KubeSphere教程FG023-KubeSphere集群弹性伸缩与自动扩缩容实战
本教程详细介绍KubeSphere中集群弹性伸缩与自动扩缩容的实战操作，包括基础概念、生产环境规划、具体实施方案和实战案例。风哥教程参考KubeSphere官方文档KubeSphere容器平台使用指南、KubeSphere集群管理等相关内容。
目录大纲
Part01-基础概念与理论知识
1.1 集群弹性伸缩核心概念
集群弹性伸缩是指根据集群的资源使用情况，自动添加或移除集群节点，以适应业务负载的变化。它包括： 风哥提示：
- 节点自动扩容：当集群资源不足时，自动添加新节点
- 节点自动缩容：当集群资源过剩时，自动移除多余节点
- 资源平衡：确保资源在节点间均匀分布
- 成本优化：根据实际需求调整资源，避免资源浪费
1.2 自动扩缩容核心概念
自动扩缩容是指根据应用的负载情况，自动调整Pod的数量，以适应业务流量的变化。它包括：
- Horizontal Pod Autoscaler (HPA)：根据CPU、内存等指标自动调整Pod数量
- Vertical Pod Autoscaler (VPA)：根据资源使用情况自动调整Pod的资源请求和限制
- Cluster Autoscaler：根据集群资源使用情况自动调整节点数量
- Custom Metrics Autoscaler：根据自定义指标自动调整Pod数量
1.3 扩缩容策略与算法
常用的扩缩容策略与算法包括：
- 基于CPU/内存使用率的扩缩容：当CPU或内存使用率超过阈值时进行扩容，低于阈值时进行缩容
- 基于自定义指标的扩缩容：根据业务自定义指标（如QPS、连接数等）进行扩缩容
- 基于时间的扩缩容：根据业务流量的时间模式（如高峰期、低峰期）进行扩缩容
- 基于预测的扩缩容：使用机器学习算法预测未来的负载情况，提前进行扩缩容
Part02-生产环境规划与建议
2.1 资源规划
在实施集群弹性伸缩与自动扩缩容时，资源规划是非常重要的：
- 节点资源规划：根据应用需求，规划节点的CPU、内存等资源
- Pod资源规划：为每个Pod设置合理的资源请求和限制
- 扩缩容范围规划：确定集群的最小和最大节点数
- 资源预留：为系统组件和关键应用预留足够的资源
2.2 扩缩容策略规划
扩缩容策略规划对于保障应用的可用性和性能也非常重要：
- 扩容策略：确定何时进行扩容，扩容的步长和上限
- 缩容策略：确定何时进行缩容，缩容的步长和下限
- 冷却时间：设置扩缩容的冷却时间，避免频繁扩缩容
- 优先级策略：为不同的应用设置不同的扩缩容优先级
2.3 监控规划
监控规划是集群弹性伸缩与自动扩缩容的基础：
- 指标监控：监控CPU、内存、网络等指标
- 应用监控：监控应用的响应时间、QPS等指标
- 扩缩容事件监控：监控扩缩容事件，及时发现问题
- 告警配置：为扩缩容相关的指标设置告警
Part03-生产环境项目实施方案
3.1 集群弹性伸缩配置
在KubeSphere中，集群弹性伸缩的配置步骤： 学习交流加群风哥微信: itpux-com
- 启用Cluster Autoscaler：在集群中启用Cluster Autoscaler
- 配置节点池：创建可自动扩缩容的节点池
- 设置扩缩容参数：设置最小和最大节点数、扩缩容触发条件等
- 测试扩缩容：验证集群弹性伸缩的效果
3.2 Horizontal Pod Autoscaler (HPA)配置
HPA的配置步骤：
- 部署应用：部署需要自动扩缩容的应用
- 配置资源请求和限制：为应用设置合理的资源请求和限制
- 创建HPA：创建HPA资源，设置扩缩容的触发条件和范围
- 测试HPA：验证HPA的自动扩缩容效果
3.3 Cluster Autoscaler配置
Cluster Autoscaler的配置步骤：
- 安装Cluster Autoscaler：在集群中安装Cluster Autoscaler
- 配置节点组：配置可自动扩缩容的节点组
- 设置扩缩容参数：设置最小和最大节点数、扩缩容触发条件等
- 测试Cluster Autoscaler：验证集群弹性伸缩的效果
Part04-生产案例与实战讲解
4.1 HPA自动扩缩容实战
下面我们来实战演示HPA自动扩缩容： 学习交流加群风哥QQ113257174
# 部署应用
cat > app-deployment.yaml << EOF
apiVersion: apps/v1
kind: Deployment
metadata:
name: app
namespace: fgedu
spec:
replicas: 2
selector:
matchLabels:
app: app
template:
metadata:
labels:
app: app
spec:
containers:
– name: app
image: nginx:1.19.10
resources:
requests:
cpu: “100m”
memory: “100Mi”
limits:
cpu: “500m”
memory: “500Mi”
ports:
– containerPort: 80
EOF
kubectl apply -f app-deployment.yaml
deployment.apps/app created
cat > app-deployment.yaml << EOF
apiVersion: apps/v1
kind: Deployment
metadata:
name: app
namespace: fgedu
spec:
replicas: 2
selector:
matchLabels:
app: app
template:
metadata:
labels:
app: app
spec:
containers:
– name: app
image: nginx:1.19.10
resources:
requests:
cpu: “100m”
memory: “100Mi”
limits:
cpu: “500m”
memory: “500Mi”
ports:
– containerPort: 80
EOF
kubectl apply -f app-deployment.yaml
deployment.apps/app created
# 创建服务
cat > app-service.yaml << EOF
apiVersion: v1
kind: Service
metadata:
name: app
namespace: fgedu
spec:
selector:
app: app
ports:
– port: 80
targetPort: 80
EOF
kubectl apply -f app-service.yaml
service/app created
cat > app-service.yaml << EOF
apiVersion: v1
kind: Service
metadata:
name: app
namespace: fgedu
spec:
selector:
app: app
ports:
– port: 80
targetPort: 80
EOF
kubectl apply -f app-service.yaml
service/app created
# 创建HPA
kubectl autoscale deployment app –namespace=fgedu –cpu-percent=50 –min=2 –max=10
horizontalpodautoscaler.autoscaling/app created
kubectl autoscale deployment app –namespace=fgedu –cpu-percent=50 –min=2 –max=10
horizontalpodautoscaler.autoscaling/app created
# 检查HPA状态
kubectl get hpa –namespace=fgedu
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
app Deployment/app 0%/50% 2 10 2 1m
kubectl get hpa –namespace=fgedu
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
app Deployment/app 0%/50% 2 10 2 1m
# 生成负载
kubectl run -i –tty load-generator –image=busybox –namespace=fgedu — /bin/sh -c “while true; do wget -q -O – http://app; done”
# 持续输出响应内容
kubectl run -i –tty load-generator –image=busybox –namespace=fgedu — /bin/sh -c “while true; do wget -q -O – http://app; done”
# 持续输出响应内容
# 查看HPA状态（在另一个终端）
kubectl get hpa –namespace=fgedu -w
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
app Deployment/app 300%/50% 2 10 2 5m
app Deployment/app 300%/50% 2 10 4 5m
app Deployment/app 150%/50% 2 10 4 6m
app Deployment/app 75%/50% 2 10 8 7m
app Deployment/app 37%/50% 2 10 8 8m
kubectl get hpa –namespace=fgedu -w
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
app Deployment/app 300%/50% 2 10 2 5m
app Deployment/app 300%/50% 2 10 4 5m
app Deployment/app 150%/50% 2 10 4 6m
app Deployment/app 75%/50% 2 10 8 7m
app Deployment/app 37%/50% 2 10 8 8m
4.2 Cluster Autoscaler集群弹性伸缩实战
下面我们来实战演示Cluster Autoscaler集群弹性伸缩： 更多视频教程www.fgedu.net.cn
# 安装Cluster Autoscaler
kubectl apply -f https://raw.githubusercontent.com/kubernetes/autoscaler/master/cluster-autoscaler/cloudprovider/aws/examples/cluster-autoscaler-autodiscover.yaml
# 具体安装命令根据云厂商不同而不同
kubectl apply -f https://raw.githubusercontent.com/kubernetes/autoscaler/master/cluster-autoscaler/cloudprovider/aws/examples/cluster-autoscaler-autodiscover.yaml
# 具体安装命令根据云厂商不同而不同
# 配置Cluster Autoscaler
kubectl edit deployment cluster-autoscaler -n kube-system
# 在命令行参数中添加以下内容：
–balance-similar-node-groups=true
–expander=least-waste
–node-group-auto-discovery=asg:tag=k8s.io/cluster-autoscaler/enabled,k8s.io/cluster-autoscaler/my-cluster
kubectl edit deployment cluster-autoscaler -n kube-system
# 在命令行参数中添加以下内容：
–balance-similar-node-groups=true
–expander=least-waste
–node-group-auto-discovery=asg:tag=k8s.io/cluster-autoscaler/enabled,k8s.io/cluster-autoscaler/my-cluster
# 检查Cluster Autoscaler状态
kubectl get pods -n kube-system | grep cluster-autoscaler
cluster-autoscaler-5f984b76c5-4x7k8 1/1 Running 0 5m
kubectl get pods -n kube-system | grep cluster-autoscaler
cluster-autoscaler-5f984b76c5-4x7k8 1/1 Running 0 5m
# 生成负载，触发集群扩容
cat > high-load-deployment.yaml << EOF
apiVersion: apps/v1
kind: Deployment
metadata:
name: high-load-app
namespace: fgedu
spec:
replicas: 20
selector:
matchLabels:
app: high-load-app
template:
metadata:
labels:
app: high-load-app
spec:
containers:
– name: app
image: nginx:1.19.10
resources:
requests:
cpu: “1”
memory: “1Gi”
limits:
cpu: “2”
memory: “2Gi”
ports:
– containerPort: 80
EOF
kubectl apply -f high-load-deployment.yaml
deployment.apps/high-load-app created
cat > high-load-deployment.yaml << EOF
apiVersion: apps/v1
kind: Deployment
metadata:
name: high-load-app
namespace: fgedu
spec:
replicas: 20
selector:
matchLabels:
app: high-load-app
template:
metadata:
labels:
app: high-load-app
spec:
containers:
– name: app
image: nginx:1.19.10
resources:
requests:
cpu: “1”
memory: “1Gi”
limits:
cpu: “2”
memory: “2Gi”
ports:
– containerPort: 80
EOF
kubectl apply -f high-load-deployment.yaml
deployment.apps/high-load-app created
# 查看集群节点状态
kubectl get nodes -w
NAME STATUS ROLES AGE VERSION
fgedu-master-01 Ready control-plane,master 1d v1.26.0
fgedu-worker-01 Ready worker 1d v1.26.0
fgedu-worker-02 Ready worker 1d v1.26.0
fgedu-worker-03 Ready worker 5m v1.26.0 # 新扩容的节点
fgedu-worker-04 Ready worker 3m v1.26.0 # 新扩容的节点
kubectl get nodes -w
NAME STATUS ROLES AGE VERSION
fgedu-master-01 Ready control-plane,master 1d v1.26.0
fgedu-worker-01 Ready worker 1d v1.26.0
fgedu-worker-02 Ready worker 1d v1.26.0
fgedu-worker-03 Ready worker 5m v1.26.0 # 新扩容的节点
fgedu-worker-04 Ready worker 3m v1.26.0 # 新扩容的节点
4.3 混合扩缩容策略实战
下面我们来实战演示混合扩缩容策略： 更多学习教程公众号风哥教程itpux_com
# 部署应用
cat > mixed-app-deployment.yaml << EOF
apiVersion: apps/v1
kind: Deployment
metadata:
name: mixed-app
，
namespace: fgedu
spec:
replicas: 3
selector:
matchLabels:
app: mixed-app
template:
metadata:
labels:
app: mixed-app
spec:
containers:
– name: app
image: nginx:1.19.10
resources:
requests:
cpu: “200m”
memory: “200Mi”
limits:
cpu: “1”
memory: “1Gi”
ports:
– containerPort: 80
EOF
kubectl apply -f mixed-app-deployment.yaml
deployment.apps/mixed-app created
cat > mixed-app-deployment.yaml << EOF
apiVersion: apps/v1
kind: Deployment
metadata:
name: mixed-app
，
namespace: fgedu
spec:
replicas: 3
selector:
matchLabels:
app: mixed-app
template:
metadata:
labels:
app: mixed-app
spec:
containers:
– name: app
image: nginx:1.19.10
resources:
requests:
cpu: “200m”
memory: “200Mi”
limits:
cpu: “1”
memory: “1Gi”
ports:
– containerPort: 80
EOF
kubectl apply -f mixed-app-deployment.yaml
deployment.apps/mixed-app created
# 创建基于自定义指标的HPA
cat > custom-metrics-hpa.yaml << EOF
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: mixed-app-hpa
namespace: fgedu
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: mixed-app
minReplicas: 3
maxReplicas: 15
metrics:
– type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 60
– type: Resource
resource:
name: memory
target:
type: Utilization
averageUtilization: 70
EOF
kubectl apply -f custom-metrics-hpa.yaml
horizontalpodautoscaler.autoscaling/mixed-app-hpa created
cat > custom-metrics-hpa.yaml << EOF
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: mixed-app-hpa
namespace: fgedu
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: mixed-app
minReplicas: 3
maxReplicas: 15
metrics:
– type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 60
– type: Resource
resource:
name: memory
target:
type: Utilization
averageUtilization: 70
EOF
kubectl apply -f custom-metrics-hpa.yaml
horizontalpodautoscaler.autoscaling/mixed-app-hpa created
# 查看HPA状态
kubectl get hpa –namespace=fgedu
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
mixed-app-hpa Deployment/mixed-app 0%/60%, 0%/70% 3 15 3 1m
kubectl get hpa –namespace=fgedu
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
mixed-app-hpa Deployment/mixed-app 0%/60%, 0%/70% 3 15 3 1m
Part05-风哥经验总结与分享
5.1 常见问题与解决方案
在实施集群弹性伸缩与自动扩缩容时，常见的问题及解决方案： from K8S+DB视频:www.itpux.com
- HPA不触发扩容：检查资源请求和限制是否设置正确，确保指标采集正常
- 集群扩容失败：检查节点池配置，确保有足够的资源配额
- 缩容后Pod调度失败：检查节点资源是否足够，调整Pod的资源请求
- 频繁扩缩容：调整扩缩容的阈值和冷却时间，避免震荡
5.2 最佳实践建议
集群弹性伸缩与自动扩缩容的最佳实践：
- 设置合理的资源请求和限制：根据应用的实际需求设置资源请求和限制
- 选择合适的扩缩容策略：根据应用的特点选择合适的扩缩容策略
- 设置合理的扩缩容阈值：避免过于敏感的扩缩容触发条件
- 配置适当的冷却时间：避免频繁扩缩容
- 监控扩缩容事件：及时发现和解决扩缩容过程中的问题
5.3 性能优化技巧
集群弹性伸缩与自动扩缩容的性能优化技巧：
- 使用节点亲和性和反亲和性：优化Pod的分布，提高资源利用率
- 合理配置Pod的资源请求和限制：避免资源浪费和资源竞争
- 使用多种扩缩容策略：结合CPU、内存和自定义指标进行扩缩容
- 优化集群的节点池配置：根据应用需求配置不同类型的节点池
- 定期分析和调整扩缩容策略：根据实际运行情况调整扩缩容策略
在实施集群弹性伸缩与自动扩缩容时，一定要设置合理的资源请求和限制，避免资源浪费和资源竞争。
本文由风哥教程整理发布,仅用于学习测试使用,转载注明出处:http://www.fgedu.net.cn/10327.html
