Grafana 系列文章（一）：基于 Grafana 的全栈可观察性 Demo

docker-compose up
BASH

- "3000:3000"
YAML

- "3123:3000"
YAML

mythical-requester:
  build:
    context: ./source
    dockerfile: docker/Dockerfile
    args:
      HTTP_PROXY: http://192.168.2.9:7890
      HTTPS_PROXY: http://192.168.2.9:7890
      SERVICE: mythical-beasts-requester
mythical-server:
  build:
    context: ./source
    dockerfile: docker/Dockerfile
    args:
      HTTP_PROXY: http://192.168.2.9:7890
      HTTPS_PROXY: http://192.168.2.9:7890              
      SERVICE: mythical-beasts-server
prometheus:
  build: 
    context: ./prometheus
    args:
      HTTP_PROXY: http://192.168.2.9:7890
      HTTPS_PROXY: http://192.168.2.9:7890    
YAML

# The Grafana dashboarding server.
grafana:
  image: grafana/grafana
  volumes:
    - "./grafana/definitions:/var/lib/grafana/dashboards"
    - "./grafana/provisioning:/etc/grafana/provisioning"
  ports:
    - "3000:3000"
  environment:
    - GF_FEATURE_TOGGLES_ENABLE=tempoSearch,tempoServiceGraph
YAML

prometheus:
  build: 
    context: ./prometheus
    args:
      HTTP_PROXY: http://192.168.2.9:7890
      HTTPS_PROXY: http://192.168.2.9:7890    
  ports:
    - "9090:9090"
YAML

global:
  scrape_interval:     15s # By default, scrape targets every 15 seconds.
remote_read:
scrape_configs:
  # Scrape Prometheus' own metrics.
  - job_name: 'prometheus'
    static_configs:
      - targets: ['localhost:9090']
        labels:
          group: 'prometheus'
  # Scrape from the Mythical Server service.
  - job_name: 'mythical-server'
    scrape_interval: 2s
    static_configs:
      - targets: ['mythical-server:4000']
        labels:
          group: 'mythical'
  # Scrape from the Mythical Requester service.
  - job_name: 'mythical-requester'
    scrape_interval: 2s
    static_configs:
      - targets: ['mythical-requester:4001']
        labels:
          group: 'mythical'
  # Scrape from the Node exporter, giving us resource usage.
  - job_name: 'node'
    scrape_interval: 5s
    static_configs:
      - targets: ['nodeexporter:9100']
        labels:
          group: 'resources'
  # Scrape from Grafana Agent, giving us metrics from traces it collects.
  - job_name: 'span-metrics'
    scrape_interval: 2s
    static_configs:
      - targets: ['agent:12348']
        labels:
          group: 'mythical'
  # Scrape from Grafana Agent, giving us metrics from traces it collects.
  - job_name: 'agent-metrics'
    scrape_interval: 2s
    static_configs:
      - targets: ['agent:12345']
        labels:
          group: 'mythical'
YAML

loki:
  image: grafana/loki
  ports:
    - "3100:3100"
YAML

tempo:
  image: grafana/tempo:1.2.1
  ports:
    - "3200:3200"
    - "4317:4317"
    - "55680:55680"
    - "55681:55681"
    - "14250:14250"
  command: ["-config.file=/etc/tempo.yaml" ]
  volumes:
    - ./tempo/tempo.yaml:/etc/tempo.yaml
YAML

server:
  http_listen_port: 3200
distributor:
  receivers:                           # 此配置将监听 tempo 能够监听的所有端口和协议。
    jaeger:                            # 更多的配置信息可以从 OpenTelemetry 收集器中获得
      protocols:                       # 在这里：https://github.com/open-telemetry/opentelemetry-collector/tree/main/receiver
        thrift_http:                   #
        grpc:                          # 对于生产部署来说，你应该只启用你需要的接收器！
        thrift_binary:
        thrift_compact:
    otlp:
      protocols:
        http:
        grpc:
ingester:
  trace_idle_period: 10s               # 在一个追踪没有收到跨度后，认为它已经完成并将其冲走的时间长度。
  max_block_bytes: 1_000_000           # 当它达到这个尺寸时，切掉头块或。..
  max_block_duration: 5m               #   这么长时间
compactor:
  compaction:
    compaction_window: 1h              # 在这个时间窗口中的块将被压缩在一起
    max_block_bytes: 100_000_000       # 压实块的最大尺寸
    block_retention: 1h
    compacted_block_retention: 10m
storage:
  trace:
    backend: local                     # 使用的后端配置
    block:
      bloom_filter_false_positive: .05 # 较低的值会产生较大的过滤器，但会产生较少的假阳性结果。
      index_downsample_bytes: 1000     # 每条索引记录的字节数
      encoding: zstd                   # 块编码 / 压缩。 选项：none, gzip, lz4-64k, lz4-256k, lz4-1M, lz4, snappy, zstd
    wal:
      path: /tmp/tempo/wal             # 在本地存储 wal 的地方
      encoding: none                   # wal 编码 / 压缩。 选项：none, gzip, lz4-64k, lz4-256k, lz4-1M, lz4, snappy, zstd
    local:
      path: /tmp/tempo/blocks
    pool:
      max_workers: 100                 # worker 池决定了对对象存储后台的并行请求的数量
      queue_depth: 10000
search_enabled: true
YAML

agent:
  image: grafana/agent:v0.24.0
  ports:
    - "12347:12345"
    - "12348:12348"
    - "6832:6832"
    - "55679:55679"
  volumes:
    - "${PWD}/agent/config.yaml:/etc/agent/agent.yaml"
  command: [
    "-config.file=/etc/agent/agent.yaml",
    "-server.http.address=0.0.0.0:12345",
  ]
YAML

server:
  log_level: debug
# 配置一个日志摄取端点，用于自动记录功能。
logs:
    configs:
    - name: loki
      clients:
        - url: http://loki:3100/loki/api/v1/push
          external_labels:
            job: agent
    positions_directory: /tmp/positions
# 配置一个 Tempo 实例来接收来自微服务的追踪。
traces:
  configs:
  - name: latencyEndpoint
    # 在 6832 端口接收 Jaeger 格式的追踪信息。
    receivers:
      jaeger:
        protocols:
          thrift_binary:
            endpoint: "0.0.0.0:6832"
    # 向 Tempo 实例发送成批的跟踪数据。
    remote_write:
      - endpoint: tempo:55680
        insecure: true
    # 从传入的跟踪跨度生成普罗米修斯指标。
    spanmetrics:
      # 添加 http.target 和 http.method span 标签作为度量数据的标签。
      dimensions:
        - name: http.method
        - name: http.target
      # 在 12348 端口暴露这些指标。
      handler_endpoint: 0.0.0.0:12348
    # 从传入的跟踪数据中自动生成日志。
    automatic_logging:
      # 使用在配置文件开始时定义的日志实例。
      backend: logs_instance
      logs_instance_name: loki
      # 每个根跨度记录一行（即每个跟踪记录一行）。
      roots: true
      processes: false
      spans: false
      # 在日志行中添加 http.method、http.target 和 http.status_code span 标签。如果有的话。
      span_attributes:
        - http.method
        - http.target
        - http.status_code
      # 强制将跟踪 ID 设置为 `traceId`。
      overrides:
        trace_id_key: "traceId"
    # 启用服务图。
    service_graphs:
      enabled: true
YAML