简介

  在业务运维场景中，需要对核心的API接口进行拨测。而各个接口需要传递的参数或者接口之间的依赖是比较复杂的，通常接口之间都是通过链式请求来完成一个业务场景。常见的就是先登录，拿到token以后，再进行后续的API请求。postman提供了基于GUI的方式完成这种场景适配，但是对于运维来讲，需要定时的基于策略的形式来对API进行监控。本篇文章就带你从0-1打造API监控体系。

知识储备

1. Postman使用方法

2. Docker基础知识

部署步骤

1. 从postman导出collection

以下文件以拨测httpbin.org为例,在Postman的GUI工具中导出拨测的json文件(httpbin.json)。示例中包含两个接口，一个模拟认证，一个模拟接口请求

2. 将导出的文件放到docker中运行

docker run -d -p 8080:8080 -v ./httpbin.json:/runner/collection.json kevinniu666/postman-prometheus:1.0.0

3. 获取拨测指标

curl 10.128.120.52:8080/metrics

# TYPE postman_lifetime_runs_total counter
postman_lifetime_runs_total{collection="httpbin"} 1

# TYPE postman_lifetime_iterations_total counter
postman_lifetime_iterations_total{collection="httpbin"} 1

# TYPE postman_lifetime_requests_total counter
postman_lifetime_requests_total{collection="httpbin"} 2

# TYPE postman_stats_iterations_total gauge
postman_stats_iterations_total{collection="httpbin"} 1

# TYPE postman_stats_iterations_failed gauge
postman_stats_iterations_failed{collection="httpbin"} 0

# TYPE postman_stats_requests_total gauge
postman_stats_requests_total{collection="httpbin"} 2

# TYPE postman_stats_requests_failed gauge
postman_stats_requests_failed{collection="httpbin"} 0

# TYPE postman_stats_tests_total gauge
postman_stats_tests_total{collection="httpbin"} 2

# TYPE postman_stats_tests_failed gauge
postman_stats_tests_failed{collection="httpbin"} 0

# TYPE postman_stats_test_scripts_total gauge
postman_stats_test_scripts_total{collection="httpbin"} 4

# TYPE postman_stats_test_scripts_failed gauge
postman_stats_test_scripts_failed{collection="httpbin"} 0

# TYPE postman_stats_assertions_total gauge
postman_stats_assertions_total{collection="httpbin"} 3

# TYPE postman_stats_assertions_failed gauge
postman_stats_assertions_failed{collection="httpbin"} 0

# TYPE postman_stats_transfered_bytes_total gauge
postman_stats_transfered_bytes_total{collection="httpbin"} 794

# TYPE postman_stats_resp_avg gauge
postman_stats_resp_avg{collection="httpbin"} 541

# TYPE postman_stats_resp_min gauge
postman_stats_resp_min{collection="httpbin"} 494

# TYPE postman_stats_resp_max gauge
postman_stats_resp_max{collection="httpbin"} 588

# TYPE postman_request_status_code gauge
postman_request_status_code{request_name="authentication",iteration="0",collection="httpbin"} 200

# TYPE postman_request_resp_time gauge
postman_request_resp_time{request_name="authentication",iteration="0",collection="httpbin"} 588

# TYPE postman_request_resp_size gauge
postman_request_resp_size{request_name="authentication",iteration="0",collection="httpbin"} 54

# TYPE postman_request_status_ok gauge
postman_request_status_ok{request_name="authentication",iteration="0",collection="httpbin"} 1

# TYPE postman_request_failed_assertions gauge
postman_request_failed_assertions{request_name="authentication",iteration="0",collection="httpbin"} 0

# TYPE postman_request_total_assertions gauge
postman_request_total_assertions{request_name="authentication",iteration="0",collection="httpbin"} 1

# TYPE postman_request_status_code gauge
postman_request_status_code{request_name="business-request",iteration="0",collection="httpbin"} 200

# TYPE postman_request_resp_time gauge
postman_request_resp_time{request_name="business-request",iteration="0",collection="httpbin"} 494

# TYPE postman_request_resp_size gauge
postman_request_resp_size{request_name="business-request",iteration="0",collection="httpbin"} 740

# TYPE postman_request_status_ok gauge
postman_request_status_ok{request_name="business-request",iteration="0",collection="httpbin"} 1

# TYPE postman_request_failed_assertions gauge
postman_request_failed_assertions{request_name="business-request",iteration="0",collection="httpbin"} 0

# TYPE postman_request_total_assertions gauge
postman_request_total_assertions{request_name="business-request",iteration="0",collection="httpbin"} 2

4. 将指标接入prometheus中，prometheus的配置文件中添加以下信息。

  - job_name: http-api-monitor
    scrape_interval: 15s
    scrape_timeout: 10s
    metrics_path: /metrics
    scheme: http
    static_configs:
    - targets:
      - 10.128.120.52:8080 #该ip为容器运行节点IP地址
      labels:
        usage: "httpbin接口拨测" 

5. 配置prometheus告警规则,详细的规则，大家可以根据prometheus的指标自己来设置。

  - alert: 接口返回码异常
    expr: postman_request_status_code  != 200 
    for: 1m
    labels:
      severity: error
    annotations:
      summary: "接口响应代码非200"
      description: "后端接口拨测失败"
  - alert: 接口返回内容判定失败
    expr: postman_request_failed_assertions  != 0
    for: 1m
    labels:
      severity: error
    annotations:
      summary: "接口返回内容判定失败"
      description: "后端接口postman测试未通过"	 

接下来的事情就是对接Alertmanager，并把告警发送给运维了。如果需要Grafana的图表，项目中有说明。源代码地址：

GitHub - kevinniu666/postman-prometheus: Run Postman collections continuously and export results as Prometheus metrics

写在后面

  整个实现方案的核心是这个容器，它将postman的运行转化成为了prometheus可识别的指标。这个容器的源代码已经在文档中提供了。如果你对nodejs有了解，可以自己去修改源码。该容器可以通过环境变量控制一些行为，列举如下：

 放一张产线的业务拨测图表：