May 2026 - V1.19.0

✨✨✨ This release introduces scheduled NodePool scaling, improves the component upgrade workflow, and adds better disk visibility across CloudPilot AI. It also includes Workload Autoscaler safety improvements for JVM and OOM scenarios, plus reliability fixes that make daily operations smoother.

🚀 Highlights

NodePool Schedule Scaling

CloudPilot AI now supports NodePool schedule scaling, allowing teams to align capacity with predictable workload patterns. Instead of keeping the same NodePool size around the clock, users can define schedule-based scaling behavior so capacity is available when demand is expected and reduced when it is no longer needed.

The schedule experience is unified across CloudPilot AI surfaces, so users can manage schedule-based scaling from the same product workflow instead of maintaining separate per-environment procedures.

NodePool schedule scaling

Automatic Component Upgrade Workflow

CloudPilot AI now supports an automatic component upgrade workflow for in-cluster components. This gives operators a clearer and more consistent path for keeping CloudPilot AI components aligned with the selected product version.

Because V1.19.0 changes upgrade behavior and component packaging, this version is marked as manual-upgrade-required so operators can review the upgrade before applying it.

Automatic component upgrade workflow

⚙️ Enhancements

Add disk support and disk monitoring, improving visibility into node and workload resource usage beyond CPU and memory.
Improve Workload Autoscaler runtime safety with OOM auto-remediation, ResourceStartupBoost improvements, enhanced JVM optimization, and safer Pod recreation when resources need to be adjusted.
Improve Data Protection Mode and proxy handling, including safer behavior when workload uploading is disabled.
Add workload resync support so workload state can be refreshed more reliably.
Improve Workload Autoscaler recommendation metrics so recommended CPU data is easier to query and display consistently.
Improve price refresh reliability so partial upstream price data does not overwrite healthy local price data.

🛠️ Bug Fixes

Fix saving report edge cases, including discount handling and date-range filtering for usage data.
Fix false record-not-found behavior in idempotent marketplace updates.
Fix controller reliability issues by reducing Kubernetes API pressure, preventing controller stalls, and reconnecting dead proxy connections.
Fix Workload Autoscaler usage calculation and JVM PromQL query behavior.

These updates further improve CloudPilot AI’s scheduling flexibility, autoscaling safety, and operational reliability. For questions or support, join our Slack community

Stay tuned for more updates! 🚀