Databricks issue: isolate workspace, compute, job, and data layer

Record cloud, region, workspace, runtime, cluster or SQL warehouse, timestamp, job or notebook, and exact failing operation. Compare service status, identity, permissions, libraries, compute health, storage, network, quotas, and recent changes. Preserve logs and job outputs before rerunning writes. Never publish tokens, workspace URLs, notebooks with proprietary code, cloud credentials, table data, secrets, cluster configs, or customer information.