Hi @SOlivero ,

Try configuring a shared all-purpose cluster and set each job to use this existing cluster rather than creating new job-specific clusters, ensuring the cluster stays warm and avoiding startup delays. Another option is to restructure your master job to run the notebooks sequentially using task dependencies within a single job definition, allowing all tasks to share the same cluster. This setup should help improve speed and efficiency, making your jobs run faster without the need for multiple clusters.

Hope this helps!

Regards,

Brahma