asp.net core项目集成OpenTelemetry输出到OpenObserve进行可观测性处理

一、安装OpenObserve和Otel-collector

使用docker-compose 安装

记得替换下USER_MAIL xxx@xx.com和PASSWORD xxx
这里解释下为啥要用named volume来映射openobserve的data目录，因为目前版本(0.7.2)在windows的docker desktop里使用会有bug，导致stream无法正确写入文件持久化，一直在生成新的stream文件，而使用named volume则正常，详情请见issue

openobserve占用了5080（http）、5081（grpc）两个端口
otel-collector占用了5078和5079两个端口来作为对外提供服务的入口，后边的配置会有解释，其他端口有注释这里就不赘述了

新建一个openobserve目录，新建docker-compose.yaml文件，内容如下

version: '3'
services:
  main:
    image: openobserve/openobserve
    volumes:
      - ob-v:/data
    ports:
      - 5080:5080
      - 5081:5081
    environment:
      - ZO_DATA_DIR=/data
      - ZO_ROOT_USER_EMAIL=xxx@xx.com
      - ZO_ROOT_USER_PASSWORD=xxxx
      - TZ=Asia/Shanghai
    restart: always
  otel-collector:
    image: otel/opentelemetry-collector-contrib
    restart: always
    volumes:
      - ./otel-collector-config.yaml:/etc/otelcol-contrib/config.yaml
    ports:
      - 1888:1888 # pprof extension
      - 8888:8888 # Prometheus metrics exposed by the Collector
      - 8889:8889 # Prometheus exporter metrics
      - 13133:13133 # health_check extension
      - 5078:5078 # OTLP   gRPC receiver
      - 5079:5079 # OTLP/2 gRPC receiver
      #- 55679:55679 # zpages extension

volumes:
  ob-v: {}

otel-collector-config配置

然后在当前目录新建otel-collector-config.yaml内容如下
指定了两个端点，5079和5078，让测试环境和生产环境走不同的流水线处理。otel-collector-config配置参考文档
记得替换下边的myapp为自己的项目名称

receivers:
  otlp:
    protocols:
      grpc:
        endpoint: 0.0.0.0:5079
  otlp/2:
    protocols:
      grpc:
        endpoint: 0.0.0.0:5078

processors:
  batch:

exporters:
  otlp:
    endpoint: main:5081
    headers:
      Authorization: "Basic xxxxxx=="
      organization: myapp_test
      stream-name: default
    tls:
      insecure: true
  otlp/2:
    endpoint: main:5081
    headers:
      Authorization: "Basic xxxxxx=="
      organization: myapp_production
      stream-name: default
    tls:
      insecure: true

extensions:
  health_check:

service:
  extensions: [health_check]
  pipelines:
    traces:
      receivers: [otlp]
      processors: [batch]
      exporters: [otlp]
    metrics:
      receivers: [otlp]
      processors: [batch]
      exporters: [otlp]
    logs:
      receivers: [otlp]
      processors: [batch]
      exporters: [otlp]
    traces/2:
      receivers: [otlp/2]
      processors: [batch]
      exporters: [otlp/2]
    metrics/2:
      receivers: [otlp/2]
      processors: [batch]
      exporters: [otlp/2]
    logs/2:
      receivers: [otlp/2]
      processors: [batch]
      exporters: [otlp/2]

然后使用命令行运行docker-compose up -d

浏览器访问 服务器ip:5080, 用户名和密码是上边设置的邮箱和密码
点击左侧采集--Trace（OpenTelemetry）复制其中的Authorization: 后边的部分，替换otel-collector-config.yaml中的Basic xxxxxx==

最后执行docker-compose restart重启容器完成配置

二、asp.net core项目增加OpenTelemetry导出

引用OpenTelemetry Nuget包

在web api项目的csproj文件中增加以下nuget包引用

    <PackageReference Include="OpenTelemetry.Exporter.Console" Version="1.7.0" />
    <PackageReference Include="OpenTelemetry.Exporter.OpenTelemetryProtocol" Version="1.7.0" />
    <PackageReference Include="OpenTelemetry.Extensions.Hosting" Version="1.7.0" />
    <PackageReference Include="OpenTelemetry.Instrumentation.AspNetCore" Version="1.7.0" />
    <PackageReference Include="OpenTelemetry.Instrumentation.EntityFrameworkCore" Version="1.0.0-beta.8" />
    <PackageReference Include="OpenTelemetry.Instrumentation.Http" Version="1.7.0" />
    <PackageReference Include="OpenTelemetry.Instrumentation.Runtime" Version="1.5.1" />

在appsettings.json中配置启用开关

  "OpenTelemetry": {
    "ServiceName": "UserService",
    "EndPoint": "http://192.168.1.101:5079",
    "Enable": true
  }

其中ServiceName是该web api项目名称，EndPoint是otel-collector的一个grpc端点
比如测试环境的项目使用了5079端口，那生成环境就改为5078端口，如果还有演示环境、预发布环境之类的就继续在配置里增加端口

代码中配置启用Logging Metrics Traces

配置Logging导出

由于ConfigureLogging方法是IHostBuilder的扩展方法，所以只能在CreateDefaultBuilder这里调用

Host.CreateDefaultBuilder(args)
            .ConfigureLogging((c,l) =>
            {
                if (bool.TryParse(c.Configuration["OpenTelemetry:Enable"], out var e) && e)
                {
                    l.AddOpenTelemetry(logging =>
                    logging.SetResourceBuilder(ResourceBuilder.CreateDefault().AddService(c.Configuration["OpenTelemetry:ServiceName"]))
                        .AddOtlpExporter(o =>
                        {
                            o.Endpoint = new Uri(c.Configuration["OpenTelemetry:Endpoint"]);
                        }));
                }
            })

配置 Tracing和Metrics导出

在有IServiceCollection的地方执行AddOpenTelemetry扩展方法

if (bool.TryParse(_appConfiguration["OpenTelemetry:Enable"], out var otelEnable) && otelEnable)
            {
                services
                    .AddOpenTelemetry()
                    .ConfigureResource(builder => builder
                        .AddService(serviceName: _appConfiguration["OpenTelemetry:ServiceName"], serviceInstanceId: Environment.MachineName))
                    .WithTracing(builder => builder
                        .SetSampler(new AlwaysOnSampler())
                        .AddEntityFrameworkCoreInstrumentation(options => options.SetDbStatementForText = true)
                        .AddHttpClientInstrumentation()
                        .AddAspNetCoreInstrumentation(options =>
                        {
                            options.RecordException = true;
                        })
                        .AddOtlpExporter(o =>
                        {
                            o.Endpoint = new Uri(_appConfiguration["OpenTelemetry:Endpoint"]);
                        })
                    )

                    .WithMetrics(builder => builder
                        .AddRuntimeInstrumentation()
                        .AddAspNetCoreInstrumentation()
                        .AddOtlpExporter(o =>
                        {
                            o.Endpoint = new Uri(_appConfiguration["OpenTelemetry:Endpoint"]);
                        })
                    );
            }

最后项目运行起来在OpenObserve的数据流中有对应数据就算成功了