部署链路追踪系统

LOOK
本人更推荐 jaeger
不过会有一个问题，日志保存在本地，jaeger服务宕机，重新启动后日志可能混乱，经过排查是其中一个组件服务重启失败，由于时间原因没排查出具体问题，换了一个链路收集系统。
当然也可以保存在es中，这样查询会更快，但es比较吃资源。

tempo系统部署

主要功能是接收日志收集数据，压缩数据， 存储数据，提供查询接口

tempo-values.yaml

# 全局配置
global:
  image:
    registry: docker.io
  clusterDomain: cluster.local

# Tempo 基础配置
tempo:
  structuredConfig:
    compactor: {}
    distributor: {}
    ingester: {}
    server:
      http_listen_port: 3100
    storage:
      trace:
        backend: s3                                     # 使用 S3 兼容存储
        s3:
          bucket: tempo                                 # MinIO 存储桶名称
          endpoint: loki-minio.logging.svc:9000         # MinIO 服务地址
          access_key: root-user                         # MinIO 访问密钥,部署loki时指明
          secret_key: supersecretpassword               # MinIO 密钥，部署loki时指明
          insecure: true
          forcepathstyle: true                          # 使用路径风格访问

# Ingester 配置
ingester:
  replicas: 3
  persistence:
    enabled: true
    inMemory: false
    size: 10Gi
    storageClass: "nfs-client"
    annotations: {}

# Distributor 配置
distributor:
  replicas: 2
  config:
    log_received_spans:
      enabled: true
      include_all_attributes: false

# Compactor 配置
compactor:
  replicas: 1
  persistence:
    enabled: true
    inMemory: false
    size: 50Gi    # 从 10Gi 改为 50Gi
    storageClass: "nfs-client"
    annotations: {}
  config:
    compaction:
      block_retention: 48h
      compaction_window: 1h
      max_block_bytes: 107374182400  # 100GB

# Query Frontend 配置
queryFrontend:
  replicas: 2
  query:
    enabled: false  # 启用 Jaeger UI
    config: |
      max_concurrent: 2000
      max_traces_per_tenant: 10000
      max_lookback_delta: 1h
    extraArgs: 
      - "-config /conf/tempo-query.yaml"
      - "--query.base-path="  # 空值以覆盖默认值
  config:
    max_outstanding_per_tenant: 2000
    max_retries: 2
    search:
      concurrent_jobs: 1000
    trace_by_id:
      query_shards: 50

# 资源请求和限制
resources:
  ingester:
    limits:
      cpu: 1
      memory: 1Gi
    requests:
      cpu: 500m
      memory: 512Mi
  distributor:
    limits:
      cpu: 1
      memory: 1Gi
    requests:
      cpu: 500m
      memory: 512Mi
  compactor:
    limits:
      cpu: 1
      memory: 1Gi
    requests:
      cpu: 500m
      memory: 512Mi
  query_frontend:
    limits:
      cpu: 1
      memory: 1Gi
    requests:
      cpu: 500m
      memory: 512Mi

# RBAC 配置
rbac:
  create: true
  pspEnabled: false

# ServiceAccount 配置
serviceAccount:
  create: true
  annotations: {}

# 指标生成器配置
metricsGenerator:
  enabled: true
  config:
    registry:
      collection_interval: 15s
    processor:
      service_graphs:
        dimensions: []
        histogram_buckets: [0.1, 0.2, 0.4, 0.8, 1.6, 3.2, 6.4, 12.8]
      span_metrics:
        dimensions: []

# 搜索配置
search_enabled: true

# 追踪保留配置
retention:
  enabled: true
  duration: 48h

# 存活和就绪探针
livenessProbe:
  enabled: true
  initialDelaySeconds: 30
  periodSeconds: 10

readinessProbe:
  enabled: true
  initialDelaySeconds: 30
  periodSeconds: 10

traces:
  otlp:
    grpc:
      enabled: true  # 启用OTLP gRPC接收器
    http:
      enabled: true  # 可选：同时启用HTTP接收器

Helm 命令

# 添加grafana仓库
helm repo add grafana https://grafana.github.io/helm-charts

# 更新仓库
helm repo update

# 部署 tempo
helm install tempo grafana/tempo-distributed \
  --namespace logging \
  -f tempo-values.yaml