如何在kubernetes環境中共享GPU

夜洛IT后端技术分享發表於2024-10-13
GPU

隨著人工智慧和大模型的快速發展，雲上GPU資源共享變得必要，因為它可以降低硬體成本，提升資源利用效率，並滿足模型訓練和推理對大規模平行計算的需求。

在kubernetes內建的資源排程功能中，GPU排程只能根據“核數”進行排程，但是深度學習等演算法程式執行過程中，資源佔用比較高的是視訊記憶體，這樣就形成了很多的資源浪費。

目前的GPU資源共享方案有兩種。一種是將一個真正的GPU分解為多個虛擬GPU，即vGPU，這樣就可以基於vGPU的數量進行排程；另一種是根據GPU的視訊記憶體進行排程。

本文將講述如何安裝kubernetes元件實現根據GPU視訊記憶體排程資源。

系統資訊

  • 系統：centos stream8

  • 核心：4.18.0-490.el8.x86_64

  • 驅動：NVIDIA-Linux-x86_64-470.182.03

  • docker：20.10.24

  • kubernetes版本：1.24.0

1. 驅動安裝

請登入nvida官網自行安裝：https://www.nvidia.com/Download/index.aspx?lang=en-us

2. docker安裝

請自行安裝docker或其他容器執行時，如果使用其他容器執行時，第三步配置請參考NVIDA官網 https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#installation-guide

注意：官方支援docker、containerd、podman，但本文件只驗證過docker的使用，如果使用其他容器執行時，請注意差異性。

3. NVIDIA Container Toolkit 安裝

  1. 設定倉庫與GPG Key
distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
   && curl -s -L https://nvidia.github.io/libnvidia-container/$distribution/libnvidia-container.repo | sudo tee /etc/yum.repos.d/nvidia-container-toolkit.repo
  1. 開始安裝
sudo dnf clean expire-cache --refresh
sudo dnf install -y nvidia-container-toolkit
  1. 修改docker配置檔案新增容器執行時實現
sudo nvidia-ctk runtime configure --runtime=docker
  1. 修改/etc/docker/daemon.json,設定nvidia為預設容器執行時（必需）
{
    "default-runtime": "nvidia",
    "runtimes": {
        "nvidia": {
            "path": "/usr/bin/nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}
  1. 重啟docker並開始驗證是否生效
sudo systemctl restart docker
sudo docker run --rm --runtime=nvidia --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi

如果返回如下資料，說明配置成功

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 450.51.06    Driver Version: 450.51.06    CUDA Version: 11.0     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  Tesla T4            On   | 00000000:00:1E.0 Off |                    0 |
| N/A   34C    P8     9W /  70W |      0MiB / 15109MiB |      0%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
​
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

4. 安裝K8S GPU排程器

  1. 首先執行以下yaml，部署排程器
# rbac.yaml
---
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: gpushare-schd-extender
rules:
  - apiGroups:
      - ""
    resources:
      - nodes
    verbs:
      - get
      - list
      - watch
  - apiGroups:
      - ""
    resources:
      - events
    verbs:
      - create
      - patch
  - apiGroups:
      - ""
    resources:
      - pods
    verbs:
      - update
      - patch
      - get
      - list
      - watch
  - apiGroups:
      - ""
    resources:
      - bindings
      - pods/binding
    verbs:
      - create
  - apiGroups:
      - ""
    resources:
      - configmaps
    verbs:
      - get
      - list
      - watch
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: gpushare-schd-extender
  namespace: kube-system
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: gpushare-schd-extender
  namespace: kube-system
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: gpushare-schd-extender
subjects:
  - kind: ServiceAccount
    name: gpushare-schd-extender
    namespace: kube-system
​
# deployment yaml
---
kind: Deployment
apiVersion: apps/v1
metadata:
  name: gpushare-schd-extender
  namespace: kube-system
spec:
  replicas: 1
  strategy:
    type: Recreate
  selector:
    matchLabels:
      app: gpushare
      component: gpushare-schd-extender
  template:
    metadata:
      labels:
        app: gpushare
        component: gpushare-schd-extender
      annotations:
        scheduler.alpha.kubernetes.io/critical-pod: ''
    spec:
      hostNetwork: true
      tolerations:
        - effect: NoSchedule
          operator: Exists
          key: node-role.kubernetes.io/master
        - effect: NoSchedule
          key: node-role.kubernetes.io/control-plane
          operator: Exists
        - effect: NoSchedule
          operator: Exists
          key: node.cloudprovider.kubernetes.io/uninitialized
      nodeSelector:
        node-role.kubernetes.io/control-plane: ""
      serviceAccount: gpushare-schd-extender
      containers:
        - name: gpushare-schd-extender
          image: registry.cn-hangzhou.aliyuncs.com/acs/k8s-gpushare-schd-extender:1.11-d170d8a
          env:
            - name: LOG_LEVEL
              value: debug
            - name: PORT
              value: "12345"
​
# service.yaml
---
apiVersion: v1
kind: Service
metadata:
  name: gpushare-schd-extender
  namespace: kube-system
  labels:
    app: gpushare
    component: gpushare-schd-extender
spec:
  type: NodePort
  ports:
    - port: 12345
      name: http
      targetPort: 12345
      nodePort: 32766
  selector:
    # select app=ingress-nginx pods
    app: gpushare
    component: gpushare-schd-extender
  1. 在/etc/kubernetes目錄下新增排程策略配置檔案
#scheduler-policy-config.yaml
---
apiVersion: kubescheduler.config.k8s.io/v1beta2
kind: KubeSchedulerConfiguration
clientConnection:
  kubeconfig: /etc/kubernetes/scheduler.conf
extenders:
    # 不知道為什麼不支援svc的方式呼叫，必須用nodeport
  - urlPrefix: "http://gpushare-schd-extender.kube-system:12345/gpushare-scheduler"
    filterVerb: filter
    bindVerb: bind
    enableHTTPS: false
    nodeCacheCapable: true
    managedResources:
      - name: aliyun.com/gpu-mem
        ignoredByScheduler: false
    ignorable: false

上面的 http://gpushare-schd-extender.kube-system:12345 注意要替換為你本地部署的{nodeIP}:{gpushare-schd-extender的nodeport埠}，否則會訪問不到

查詢命令如下：

kubectl get service gpushare-schd-extender -n kube-system -o jsonpath='{.spec.ports[?(@.name=="http")].nodePort}'
  1. 修改kubernetes排程配置 /etc/kubernetes/manifests/kube-scheduler.yaml
1. 在commond中新增
 - --config=/etc/kubernetes/scheduler-policy-config.yaml
​
2. 新增pod掛載目錄
在volumeMounts:中新增
- mountPath: /etc/kubernetes/scheduler-policy-config.yaml
  name: scheduler-policy-config
  readOnly: true
在volumes:中新增
- hostPath:
      path: /etc/kubernetes/scheduler-policy-config.yaml
      type: FileOrCreate
  name: scheduler-policy-config

注意：這裡千萬不要改錯，否則可能會出現莫名其妙的錯誤
示例如下：

  1. 配置rbac及安裝device外掛
kubectl create -f https://raw.githubusercontent.com/AliyunContainerService/gpushare-device-plugin/master/device-plugin-rbac.yaml
kubectl create -f https://raw.githubusercontent.com/AliyunContainerService/gpushare-device-plugin/master/device-plugin-ds.yaml

5. 在GPU節點上新增標籤

kubectl label node <target_node> gpushare=true

6. 安裝kubectl Gpu 外掛

cd /usr/bin/
wget https://github.com/AliyunContainerService/gpushare-device-plugin/releases/download/v0.3.0/kubectl-inspect-gpushare
chmod u+x /usr/bin/kubectl-inspect-gpushare

7. 驗證

  1. 使用kubectl查詢GPU資源使用情況
# kubectl inspect gpushare
NAME                                IPADDRESS     GPU0(Allocated/Total)  GPU Memory(GiB)
cn-shanghai.i-uf61h64dz1tmlob9hmtb  192.168.0.71  6/15                   6/15
cn-shanghai.i-uf61h64dz1tmlob9hmtc  192.168.0.70  3/15                   3/15
------------------------------------------------------------------------------
Allocated/Total GPU Memory In Cluster:
9/30 (30%)
  1. 建立一個有GPU需求的資源，檢視其資源排程情況
apiVersion: apps/v1
kind: Deployment
metadata:
  name: binpack-1
  labels:
    app: binpack-1
spec:
  replicas: 1
  selector: # define how the deployment finds the pods it manages
    matchLabels:
      app: binpack-1
  template: # define the pods specifications
    metadata:
      labels:
        app: binpack-1
    spec:
      tolerations:
        - effect: NoSchedule
          key: cloudClusterNo
          operator: Exists        
      containers:
        - name: binpack-1
          image: cheyang/gpu-player:v2
          resources:
            limits:
              # 單位GiB
              aliyun.com/gpu-mem: 3

8. 問題排查

如果在安裝過程中發現資源未安裝成功，可以透過pod檢視日誌

kubectl get po -n kube-system -o=wide | grep gpushare-device 
kubecl logs -n kube-system <pod_name>

參考地址：
NVIDA官網container-toolkit安裝文件: https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html#docker
阿里雲GPU外掛安裝：https://github.com/AliyunContainerService/gpushare-scheduler-extender/blob/master/docs/install.md

相關文章