求助 es 多层聚合查询问题

V2EX = way to explore

V2EX 是一个关于分享和探索的地方

For Existing Member Sign In

› Elasticsearch 参考文档

› Elasticsearch: The Definitive Guide

› elasticseaerch-analysis-ik

This topic created in 2005 days ago, the information mentioned may be changed or developed.

数据结构如下所示

{
  "id": 1234,
  "operationName": "harbor-harbor-core.harbor.svc.cluster.local:80/service/*",
  "duration": 145428,
  "tags": [
    {
      "key": "guid:x-request-id",
      "type": "string",
      "value": "bccfdfa4-86e6-9f47-a41e-d56f7b4e5a88"
    },
    {
      "key": "http.url",
      "type": "string",
      "value": "http://baidu.com"
    }
  ]
}

我想根据 tags 中 key 为 http.url 这个字段，聚合出 url 的数量，并且算出每个 url 的平均访问时间，也就是算 avg duration 字段。现在我用下面的查询可以对 url 以及数量进行聚合，但是不知道怎么算 avg duration 。

{
  "size": 0,
  "aggs": {
    "tags": {
      "nested": {
        "path": "tags"
      },
      "aggs": {
        "key": {
          "filter": {
            "term": {
              "tags.key": "http.url"
            }
          },
          "aggs": {
            "values": {
              "terms": {
                "field": "tags.value"
              }
            }
          }
        }
      }
    }
  }
}

求助 es 大神，如何写查询能够计算出 avg duration

1 replies • 2020-11-11 17:17:48 +08:00

weofuh

Nov 11, 2020

你需要 reverse_nested

{
"size": 0,
"aggs": {
"tags": {
"nested": {
"path": "tags"
},
"aggs": {
"key": {
"filter": {
"term": {
"tags.key": "http.url"
}
},
"aggs": {
"values": {
"terms": {
"field": "tags.value"
},
"aggs": {
"url_duration_avg": {
"reverse_nested": {},
"aggs": {
"duration_avg":{
"avg": {
"field": "duration"
}
}
}
}
}
}
}
}
}
}
}
}