[{"data":1,"prerenderedAt":17},["ShallowReactive",2],{"blog-tag-ja-oomkilled":3},[4],{"path":5,"title":6,"description":7,"date":8,"tags":9},"\u002Fblog\u002Fja\u002Fk3s-gpu-oomkilled-vpa-resource-estimation","GPUメモリ不足でAIトレーニングジョブがOOMKilled。K3sでVPAが効かない理由とリソース設計","K3s上でAIトレーニングジョブがGPUメモリ不足でOOMKilledされる原因と、Vertical Pod Autoscalerが GPU リソースには効かない理由を解説。requests\u002Flimitsの正しい見積もり方法と実測ベースの設計手順まで、運用者向けに具体的に紹介します。","2026-09-20",[10,11,12,13,14,15,16],"k3s","kubernetes","gpu","oomkilled","vpa","ai-infrastructure","resource-management",1792592672728]