Explanation

The Cluster Event Log (also known as the compute event log) in Databricks captures all lifecycle events for a cluster, including starting, stopping, and autoscaling actions (resizing up or down). Each event is timestamped, allowing administrators to correlate cluster activity with workload demands. These logs can be viewed via the Event Log tab on the cluster's detail page or retrieved via the Clusters API.

Why the other options are incorrect:

• Workspace audit logs: These track administrative and security actions across the workspace (e.g., who created a job or changed a permission) but do not record automated infrastructure scaling events.
• The Driver's log files: These contain application-level logs (Spark output and debugging info) related to the execution of code, not the cluster's infrastructure lifecycle.
• Ganglia: This provides real-time and historical resource usage metrics (CPU, Memory, Network), but it does not provide a discrete log of resize events.