wondering if there is a way in Databricks to run a job continuously except for 1 or 2 hours every night in which the cluster could restart. We are using interactive clusters for our jobs and development in Dev and UAT. In Prod we are still using interactive clusters, just not developing with them or doing anything manually. We had wanted to use job clusters for the actual production jobs, but were concerned about start-up time for the job clusters given each job would have to wait for startup each time, which we cannot afford.
My concern with continuous jobs is the cluster will never restart, so I am assuming there will be performance issues over time? Or that the cache is never cleared and there could be more memory issues.
If this is not a concern, than I would just proceed with the continuous jobs. Otherwise, is there a way to stop continuous jobs for a 1-2 hours window and restart the cluster?
Thanks
