k8s helm安装VictoriaMetrics介绍

以下内容由ai生成

一、VictoriaMetrics 介绍

VictoriaMetrics 是一款高性能、高压缩率、易扩展的开源时序数据库（TSDB），专为云原生监控、APM、IoT 等大规模时序数据场景设计，是 Prometheus 主流的高性能替代 / 长期存储方案。

1. 核心优势

极致性能
- 写入：单机支持数百万指标 / 秒写入VictoriaMetrics
- 查询：复杂查询比 Prometheus 快 2~10 倍，大时间范围查询优势更明显
- 内存占用：比 Prometheus 低 3~4 倍
超高压缩
- 自研高效压缩算法，相同数据磁盘占用仅为 Prometheus 的 1/7~1/10VictoriaMetrics
架构简洁
- 单二进制文件，无任何外部依赖（无需 ZooKeeper、Etcd 等）
- 数据存储在单一目录，备份 / 迁移 / 运维极简
完美兼容 Prometheus
- 完全兼容 PromQL（扩展为 MetricsQL）
- 兼容 Prometheus 远程读写、服务发现、AlertManager、Grafana 面板
- 可无缝替代 Prometheus 或作为其远程长期存储
两种部署模式
- 单机版（VMSingle）：中小规模，一键部署，高可用靠备份
- 集群版（Cluster）：大规模场景，无共享分布式架构（SN 架构），组件独立伸缩VictoriaMetrics

2. 集群版核心组件VictoriaMetrics

vmstorage：有状态存储节点，负责数据存储与查询，节点间无通信、不共享数据
vminsert：无状态写入代理，接收指标并哈希分片到多个 vmstorage
vmselect：无状态查询代理，分发查询到所有 vmstorage 并聚合结果

vmagent：轻量采集器，替代 Prometheus 采集，支持 PULL/PUSH，可远程写入

二、K8s 环境部署 VictoriaMetrics（推荐方式二）

推荐两种主流方式：Helm 部署集群版（生产推荐）、Operator 部署（CRD 化管理）。

方式一：Helm 部署 VictoriaMetrics 集群（生产标准）

前提

Kubernetes 1.21+
Helm 3.10+
kubectl 有权限

1. 添加 Helm 仓库

helm repo add vm https://victoriametrics.github.io/helm-charts/
helm repo update

2. 部署集群（3 副本高可用）

bash

运行

helm install victoria-metrics vm/victoria-metrics-cluster \
  --namespace monitoring \
  --create-namespace \
  --set vmstorage.replicaCount=3 \
  --set vminsert.replicaCount=2 \
  --set vmselect.replicaCount=2 \
  --set vmstorage.persistentVolume.size=50Gi \
  --set vmstorage.persistentVolume.storageClassName=gp2

3. 验证部署

bash

运行

kubectl get pods -n monitoring -l app=victoria-metrics-cluster

4. 服务访问（内部）

写入地址：http://victoria-metrics-cluster-vminsert.monitoring.svc:8480/insert/0/prometheus/
查询地址：http://victoria-metrics-cluster-vmselect.monitoring.svc:8481/select/0/prometheus/

5. 暴露服务（可选，NodePort/Ingress）

bash

运行

# NodePort 暴露查询
kubectl expose svc victoria-metrics-cluster-vmselect \
  --name vmselect-nodeport \
  --type NodePort \
  --port 8481 \
  --namespace monitoring

方式二：Operator 部署（CRD 化，灵活管理）

1. 安装 Operator

helm repo add vm https://victoriametrics.github.io/helm-charts/
helm repo update

helm install vm-operator vm/victoria-metrics-operator \
--namespace monitoring \
--set operator.replicaCount=1 \
--set tolerations[0].key=app \
--set tolerations[0].operator=Equal \
--set tolerations[0].value=vmstack-saas \
--set tolerations[0].effect=NoSchedule

（可选）指定节点组

helm install vm-operator vm/victoria-metrics-operator \
  --namespace monitoring \
  --set operator.replicaCount=1 \
  --set tolerations[0].key=app \
  --set tolerations[0].operator=Equal \
  --set tolerations[0].value=vmstack-saas \
  --set tolerations[0].effect=NoSchedule \
  --set operator.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms[0].matchExpressions[0].key=eks.amazonaws.com/nodegroup \
  --set operator.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms[0].matchExpressions[0].operator=In \
  --set operator.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms[0].matchExpressions[0].values[0]=eks-prod-paimatix-shared-critical-Az2 \
  --set operator.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms[0].matchExpressions[0].values[1]=eks-prod-paimatix-shared-critical-Az3

2. 安装 victoria组件

部署单机版（VMSingle）示例

yaml

# vm-single.yaml
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMSingle
metadata:
  name: vm-single
  namespace: monitoring
spec:
  retentionPeriod: "30d"
  storage:
    volumeClaimTemplate:
      spec:
        resources:
          requests:
            storage: 50Gi
        storageClassName: local-path
  resources:
    requests:
      cpu: "1"
      memory: "2Gi"

kubectl apply -f vm-single.yaml

部署集群版（VMCluster）示例

cat > vmcluster.yaml << EOF
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMCluster
metadata:
  name: vm
  namespace: monitoring
spec:
  retentionPeriod: "365d"
  vmstorage:
    replicaCount: 3
    storage:
      volumeClaimTemplate:
        spec:
          storageClassName: alicloud-disk-ssd
          resources:
            requests:
              storage: 50Gi
    tolerations:
    - key: "app"
      operator: "Equal"
      value: "vmstack-saas"
      effect: "NoSchedule"
  vminsert:
    replicaCount: 2
    tolerations:
    - key: "app"
      operator: "Equal"
      value: "vmstack-saas"
      effect: "NoSchedule"
  vmselect:
    replicaCount: 2
    tolerations:
    - key: "app"
      operator: "Equal"
      value: "vmstack-saas"
      effect: "NoSchedule"
EOF

（选做）部署集群版（VMCluster），配置指定节点组示例

cat > vmcluster.yaml << EOF
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMCluster
metadata:
  name: vm
  namespace: monitoring
spec:
  retentionPeriod: "365d"
  vmstorage:
    replicaCount: 3
    storage:
      volumeClaimTemplate:
        spec:
          storageClassName: alicloud-disk-ssd
          resources:
            requests:
              storage: 50Gi
    tolerations:
    - key: "app"
      operator: "Equal"
      value: "vmstack-saas"
      effect: "NoSchedule"
    # 添加节点亲和性
    affinity:
      nodeAffinity:
        requiredDuringSchedulingIgnoredDuringExecution:
          nodeSelectorTerms:
          - matchExpressions:
            - key: eks.amazonaws.com/nodegroup
              operator: In
              values:
              - eks-prod-paimatix-shared-critical-Az2
              - eks-prod-paimatix-shared-critical-Az3
  vminsert:
    replicaCount: 2
    tolerations:
    - key: "app"
      operator: "Equal"
      value: "vmstack-saas"
      effect: "NoSchedule"
    # 添加节点亲和性
    affinity:
      nodeAffinity:
        requiredDuringSchedulingIgnoredDuringExecution:
          nodeSelectorTerms:
          - matchExpressions:
            - key: eks.amazonaws.com/nodegroup
              operator: In
              values:
              - eks-prod-paimatix-shared-critical-Az2
              - eks-prod-paimatix-shared-critical-Az3
  vmselect:
    replicaCount: 2
    tolerations:
    - key: "app"
      operator: "Equal"
      value: "vmstack-saas"
      effect: "NoSchedule"
    # 添加节点亲和性
    affinity:
      nodeAffinity:
        requiredDuringSchedulingIgnoredDuringExecution:
          nodeSelectorTerms:
          - matchExpressions:
            - key: eks.amazonaws.com/nodegroup
              operator: In
              values:
              - eks-prod-paimatix-shared-critical-Az2
              - eks-prod-paimatix-shared-critical-Az3
EOF

运行

kubectl apply -f vmcluster.yaml

部署完成后有如下信息，（方式一举例）

[ec2-user@paimatix-dev-kubectl ~]$ helm install victoria-metrics vm/victoria-metrics-cluster \
  --namespace monitoring \
  --create-namespace \
  --set vmstorage.replicaCount=3 \
  --set vminsert.replicaCount=2 \
  --set vmselect.replicaCount=2 \
  --set vmstorage.persistentVolume.size=50Gi \
  --set vmstorage.persistentVolume.storageClass=gp2
NAME: victoria-metrics
LAST DEPLOYED: Tue Apr 14 08:03:16 2026
NAMESPACE: monitoring
STATUS: deployed
REVISION: 1
TEST SUITE: None
NOTES:
Write API:

The Victoria Metrics write api can be accessed via port 8480 with the following DNS name from within your cluster:
victoria-metrics-victoria-metrics-cluster-vminsert.monitoring.svc.cluster.local.

Get the Victoria Metrics insert service URL by running these commands in the same shell:
  export POD_NAME=$(kubectl get pods --namespace monitoring -l "app=vminsert" -l "app.kubernetes.io/instance=victoria-metrics" -o jsonpath="{.items[0].metadata.name}")
  kubectl --namespace monitoring port-forward $POD_NAME 8480

You need to update your Prometheus configuration file and add the following lines to it:

prometheus.yml

    remote_write:
      - url: "http://<insert-service>/insert/0/prometheus/"

for example -  inside the Kubernetes cluster:

    remote_write:
      - url: http://victoria-metrics-victoria-metrics-cluster-vminsert.monitoring.svc.cluster.local.:8480/insert/0/prometheus/
Read API:

The VictoriaMetrics read api can be accessed via port 8481 with the following DNS name from within your cluster:
victoria-metrics-victoria-metrics-cluster-vmselect.monitoring.svc.cluster.local.

Get the VictoriaMetrics select service URL by running these commands in the same shell:
  export POD_NAME=$(kubectl get pods --namespace monitoring -l "app=vmselect" -l "app.kubernetes.io/instance=victoria-metrics" -o jsonpath="{.items[0].metadata.name}")
  kubectl --namespace monitoring port-forward $POD_NAME 8481

You need to specify select service URL into your Grafana:
 NOTE: you need to use the Prometheus Data Source

Input this URL field into Grafana

    http://<select-service>/select/0/prometheus/


for example - inside the Kubernetes cluster:

    http://victoria-metrics-victoria-metrics-cluster-vmselect.monitoring.svc.cluster.local.:8481/select/0/prometheus/

三、安装vmagent

方式一：Helm 部署 VictoriaMetrics 集群

例子：监控kafka

1、安装vmagent

cat > vmagent.yaml << 'EOF'
apiVersion: v1
kind: ConfigMap
metadata:
  name: vmagent-config
  namespace: monitoring
data:
  prometheus.yml: |
    global:
      scrape_interval: 15s

    # 专门采集 kafka-exporter → job 名正确
    scrape_configs:
      - job_name: 'kafka-exporter'
        static_configs:
          - targets:
            - 'kafka-exporter.monitoring.svc:9308'

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: vmagent
  namespace: monitoring
spec:
  replicas: 1
  selector:
    matchLabels:
      app: vmagent
  template:
    metadata:
      labels:
        app: vmagent
    spec:
      serviceAccountName: vmagent
      containers:
      - name: vmagent
        image: victoriametrics/vmagent:latest
        args:
        - --promscrape.config=/etc/prometheus/prometheus.yml
        - --remoteWrite.url=http://victoria-metrics-victoria-metrics-cluster-vminsert.monitoring.svc:8480/insert/0/prometheus/
        volumeMounts:
        - name: config
          mountPath: /etc/prometheus
      volumes:
      - name: config
        configMap:
          name: vmagent-config
EOF

2、给vmagent权限，不然vmagent它没有权限去 monitoring 命名空间 查看 Pod

cat > vmagent-rbac.yaml << EOF
apiVersion: v1
kind: ServiceAccount
metadata:
  name: vmagent
  namespace: monitoring
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: vmagent
rules:
- apiGroups: [""]
  resources:
  - nodes
  - nodes/proxy
  - services
  - endpoints
  - pods
  verbs: ["get", "list", "watch"]
- apiGroups:
  - extensions
  - networking.k8s.io
  resources:
  - ingresses
  verbs: ["get", "list", "watch"]
- nonResourceURLs: ["/metrics"]
  verbs: ["get"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: vmagent
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: vmagent
subjects:
- kind: ServiceAccount
  name: vmagent
  namespace: monitoring
EOF

3、安装vmagent

kubectl apply -f  vmagent-rbac.yaml
kubectl apply -f  vmagent.yaml

4、安装kafka exporter

vim kafka-exporter.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: kafka-exporter
  namespace: monitoring
  labels:
    app: kafka-exporter
spec:
  replicas: 1
  selector:
    matchLabels:
      app: kafka-exporter
  template:
    metadata:
      labels:
        app: kafka-exporter
    spec:
      containers:
      - name: kafka-exporter
        image: danielqsj/kafka-exporter:latest
        args:
        - "--kafka.server=alikafka-pre-cn-dew4qq0ut001-1-vpc.alikafka.aliyuncs.com:9092"
        - "--kafka.server=alikafka-pre-cn-dew4qq0ut001-2-vpc.alikafka.aliyuncs.com:9092"
        - "--kafka.server=alikafka-pre-cn-dew4qq0ut001-3-vpc.alikafka.aliyuncs.com:9092"
        - "--web.listen-address=:9308"
        ports:
        - containerPort: 9308
          name: metrics
        resources:
          requests:
            cpu: 100m
            memory: 100Mi
          limits:
            cpu: 2000m
            memory: 2000Mi
---
# Service 带正确标签！
apiVersion: v1
kind: Service
metadata:
  name: kafka-exporter
  namespace: monitoring
  labels:
    app: kafka-exporter  # 关键标签！VMServiceScrape 靠它发现
spec:
  ports:
  - port: 9308
    name: metrics
    targetPort: 9308
  selector:
    app: kafka-exporter

kc apply -f kafka-exporter.yaml

5、验证：起一个终端

kubectl port-forward -n monitoring svc/victoria-metrics-victoria-metrics-cluster-vmselect 8481

6、验证：另外起一个终端运行

curl 'http://127.0.0.1:8481/select/0/prometheus/api/v1/query?query=kafka_brokers'

{"status":"success","isPartial":false,"data":{"resultType":"vector","result":[{"metric":{"__name__":"kafka_brokers","instance":"kafka-exporter.monitoring.svc:9308","job":"kafka-exporter"},"value":[1776158258,"3"]},{"metric":{"__name__":"kafka_brokers","instance":"10.52.99.85:9308","job":"kafka-exporter"},"value":[1776158258,"3"]}]},"stats":{"seriesFetched": "2","executionTimeMsec":3}}

方式二：Operator 部署

例子：监控kafka

1、安装vmagent,给vmagent权限，不然vmagent它没有权限去 monitoring 命名空间 查看 Pod

vim vmagent-rbac.yaml

apiVersion: v1
kind: ServiceAccount
metadata:
  name: vmagent
  namespace: monitoring
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: vmagent
rules:
- apiGroups: [""]
  resources:
  - nodes
  - services
  - endpoints
  - pods
  - secrets      # 关键：加上这个！
  verbs: ["get", "list", "watch"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: vmagent
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: vmagent
subjects:
- kind: ServiceAccount
  name: vmagent
  namespace: monitoring

vim vmagent.yaml

apiVersion: operator.victoriametrics.com/v1beta1
kind: VMAgent
metadata:
  name: vmagent
  namespace: monitoring
spec:
  replicaCount: 1
  serviceAccountName: vmagent

  # 👇 正确：选择哪些命名空间里的 VMServiceScrape
  serviceScrapeNamespaceSelector:
    matchNames:
      - monitoring
      - internal
      - production
      - stagging
  # 为 VMStaticScrape 添加命名空间选择器，让它去 internal 命名空间里找
  staticScrapeNamespaceSelector:
    matchNames:
      - monitoring
      - internal
      - production
      - stagging
  podScrapeNamespaceSelector:
    matchNames:
      - monitoring
      - internal
      - production
      - stagging
  # 👇 正确：选择所有 VMServiceScrape（空选择器 + selectAllByDefault）
  selectAllByDefault: true
  serviceScrapeSelector: {}

  podScrapeSelector: {}
  staticScrapeSelector: {}

  remoteWrite:
  - url: "http://vminsert-vm.monitoring.svc:8480/insert/0/prometheus/"

kubectl apply -f vmagent-rbac.yaml
kubectl apply -f vmagent.yaml

2、创建 VMServiceScrape

cat > vmservicescrape-kafka.yaml << EOF
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMServiceScrape
metadata:
  name: kafka-exporter
  namespace: monitoring
spec:
  # 你要的：job 名称
  jobLabel: kafka-exporter
  
  # 自动发现 Service
  selector:
    matchLabels:
      app: kafka-exporter

  # 采集端口
  endpoints:
  - port: 9308
    interval: 15s
    path: /metrics
EOF

3、安装kafka exporter

vim kafka-exporter.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: kafka-exporter
  namespace: monitoring
  labels:
    app: kafka-exporter
spec:
  replicas: 1
  selector:
    matchLabels:
      app: kafka-exporter
  template:
    metadata:
      labels:
        app: kafka-exporter
    spec:
      containers:
      - name: kafka-exporter
        image: danielqsj/kafka-exporter:latest
        args:
        - "--kafka.server=alikafka-pre-cn-dew4qq0ut001-1-vpc.alikafka.aliyuncs.com:9092"
        - "--kafka.server=alikafka-pre-cn-dew4qq0ut001-2-vpc.alikafka.aliyuncs.com:9092"
        - "--kafka.server=alikafka-pre-cn-dew4qq0ut001-3-vpc.alikafka.aliyuncs.com:9092"
        - "--web.listen-address=:9308"
        ports:
        - containerPort: 9308
          name: metrics
        resources:
          requests:
            cpu: 100m
            memory: 100Mi
          limits:
            cpu: 2000m
            memory: 2000Mi
---
# Service 带正确标签！
apiVersion: v1
kind: Service
metadata:
  name: kafka-exporter
  namespace: monitoring
  labels:
    app: kafka-exporter  # 关键标签！VMServiceScrape 靠它发现
spec:
  ports:
  - port: 9308
    name: metrics
    targetPort: 9308
  selector:
    app: kafka-exporter

kc apply -f kafka-exporter.yaml

4、创建kafka的VMServiceScrape（生产级自动采集）

vim vmservicescrape-kafka.yaml

apiVersion: operator.victoriametrics.com/v1beta1
kind: VMServiceScrape
metadata:
  name: kafka-exporter
  namespace: monitoring
spec:
  jobName: "kafka-exporter" 
  # 自动发现 Service
  selector:
    matchLabels:
      app: kafka-exporter

  # 采集端口
  endpoints:
  - port: "metrics"
    interval: 15s
    path: /metrics

kc apply -f  vmservicescrape-kafka.yaml

5、验证：起一个端口监听

kubectl port-forward -n monitoring svc/vmselect-vm 8481

6、验证：另外起一个终端访问

[ec2-user@paimatix-dev-kubectl VictoriaMetrics-operator]$  curl 'http://127.0.0.1:8481/select/0/prometheus/api/v1/query?query=kafka_brokers'
{"status":"success","isPartial":false,"data":{"resultType":"vector","result":[{"metric":{"__name__":"kafka_brokers","container":"kafka-exporter","instance":"10.52.98.189:9308","job":"kafka-exporter","namespace":"monitoring","pod":"kafka-exporter-5b7567f9d5-2p9md","prometheus":"monitoring/vmagent","service":"kafka-exporter"},"value":[1776164088,"3"]}]},"stats":{"seriesFetched": "1","executionTimeMsec":3}}

例外：查有哪些jobs

curl 'http://127.0.0.1:8481/select/0/prometheus/api/v1/label/job/values'

四、对接 Prometheus/Grafana

1. vmselect配置ingress

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  annotations:
    alb.ingress.kubernetes.io/listen-ports: '[{"HTTP": 80}]'
    alb.ingress.kubernetes.io/scheme: internet-facing
    alb.ingress.kubernetes.io/target-type: ip
    kubernetes.io/ingress.class: alb
  name: vmselect-alb
  namespace: monitoring
spec:
  rules:
  - host: vmselect-paimatix.shoplazza.site
    http:
      paths:
      - backend:
          service:
            name: vmselect-vm
            port:
              number: 8481
        path: /
        pathType: Prefix
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  annotations:
    alb.ingress.kubernetes.io/listen-ports: '[{"HTTP": 80}]'
    alb.ingress.kubernetes.io/scheme: internet-facing
    alb.ingress.kubernetes.io/target-type: ip
    kubernetes.io/ingress.class: alb
  name: vmselect-alb
  namespace: monitoring
spec:
  rules:
  - host: vmselect-paimatix.shoplazza.site
    http:
      paths:
      - backend:
          service:
            name: vmselect-vm
            port:
              number: 8481
        path: /
        pathType: Prefix

2. Prometheus 远程读入(选做)

remote_write:
  - url: "http://victoria-metrics-cluster-vminsert.monitoring.svc:8480/insert/0/prometheus/api/v1/write"

3. Grafana 配置数据源

Type：Prometheus
URL：http://vmselect.monitoring.svc:8481/select/0/prometheus（这是k8s内部地址，如果grafana和vmselect在不同k8s集群，则写alb地址）
Access：Server（推荐）

posted @ 2026-04-14 16:21 苦逼yw 阅读(133) 评论(0) 收藏举报

刷新页面返回顶部

lfxx

k8s helm安装VictoriaMetrics介绍

一、VictoriaMetrics 介绍

1. 核心优势

2. 集群版核心组件VictoriaMetrics

二、K8s 环境部署 VictoriaMetrics（推荐方式二）

方式一：Helm 部署 VictoriaMetrics 集群（生产标准）

1. 添加 Helm 仓库

2. 部署集群（3 副本高可用）

3. 验证部署

4. 服务访问（内部）

5. 暴露服务（可选，NodePort/Ingress）

方式二：Operator 部署（CRD 化，灵活管理）

1. 安装 Operator

（可选）指定节点组

2. 安装 victoria组件

部署单机版（VMSingle）示例

部署集群版（VMCluster）示例

（选做）部署集群版（VMCluster），配置指定节点组示例

部署完成后有如下信息，（方式一举例）

三、安装vmagent

方式一：Helm 部署 VictoriaMetrics 集群

方式二：Operator 部署

四、对接 Prometheus/Grafana

1. vmselect配置ingress

2. Prometheus 远程读入(选做)

3. Grafana 配置数据源