v1.5.0
✨ New Features & Enhancements
-
Pre and Post-Delete Validation Checks (#1590): Introduces the ability to define health checks that must pass before and after a resource is deleted. This ensures that addons are only removed when the cluster is in a safe state and verifies that the removal was successful and didn't leave the system in an unstable condition.
-
Dynamic Patches via patchesFrom (#1585): Profiles and ClusterProfiles can now reference ConfigMaps or Secrets containing StrategicMerge or JSON patches.
- Templating Support: The name and namespace of the referenced resource can be defined using templates (e.g., {{ .Cluster.metadata.name }}-patches), allowing for cluster-specific patches managed centrally.
- Flexible Formats: Supports both standard Kubernetes strategic merge patches and JSON patches (RFC 6902).
⚡ Optimizations
-
Agent Efficiency: Optimized memory utilization across Sveltos agents, reducing the overhead for both sveltos-agent and drift-detection-manager
-
Log Refinement: Reduced log noise in info mode. High-volume, non-essential logs have been suppressed to provide a clearer view of system health and deployment progress.
🐞 Bug Fixes
-
Fixed Accidental Helm Uninstalls on Restart (#1608): Resolved a critical issue where templated Helm release names or namespaces caused the Chart Manager to incorrectly identify releases as "stale" during controller restarts. This fix ensures that templated values are properly instantiated before being compared against existing releases, preventing accidental uninstalls.
-
Race Condition Protection in Request Queue (#1600): Improved reconciliation reliability by discarding results if a newer request for the same key is already in the dirty queue. This ensures the controller always acts on the most recent cluster state and prevents outdated results from overriding newer pending changes.