Kubernetes Events

GPUPool Events

• ManagedNodeCreated： Normal, Managed nodes created from cloud provider, to meet the capacity requirements
• Compaction: Normal, GPU nodes auto-compaction and node termination
• MaxResourceConstraintReached: Warning, GPU pool resources reach the preset maximum, unable to continue scaling

Pod Events

• GPUQuotaOrCapacityNotEnough: Warning, unable to continue scheduling because of not enough GPU resources or not enough GPU quota
• GPUDeviceAllocatedFailed: Warning, BindingCycle error after scheduling computation, unable to write GPU information to Pod's Annotation
• PreScheduleDone: Normal, SchedulingCycle completed successfully, resource reservation successful
• ScheduleWithNativeGPU: Normal, may occur when maintaining compatibility with existing Device Plugin, allows Pods not yet migrated to TensorFusion to run using physical GPU mode
• GPUDeviceAllocated: Normal, indicates that TensorFusion scheduler has completed the BindingCycle, Pod is about to start running with vGPU