AWS CloudWatch Agent指标错误：MetricDataQueries中Id参数值不唯一

云上老周

2026-8-9

解决CloudWatch GetMetricData接口的Id重复问题

问题场景

在Lambda中通过CloudWatch Agent调用get_metric_data接口获取指标数据，代码如下：

metrics_response = cw.get_metric_data(
        MetricDataQueries = metrics_data_queries,
        StartTime = datetime.now()+timedelta(minutes=-60),
        EndTime=datetime.now()
        )

对应的metrics_data_queries配置：

[{"Id": "cpu_time_idle", "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_idle", 
"Dimensions": [{"Name": "InstanceId", "Value": "i-XXXX"}, {"Name": "cpu", "Value": "cpu0"}]}, "Period": 6000, "Stat": "Average"}, 
"Label": "cpu_time_idleResponse", "ReturnData": True}, 

{"Id": "cpu_time_idle", "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_idle", 
"Dimensions": [{"Name": "InstanceId", "Value": "i-XXXX"}]}, 
"Period": 6000, "Stat": "Average"}, "Label": "cpu_time_idleResponse", "ReturnData": True}, 

{"Id": "cpu_time_active", "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_active", 
"Dimensions": [{"Name": "InstanceId", "Value": "XXXX"}, {"Name": "cpu", "Value": "cpu0"}]}, 
"Period": 6000, "Stat": "Average"}, "Label": "cpu_time_activeResponse", "ReturnData": True}, 

{"Id": "cpu_time_active", "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_active", 
"Dimensions": [{"Name": "InstanceId", "Value": "i-XXXX"}]}, 
"Period": 6000, "Stat": "Average"}, "Label": "cpu_time_activeResponse", "ReturnData": True}]

触发错误：

[ERROR] ClientError: An error occurred (ValidationError) when calling the GetMetricData operation: **The values for parameter id in MetricDataQueries are not unique.**

Traceback (most recent call last):
  File "/var/task/lambda_function.py", line 27, in lambda_handler
    metrics_response = get_metrics(metrics_list_response,cw)
  File "/var/task/lambda_function.py", line 81, in get_metrics
    metrics_response = cw.get_metric_data(
  File "/var/runtime/botocore/client.py", line 391, in _api_call
    return self._make_api_call(operation_name, kwargs)
  File "/var/runtime/botocore/client.py", line 719, in _make_api_call
    raise error_class(parsed_response, operation_name)

使用的CWAgent配置config.json：

{
        "agent": {
                "metrics_collection_interval": 60,
                "run_as_user": "root"
        },
        "metrics": {
                "aggregation_dimensions": [
                        [
                                "InstanceId"
                        ]
                ],
                "append_dimensions": {
                        "InstanceId": "${aws:InstanceId}"
                },
                "metrics_collected": {
                        "collectd": {
                                "metrics_aggregation_interval": 60
                        },
                        "disk": {
                                "measurement": [
                                        "total",
                                        "used",
                                        "free",
                                        "used_percent"                          
                                ],
                                "metrics_collection_interval": 60,
                                "resources": [
                                        "/"
                                ]
                        },
                        "cpu": {
                                "measurement": [
                                        "time_active",
                                        "time_idle",
                                        "usage_active",
                                        "usage_idle",
                                        "usage_user",
                                        "usage_system"
                                ],
                                "totalcpu": false,
                                "metrics_collection_interval": 60,
                                "resources": [
                                        "*"
                                ]
                        },
                        "mem": {
                                "measurement": [
                                        "total",
                                        "used",
                                        "active",
                                        "available",
                                        "available_percent",
                                        "free",
                                        "inactive",
                                        "used_percent",
                                        "cached"
                                ],
                                "metrics_collection_interval": 60
                        },
                        "processes": {
                                "measurement": [
                                        "total",
                                        "running",
                                        "sleeping",
                                        "blocked",
                                        "dead"                          
                                ],
                                "metrics_collection_interval": 60
                        }
                        
                }
        }
}

问题原因

CloudWatch的GetMetricData接口要求每个MetricDataQuery的Id必须唯一，用于标识每个查询的结果。当前配置中，查询单CPU核心（带cpu: cpu0维度）和实例汇总（仅InstanceId维度）的同名称指标时，使用了相同的Id，导致校验失败。

另外，CWAgent配置中cpu部分设置了totalcpu: false且resources: ["*"]，会收集每个CPU核心的独立指标；同时aggregation_dimensions配置了按InstanceId聚合，因此会生成两种维度的指标数据，查询时需要区分但未给Id做唯一标识。

解决方案

1. 为每个查询设置唯一Id

直接修改metrics_data_queries中的Id字段，结合指标名、维度信息生成唯一标识，示例如下：

[{"Id": "cpu_time_idle_cpu0", "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_idle", 
"Dimensions": [{"Name": "InstanceId", "Value": "i-XXXX"}, {"Name": "cpu", "Value": "cpu0"}]}, "Period": 6000, "Stat": "Average"}, 
"Label": "cpu_time_idle_cpu0_Response", "ReturnData": True}, 

{"Id": "cpu_time_idle_instance", "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_idle", 
"Dimensions": [{"Name": "InstanceId", "Value": "i-XXXX"}]}, 
"Period": 6000, "Stat": "Average"}, "Label": "cpu_time_idle_instance_Response", "ReturnData": True}, 

{"Id": "cpu_time_active_cpu0", "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_active", 
"Dimensions": [{"Name": "InstanceId", "Value": "XXXX"}, {"Name": "cpu", "Value": "cpu0"}]}, 
"Period": 6000, "Stat": "Average"}, "Label": "cpu_time_active_cpu0_Response", "ReturnData": True}, 

{"Id": "cpu_time_active_instance", "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_active", 
"Dimensions": [{"Name": "InstanceId", "Value": "i-XXXX"}]}, 
"Period": 6000, "Stat": "Average"}, "Label": "cpu_time_active_instance_Response", "ReturnData": True}]

2. 动态生成查询时确保Id唯一

如果通过代码动态生成metrics_data_queries，可以通过以下方式保证Id唯一：

为每个查询添加递增序号（如cpu_time_idle_1、cpu_time_idle_2）
拼接维度值到Id中（比如将cpu维度的cpu0拼接到Id末尾）
使用UUID生成唯一字符串作为Id

示例代码片段：

import uuid

# 动态生成查询项
queries = []
# 单CPU查询
single_cpu_query = {
    "Id": f"cpu_time_idle_{uuid.uuid4().hex[:8]}",
    "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_idle", "Dimensions": [{"Name": "InstanceId", "Value": "i-XXXX"}, {"Name": "cpu", "Value": "cpu0"}]}, "Period": 6000, "Stat": "Average"},
    "Label": "cpu_time_idle_cpu0_Response",
    "ReturnData": True
}
queries.append(single_cpu_query)
# 实例汇总查询
instance_query = {
    "Id": f"cpu_time_idle_{uuid.uuid4().hex[:8]}",
    "MetricStat": {"Metric": {"Namespace": "CWAgent", "MetricName": "cpu_time_idle", "Dimensions": [{"Name": "InstanceId", "Value": "i-XXXX"}]}, "Period": 6000, "Stat": "Average"},
    "Label": "cpu_time_idle_instance_Response",
    "ReturnData": True
}
queries.append(instance_query)

3. 按需精简查询（可选）

如果不需要同时获取单CPU和实例汇总的指标，可以根据业务需求只保留其中一种查询，既避免重复Id问题，又减少查询开销。

内容的提问来源于stack exchange，提问作者Santosh Kumar Sahoo