当前位置：首页 > news >正文

VPA 垂直扩缩容实践：从 Pod 到容器

news 2026/5/26 11:53:21

VPA 垂直扩缩容实践：从 Pod 到容器

前言

哥们，别整那些花里胡哨的理论。今天直接上硬菜——我在大厂一线使用 VPA 实现垂直扩缩容的真实经验总结。作为一个白天写前端、晚上打鼓的硬核工程师，我对资源优化的追求就像对鼓点节奏的把控一样严格。

背景

最近我们团队的 Kubernetes 集群出现了资源分配不合理的问题，有些 Pod 资源过剩，有些则资源不足。经过一周的 VPA 实践，我们实现了资源的动态调整，成本降低了 30%，应用性能提升了 20%。今天就把这些干货分享给大家。

VPA 基础配置

1. 安装 VPA

问题：如何在 Kubernetes 集群中安装 VPA。

解决方案：直接上代码

# 克隆 VPA 仓库 git clone https://github.com/kubernetes/autoscaler.git cd autoscaler/vertical-pod-autoscaler # 部署 VPA ./hack/vpa-up.sh # 验证安装 kubectl get pods -n kube-system | grep vpa

2. 基础 VPA 配置

问题：如何配置 VPA 实现自动资源调整。

解决方案：

# VPA 基础配置 apiVersion: autoscaling.k8s.io/v1 kind: VerticalPodAutoscaler metadata: name: music-app-vpa namespace: default spec: targetRef: apiVersion: apps/v1 kind: Deployment name: music-app updatePolicy: updateMode: "Auto" resourcePolicy: containerPolicies: - containerName: '*' minAllowed: cpu: 100m memory: 128Mi maxAllowed: cpu: 2000m memory: 2Gi controlledResources: ["cpu", "memory"] controlledValues: RequestsAndLimits

更新策略

1. 自动更新模式

问题：如何实现 Pod 的自动重启更新。

解决方案：

# 自动更新模式 apiVersion: autoscaling.k8s.io/v1 kind: VerticalPodAutoscaler metadata: name: music-app-vpa-auto namespace: default spec: targetRef: apiVersion: apps/v1 kind: Deployment name: music-app updatePolicy: updateMode: "Auto" evictionRequirements: - resources: ["cpu", "memory"] changeRequirement: TargetHigherThanRequests resourcePolicy: containerPolicies: - containerName: music-app minAllowed: cpu: 100m memory: 128Mi maxAllowed: cpu: 2000m memory: 2Gi

2. 推荐模式

问题：如何只获取推荐而不自动更新。

解决方案：

# 推荐模式 apiVersion: autoscaling.k8s.io/v1 kind: VerticalPodAutoscaler metadata: name: music-app-vpa-rec namespace: default spec: targetRef: apiVersion: apps/v1 kind: Deployment name: music-app updatePolicy: updateMode: "Off" resourcePolicy: containerPolicies: - containerName: music-app minAllowed: cpu: 100m memory: 128Mi maxAllowed: cpu: 2000m memory: 2Gi

# 查看推荐 kubectl get vpa music-app-vpa-rec -n default -o yaml # 获取推荐值 kubectl describe vpa music-app-vpa-rec -n default

资源策略

1. 容器级别策略

问题：如何为不同容器设置不同的资源策略。

解决方案：

# 多容器 VPA 配置 apiVersion: autoscaling.k8s.io/v1 kind: VerticalPodAutoscaler metadata: name: multi-container-vpa namespace: default spec: targetRef: apiVersion: apps/v1 kind: Deployment name: music-app updatePolicy: updateMode: "Auto" resourcePolicy: containerPolicies: - containerName: music-app minAllowed: cpu: 200m memory: 256Mi maxAllowed: cpu: 2000m memory: 2Gi controlledResources: ["cpu", "memory"] - containerName: sidecar-proxy minAllowed: cpu: 50m memory: 64Mi maxAllowed: cpu: 500m memory: 512Mi controlledResources: ["cpu"] - containerName: logging-agent mode: "Off"

2. 历史数据权重

问题：如何调整历史数据的权重。

解决方案：

# 配置历史数据权重 apiVersion: autoscaling.k8s.io/v1 kind: VerticalPodAutoscaler metadata: name: music-app-vpa-history namespace: default spec: targetRef: apiVersion: apps/v1 kind: Deployment name: music-app updatePolicy: updateMode: "Auto" resourcePolicy: containerPolicies: - containerName: '*' minAllowed: cpu: 100m memory: 128Mi maxAllowed: cpu: 2000m memory: 2Gi controlledResources: ["cpu", "memory"]

VPA 与 HPA 共存

1. 分离资源控制

问题：如何让 VPA 和 HPA 同时工作。

解决方案：

# HPA 只控制内存，VPA 控制 CPU apiVersion: autoscaling/v2 kind: HorizontalPodAutoscaler metadata: name: music-app-hpa namespace: default spec: scaleTargetRef: apiVersion: apps/v1 kind: Deployment name: music-app minReplicas: 3 maxReplicas: 50 metrics: - type: Resource resource: name: memory target: type: Utilization averageUtilization: 80 --- apiVersion: autoscaling.k8s.io/v1 kind: VerticalPodAutoscaler metadata: name: music-app-vpa namespace: default spec: targetRef: apiVersion: apps/v1 kind: Deployment name: music-app updatePolicy: updateMode: "Auto" resourcePolicy: containerPolicies: - containerName: '*' minAllowed: cpu: 100m memory: 128Mi maxAllowed: cpu: 2000m memory: 2Gi controlledResources: ["cpu"]

2. 基于自定义指标的共存

问题：如何基于自定义指标实现 HPA 和 VPA 共存。

解决方案：

# 基于自定义指标的 HPA apiVersion: autoscaling/v2 kind: HorizontalPodAutoscaler metadata: name: music-app-hpa-custom namespace: default spec: scaleTargetRef: apiVersion: apps/v1 kind: Deployment name: music-app minReplicas: 3 maxReplicas: 100 metrics: - type: Pods pods: metric: name: http_requests_per_second target: type: AverageValue averageValue: "100" --- apiVersion: autoscaling.k8s.io/v1 kind: VerticalPodAutoscaler metadata: name: music-app-vpa-resources namespace: default spec: targetRef: apiVersion: apps/v1 kind: Deployment name: music-app updatePolicy: updateMode: "Auto" resourcePolicy: containerPolicies: - containerName: '*' minAllowed: cpu: 100m memory: 128Mi maxAllowed: cpu: 2000m memory: 2Gi controlledResources: ["cpu", "memory"]