Karmada
v1.16.3Orchestration & ManagementKarmada v1.16.3 is a bug-fix release addressing critical issues including controller race conditions, scheduler panics, and graceful eviction problems that could impact multi-cluster operations stability.
breakingUpgrade immediately to prevent scheduler crashes
The scheduler panic fix for divide-by-zero errors is critical for production stability. Schedule maintenance window to upgrade karmada-scheduler component, especially if you use spread constraints across clusters with dynamic availability.
enhancementReview CronFederatedHPA configurations after upgrade
The fix for CronFederatedHPA scale-up failures when replicas field is missing improves reliability. After upgrading, verify your CronFederatedHPA resources are scaling correctly from zero replicas, and consider adding explicit replicas fields to avoid edge cases.
enhancementMonitor graceful eviction behaviors post-upgrade
The graceful eviction timing fix prevents GracePeriodSeconds leakage between tasks. After upgrade, monitor your workload evictions to ensure they complete within expected timeframes and adjust grace periods if needed for improved resource cleanup.
Key changes (5)
- Fixed job status aggregator race condition causing error loops in controller manager
- Resolved CronFederatedHPA scaling failures when replicas field is missing
- Corrected graceful eviction task timing issues preventing proper resource cleanup
- Fixed scheduler divide-by-zero panic when no valid clusters available for spread constraints
- Improved backoff queue sorting to prevent priority inversion in scheduler operations