본문 바로가기

전체 글44

[k8s] NodeSelector, PodNodeSelector, NodeLabel 🔹 NodeLabelNode에 적용되는 Key-Value 쌍명령어# 모든 Node들의 라벨 조회kubectl get nodes --show-labels# 특정 Node의 라벨 조회kubectl get node --show-labels# 특정 라벨=값을 가진 Node 리스트 조회kubectl get node -l # 특정 라벨을 가진 Node 리스트 조회kubectl get node -l =# 특정 라벨 컬럼과 함께 모든 Node 리스트 출력kubectl get node -L # 특정 Node에 라벨 추가kubectl label nodes =# 특정 Node에 라벨 제거kubectl label nodes -# 특정 Node에 라벨 변경kubectl label nodes = --overwrite예시.. 2025. 2. 14.
nvidia3 Dynamic Resource Allocation (DRA) for GPUs in Kubernetes Kevin Klues kklues@nvidia.com>Last Updated: February 21st, 2023 IntroductionDynamic Resource Allocation (DRA) is an upcoming Kubernetes feature that puts resource scheduling in the hands of 3rd-party developers. From an end-user’s perspective, It moves away from the limited "countable" interface for requesting access to resources  (e.g. "n.. 2025. 2. 14.
nvidia2 Using a config file to configure the k8s-device-plugin and gpu-feature-discoveryKevin Klues kklues@nvidia.com>Last Updated:03-June-2022Table of ContentsOverview 2Design Details 2Deploying and Testing 4Deployment via helm 4Specifying Multiple Configuration Files 5Updating Per-Node Configuration With a Node Label 6Enabling gpu-feature-discovery 6OverviewAt present, the only way to configure the k8.. 2025. 2. 14.
nvidia Custom Resource Naming and Supporting Multiple GPU SKUs on a Single Node in KubernetesKevin Klues kklues@nvidia.com>Last Updated:12-May-2022Table of ContentsOverview 2Assumptions, Constraints, Dependencies 2Design Details 3OverviewA common question that often gets asked is: Can I have multiple GPUs with different SKUs on the same node in my Kubernetes cluster? As of today, the short answer is no.. 2025. 2. 14.
[k8s] Prometheus Error scraping target 해결하기 (connect: connection refused) (controller-manager, etcd, proxy, scheduler) 🔹 구성 환경Master Node: 3대Worker Node: 10대kube-prometheus-stack 버전: 67.5.0🔹 사전 지식Static Pod란?kubelet이 직접 관리하는 Pod로 일반적인 Pod와 다르게 API 서버와 상관없이 동작한다. Kubernetes API 서버에 의해 스케줄링되지 않고, 각 노드의 Kubelet이 로컬 설정 파일(/etc/kubernetes/manifests/)을 기반으로 실행한다.kube-apiserver.yamlkube-controller-manager.yamlkube-scheduler.yamletcd.yaml /etc/kubernetes/manifests/ 하위 yaml을 수정하면 즉시 적용된다.  yaml을 수정하니 Pod가 terminated되고 .. 2025. 2. 13.
반응형

티스토리툴바