如何用Python通过API获取GCP API的请求、错误等数据及嵌入图表？

一、用Python通过API提取监控指标 #

1. 前置准备 #

• 安装Google Cloud监控的Python SDK：

  pip install google-cloud-monitoring
  

• 完成身份认证：
  • 本地开发：执行gcloud auth application-default login，按提示完成登录
  • 服务器部署：将服务账号密钥文件放在服务器上，设置环境变量export GOOGLE_APPLICATION_CREDENTIALS="/path/to/your/service-account-key.json"

2. 获取API整体的Requests、Errors、Latency数据 #

以下代码可查询指定API在过去1小时的核心指标：

from google.cloud import monitoring_v3
from google.protobuf.timestamp_pb2 import Timestamp
import datetime

# 初始化客户端
client = monitoring_v3.MetricServiceClient()
# 替换为你的项目ID
PROJECT_ID = "your-project-id"
project_name = f"projects/{PROJECT_ID}"
# 替换为目标API名称，比如storage.googleapis.com
TARGET_API = "your-target-api"

# 设置查询时间范围：过去1小时
end_time = datetime.datetime.now(datetime.timezone.utc)
start_time = end_time - datetime.timedelta(hours=1)

# 转换为监控API要求的Timestamp格式
start_ts = Timestamp()
start_ts.FromDatetime(start_time)
end_ts = Timestamp()
end_ts.FromDatetime(end_time)

# 查询Requests总数
def fetch_requests():
    request_query = monitoring_v3.ListTimeSeriesRequest(
        name=project_name,
        filter=f'metric.type="serviceruntime.googleapis.com/api/request_count" AND resource.type="api" AND resource.label."api_service"="{TARGET_API}"',
        interval=monitoring_v3.TimeInterval(start_time=start_ts, end_time=end_ts),
        view=monitoring_v3.ListTimeSeriesRequest.TimeSeriesView.FULL
    )
    for series in client.list_time_series(request_query):
        print("=== Requests 数据 ===")
        for point in series.points:
            print(f"时间: {point.interval.start_time}, 数值: {point.value.int64_value}")

# 查询Errors总数
def fetch_errors():
    error_query = monitoring_v3.ListTimeSeriesRequest(
        name=project_name,
        filter=f'metric.type="serviceruntime.googleapis.com/api/error_count" AND resource.type="api" AND resource.label."api_service"="{TARGET_API}"',
        interval=monitoring_v3.TimeInterval(start_time=start_ts, end_time=end_ts),
        view=monitoring_v3.ListTimeSeriesRequest.TimeSeriesView.FULL
    )
    for series in client.list_time_series(error_query):
        print("\n=== Errors 数据 ===")
        for point in series.points:
            print(f"时间: {point.interval.start_time}, 数值: {point.value.int64_value}")

# 查询Latency分位数（p50/p95/p99）
def fetch_latency():
    latency_query = monitoring_v3.ListTimeSeriesRequest(
        name=project_name,
        filter=f'metric.type="serviceruntime.googleapis.com/api/latencies" AND resource.type="api" AND resource.label."api_service"="{TARGET_API}"',
        interval=monitoring_v3.TimeInterval(start_time=start_ts, end_time=end_ts),
        view=monitoring_v3.ListTimeSeriesRequest.TimeSeriesView.FULL
    )
    for series in client.list_time_series(latency_query):
        percentile = series.metric.labels["percentile"]
        print(f"\n=== Latency {percentile} 数据 ===")
        for point in series.points:
            print(f"时间: {point.interval.start_time}, 数值: {point.value.double_value} 秒")

# 执行所有查询
fetch_requests()
fetch_errors()
fetch_latency()

3. 获取单个API Method的指标数据 #

如果要查询特定方法（比如storage.objects.get）的指标，只需在过滤条件中添加metric.label."method"即可，示例代码：

def fetch_method_metrics(method_name):
    # 查询指定Method的Requests
    method_request_query = monitoring_v3.ListTimeSeriesRequest(
        name=project_name,
        filter=f'metric.type="serviceruntime.googleapis.com/api/request_count" AND resource.type="api" AND resource.label."api_service"="{TARGET_API}" AND metric.label."method"="{method_name}"',
        interval=monitoring_v3.TimeInterval(start_time=start_ts, end_time=end_ts),
        view=monitoring_v3.ListTimeSeriesRequest.TimeSeriesView.FULL
    )
    for series in client.list_time_series(method_request_query):
        print(f"\n=== Method {method_name} Requests ===")
        for point in series.points:
            print(f"时间: {point.interval.start_time}, 数值: {point.value.int64_value}")

# 调用示例，替换为目标Method名称
fetch_method_metrics("storage.objects.get")

如果要获取该API下所有Method的指标，去掉过滤条件中的metric.label."method"，遍历结果时通过series.metric.labels["method"]区分不同方法即可。

二、嵌入API监控图表到其他网站 #

Google Cloud Console的API详情页图表无法直接通过IFrame嵌入，原因是控制台有严格的身份验证和同源策略限制，浏览器会阻止跨域IFrame加载。

可行的替代方案：

• 用上面的Python代码获取原始指标数据，然后结合Chart.js、Plotly等前端图表库，在自己的网站上生成自定义监控图表
• 使用Google Cloud Monitoring的自定义仪表盘功能，创建包含目标API指标的仪表盘，然后通过仪表盘的分享功能生成链接（但用户仍需登录Google Cloud账号才能查看）

内容的提问来源于stack exchange，提问作者GeekGeek4