Kostavo scans both Databricks Jobs and Declarative Pipelines for runaway resources, silent failures, cluster misconfigurations, and security risks, without manual reviews.
Continuous-mode pipelines and jobs run 24/7 without review. Pipelines left in development mode burn compute indefinitely.
Repeatedly failing jobs and pipelines stuck in FAILED state notify no one. Teams discover data issues long after the fact.
Jobs running on all-purpose clusters, secrets hardcoded in env vars, SSH keys left active, missing cost tags: all invisible without systematic policies.
Flags continuous-mode pipelines and continuously-running jobs. Also detects long-running stuck job runs. Can stop them automatically.
Detects repeatedly failing jobs, pipelines stuck in FAILED state, and resources with no failure notifications or retry policies configured.
Scans job clusters for all-purpose cluster use, SSH access, plaintext secrets in env vars, missing cost tags, and overly permissive job permissions.
Job and pipeline findings: failing runs, missing timeouts, retry policy, and continuous mode.
Continuous mode, failed state, development mode, edition, preview channel, target schema, notifications, asset bundle
Failing runs, long-running runs, continuous jobs, concurrency, cluster configs, permissions, notifications, retry policies, asset bundle
Add your Databricks workspace to Kostavo. Kostavo becomes a workspace admin and uses a minimal service principal (Azure or AWS), read-only for most policies.
Choose which policies to enable and set action modes: Notify to alert resource owners and workspace admins, or Fix/Remove to stop runaway resources automatically.
Kostavo scans your workspace on a schedule. Findings appear in the dashboard and trigger notifications to the resource owner and workspace admins.
See which jobs and pipelines are failing, retrying, or running without limits.