如何基于CI Tests数据展示仅最后一次构建的测试结果饼图？

问题 #

我尝试可视化以JUnit XML格式上传的CI测试结果，现有ci_tests和ci_pipelines两个数据源，均包含status变量。需求是基于用户选择的时间窗口，展示仅最后一次构建的测试结果（通过/失败/跳过）分布的饼图，但当前查询显示的是多个构建的累计结果，即使同一天有多个构建运行。

以下是我尝试的JSON查询：

{
    "title": "Last Build Status",
    "requests": [
        {
            "response_format": "scalar",
            "queries": [
                {
                    "name": "query1",
                    "data_source": "ci_tests",
                    "search": {
                        "query": "test_level:test @test.name:* @test.suite:TEMI*"
                    },
                    "indexes": ["*"],
                    "group_by": [
                        {
                            "facet": "@test.status",
                            "limit": 1000,
                            "sort": {
                                "aggregation": "count",
                                "order": "desc",
                                "metric": "count"
                            },
                            "should_exclude_missing": true
                        }
                    ],
                    "compute": {
                        "aggregation": "cardinality",
                        "metric": "@test.name"
                    },
                    "storage": "hot"
                }
            ],
            "style": {
                "palette": "semantic"
            },
            "formulas": [
                { "formula": "query1" }
            ],
            "sort": {
                "count": 1000,
                "order_by": [
                    {
                        "type": "formula",
                        "index": 0,
                        "order": "desc"
                    }
                ]
            }
        }
    ],
    "type": "sunburst",
    "legend": { "type": "automatic" }
}

该可视化可运行但展示的是累计结果而非最新构建的结果，请问如何修改Datadog仪表盘查询（或添加过滤器），使饼图仅反映最近一次构建的测试结果而非多个构建的累计数据？

解决方案 #

要只展示最近一次构建的测试结果，需先从ci_pipelines数据源获取最新构建的标识（如@pipeline.id），再用该标识过滤ci_tests中的数据，具体修改如下：

修改后的完整JSON查询 #

{
    "title": "Last Build Status",
    "requests": [
        {
            "response_format": "scalar",
            "queries": [
                {
                    "name": "latest_pipeline",
                    "data_source": "ci_pipelines",
                    "search": {
                        "query": "status:*"
                    },
                    "indexes": ["*"],
                    "group_by": [
                        {
                            "facet": "@pipeline.id",
                            "limit": 1,
                            "sort": {
                                "aggregation": "max",
                                "order": "desc",
                                "metric": "@pipeline.timestamp"
                            }
                        }
                    ],
                    "compute": {
                        "aggregation": "max",
                        "metric": "@pipeline.timestamp"
                    },
                    "storage": "hot"
                },
                {
                    "name": "query1",
                    "data_source": "ci_tests",
                    "search": {
                        "query": "test_level:test @test.name:* @test.suite:TEMI* @pipeline.id:{{latest_pipeline.value}}"
                    },
                    "indexes": ["*"],
                    "group_by": [
                        {
                            "facet": "@test.status",
                            "limit": 1000,
                            "sort": {
                                "aggregation": "count",
                                "order": "desc",
                                "metric": "count"
                            },
                            "should_exclude_missing": true
                        }
                    ],
                    "compute": {
                        "aggregation": "cardinality",
                        "metric": "@test.name"
                    },
                    "storage": "hot"
                }
            ],
            "style": {
                "palette": "semantic"
            },
            "formulas": [
                { "formula": "query1" }
            ],
            "sort": {
                "count": 1000,
                "order_by": [
                    {
                        "type": "formula",
                        "index": 0,
                        "order": "desc"
                    }
                ]
            }
        }
    ],
    "type": "sunburst",
    "legend": { "type": "automatic" }
}

关键修改说明 #

• 新增latest_pipeline查询：从ci_pipelines中按@pipeline.timestamp降序排序，取第一个结果（最新构建）的@pipeline.id。
• 过滤测试数据：在ci_tests的查询中加入@pipeline.id:{{latest_pipeline.value}}，通过模板变量引用最新构建的ID，确保只统计该构建下的测试结果。
• 保留原有group_by和compute逻辑，维持测试结果的统计维度（通过/失败/跳过）和计算方式（去重测试用例数）不变。

内容的提问来源于stack exchange，提问作者goodman