4. 部署频率采集：通过CI/CD Pipeline事件采集部署时间戳

部署频率（Deployment Frequency）是DORA四大关键指标之一，衡量团队向生产环境交付软件的频次。高部署频率通常意味着更短的交付周期、更快的反馈循环和更高的组织效能。本章聚焦于如何通过CI/CD Pipeline事件，使用OpenTelemetry SDK精确采集每次部署的时间戳，从而计算部署频率。

4.1 部署频率指标定义与采集策略

部署频率通常以“每日部署次数”或“每周部署次数”为单位。采集策略需明确以下要素：

4.2 基于GitHub Actions的部署事件采集

GitHub Actions通过 workflow_run 或 deployment 事件提供部署完成回调。推荐在部署Job的最后一步嵌入OTel SDK发送事件。

GitHub Actions Workflow 示例（部署Job末尾）：

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Deploy to production
        run: |
          # 实际的部署命令
          ./deploy.sh

      - name: Send deployment event via OTel SDK
        run: |
          # 使用Node.js OTel SDK发送事件
          node send-deployment-event.js
        env:
          OTEL_EXPORTER_OTLP_ENDPOINT: "https://otel-collector.example.com:4318"
          DEPLOYMENT_ENVIRONMENT: "production"
          SERVICE_NAME: "my-service"
          SERVICE_VERSION: ${{ github.sha }}

send-deployment-event.js 核心代码：

const { trace, context, SpanStatusCode } = require('@opentelemetry/api');
const { OTLPTraceExporter } = require('@opentelemetry/exporter-trace-otlp-http');
const { BasicTracerProvider, SimpleSpanProcessor } = require('@opentelemetry/sdk-trace-base');

// 初始化TracerProvider
const provider = new BasicTracerProvider();
const exporter = new OTLPTraceExporter({
  url: process.env.OTEL_EXPORTER_OTLP_ENDPOINT + '/v1/traces',
});
provider.addSpanProcessor(new SimpleSpanProcessor(exporter));
provider.register();

const tracer = trace.getTracer('deployment-tracker');

// 创建部署事件Span
const span = tracer.startSpan('deployment.completed', {
  attributes: {
    'deployment.environment': process.env.DEPLOYMENT_ENVIRONMENT,
    'service.name': process.env.SERVICE_NAME,
    'service.version': process.env.SERVICE_VERSION,
    'deployment.status': 'success',
    'deployment.timestamp': Date.now(),
    'ci.pipeline.id': process.env.GITHUB_RUN_ID,
    'ci.pipeline.run_number': process.env.GITHUB_RUN_NUMBER,
  },
});
span.setStatus({ code: SpanStatusCode.OK });
span.end();

// 强制导出
provider.forceFlush().then(() => {
  console.log('Deployment event sent successfully');
});

4.3 基于Jenkins的部署事件采集

Jenkins Pipeline通过Post阶段或Groovy脚本在部署完成后调用OTel SDK。推荐使用Jenkins的 withEnv 和 sh 步骤执行事件发送。

Jenkins Pipeline 示例（Declarative Pipeline）：

pipeline {
    agent any
    environment {
        OTEL_EXPORTER_OTLP_ENDPOINT = 'https://otel-collector.example.com:4318'
        SERVICE_NAME = 'my-service'
        DEPLOYMENT_ENVIRONMENT = 'production'
    }
    stages {
        stage('Build') {
            steps {
                sh 'make build'
            }
        }
        stage('Deploy') {
            steps {
                sh './deploy.sh'
            }
        }
    }
    post {
        always {
            script {
                def serviceVersion = sh(script: 'git rev-parse HEAD', returnStdout: true).trim()
                def deploymentStatus = currentBuild.result == 'SUCCESS' ? 'success' : 'failure'
                
                sh """
                    python3 send_deployment_event.py \
                        --service-name ${SERVICE_NAME} \
                        --service-version ${serviceVersion} \
                        --environment ${DEPLOYMENT_ENVIRONMENT} \
                        --status ${deploymentStatus} \
                        --pipeline-id ${BUILD_NUMBER}
                """
            }
        }
    }
}

send_deployment_event.py 核心代码：

import argparse
from opentelemetry import trace
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import SimpleSpanProcessor
from opentelemetry.sdk.resources import Resource

def send_deployment_event():
    parser = argparse.ArgumentParser()
    parser.add_argument('--service-name', required=True)
    parser.add_argument('--service-version', required=True)
    parser.add_argument('--environment', required=True)
    parser.add_argument('--status', required=True)
    parser.add_argument('--pipeline-id', required=True)
    args = parser.parse_args()

    # 配置资源
    resource = Resource.create({
        'service.name': args.service_name,
        'service.version': args.service_version,
    })

    # 初始化TracerProvider
    provider = TracerProvider(resource=resource)
    exporter = OTLPSpanExporter(
        endpoint=f"{os.environ['OTEL_EXPORTER_OTLP_ENDPOINT']}/v1/traces"
    )
    provider.addSpanProcessor(SimpleSpanProcessor(exporter))
    trace.set_tracer_provider(provider)

    tracer = trace.get_tracer(__name__)

    # 创建部署事件Span
    with tracer.start_as_current_span('deployment.completed') as span:
        span.set_attribute('deployment.environment', args.environment)
        span.set_attribute('deployment.status', args.status)
        span.set_attribute('deployment.timestamp', int(time.time() * 1000))
        span.set_attribute('ci.pipeline.id', args.pipeline_id)
        span.set_attribute('ci.pipeline.type', 'jenkins')

        if args.status == 'success':
            span.set_status(trace.Status(trace.StatusCode.OK))
        else:
            span.set_status(trace.Status(trace.StatusCode.ERROR))

    # 强制导出
    provider.force_flush()

if __name__ == '__main__':
    import os
    import time
    send_deployment_event()

4.4 事件数据结构与属性规范

为确保不同CI/CD系统采集的部署事件能够统一聚合，建议遵循以下属性规范：

4.5 部署频率计算与可视化

采集到部署事件后，可通过以下方式计算部署频率：

• 时间窗口聚合：按天、周、月对 deployment.completed Span进行计数
• 过滤条件：仅统计 deployment.environment = "production" 且 deployment.status = "success" 的事件
• 计算公式：部署频率 = 成功部署次数 / 时间窗口天数

4.6 常见问题与避坑指南