Repository Issue Activity (beta)

#3073 [Proposal] HAMi-aware fractional GPU cost attribution with OpenCost

#1878 [Feature] Pod Preemption Support for HAMi Scheduler

#3076 device plugin should serve only the pod the node lock names

#3079 Scheduler assigns GPU allocation to init container index, breaking vGPU monitor metrics

#3083 Add allocation outcome and rollback metrics

#3016 enflame: resource name flags use the same variable

#3078 vendor device plugins do not validate the allocation annotation

#3081 [Feature] Support all four NeuronCores on Inferentia1 devices

#2581 Feature Request: google gVisor (runsc) compatibility for NVIDIA vGPU enforcement

#3055 Use `values.yaml` as unified `device-config` value source and standardize values structure

#3077 allocation check skips MIG

#2764 HAMi v2.11 RoadMap

#1899 Scheduler health check incorrectly marks GPU nodes as unhealthy after 60s, removing them from cache

#2126 [LFX Mentorship] Improve HAMi GPU observability metrics and dashboards

#3066 CNCF AI Technical Community Group Discussion/Presentation

#3057 [Ascend] Defer memory trimming for mode-agnostic Pods until node filtering

#3065 test: add a make bench target and run the existing Go benchmarks in CI

#3068 [Feature] nodeconfig: match nodes by label selector (nodelabelselector), not only by exact name

#2833 Enhance Dynamic MIG: Add CDI support

#3062 honor preferred nodeAffinity in HAMi's internal node score

#3059 bug: Cambricon setNodeLock and releaseNodeLock silently drop the original API error

#3058 perf(metrics): index device UUIDs once per scrape instead of scanning all nodes per container device

#2873 Support all 4 NeuronCores on AWS Inferentia (Inf1) devices

#3051 Hygon HCU rename leaves dcu-vgpu-device-plugin advertising the old resource names

#3052 fix(scheduler): count GPU app containers correctly with init containers