如何使用elasticsearch创建子查询

Question

我需要在 elasticsearch 上为以下条件创建查询。

当添加了 "rabbitmq.queue.name" 属性的最后一个元素等于 "service_test_error" 且 "rabbitmq.queue.messages.total.count" 的值不同于“0”时

下面的 sql 查询适用于我的搜索，但我无法使用 elasticsearch 执行相同的查询

select * from metric where rabbitmq.queue.messages.total.count != '0' and rabbitmq.queue.name = 'service_test_error' and timestamp = (select max(timestamp) from metric where rabbitmq.queue.name = 'service_test_error')

以下这些记录是我的 metric-xpto 索引中存在的示例

[
  {
    "_index": "metric-xpto",
    "_type": "_doc",
    "_id": "jYP1WnEBmYyEo7K68Zme",
    "_version": 1,
    "_score": null,
    "_source": {
      "@timestamp": "2020-04-08T18:03:14.899Z",
      "rabbitmq": {
        "queue": {
          "name": "service_test_error",
          "messages": {
            "total": {
              "count": 0
            }
          }
        }
      }
    }
  },
  {
    "_index": "metric-xpto",
    "_type": "_doc",
    "_id": "jYP1WnEBmYyEo7K68Zme",
    "_version": 1,
    "_score": null,
    "_source": {
      "@timestamp": "2020-04-07T18:03:14.899Z",
      "rabbitmq": {
        "queue": {
          "name": "service_test_error",
          "messages": {
            "total": {
              "count": 3
            }
          }
        }
      }
    }
  },
  {
    "_index": "metric-xpto",
    "_type": "_doc",
    "_id": "jYP1WnEBmYyEo7K68Zme",
    "_version": 1,
    "_score": null,
    "_source": {
      "@timestamp": "2020-04-03T17:03:14.899Z",
      "rabbitmq": {
        "queue": {
          "name": "service_alpha_test_error",
          "messages": {
            "total": {
              "count": 8
            }
          }
        }
      }
    }
  },
  {
    "_index": "metric-xpto",
    "_type": "_doc",
    "_id": "jYP1WnEBmYyEo7K68Zme",
    "_version": 1,
    "_score": null,
    "_source": {
      "@timestamp": "2020-04-03T18:03:14.899Z",
      "rabbitmq": {
        "queue": {
          "name": "service_test_error",
          "messages": {
            "total": {
              "count": 8
            }
          }
        }
      }
    }
  }
]

如何使用 elasticsearch 创建类似的查询？

Answer 1

在查询部分，您只能获取最上面的记录，然后必须在客户端检查消息数。

在聚合部分可以做到这一点。

{
  "size": 0,
  "query": {
    "term": {
      "rabbitmq.queue.name.keyword": {
        "value": "service_test_error"
      }
    }
  },
  "aggs": {
    "date": {
      "terms": {
        "field": "@timestamp",
        "size": 1,
        "order": {
          "_term": "desc"
        }
      },
      "aggs": {
        "message_count": {
          "terms": {
            "field": "rabbitmq.queue.messages.total.count",
            "size": 10
          },
          "aggs": {
            "filter_count": {
              "bucket_selector": {
                "buckets_path": {
                  "key": "_key"
                },
                "script": "if(params.key>0) return true; else return false;"
              }
            }
          }
        },
        "select_timestamp": {
          "bucket_selector": {
            "buckets_path": {
              "key": "message_count._bucket_count"
            },
            "script": "if(params.key>0) return true;else false;"
          }
        }
      }
    }
  }
}

结果：如果最上面的记录消息计数为零，则桶将为空，否则会有数据

 "hits" : {
    "total" : {
      "value" : 7,
      "relation" : "eq"
    },
    "max_score" : null,
    "hits" : [ ]
  },
  "aggregations" : {
    "date" : {
      "doc_count_error_upper_bound" : 0,
      "sum_other_doc_count" : 6,
      "buckets" : [ ]
    }
  }

Elastic search 没有连接，所以在查询中不可能将一个文档与另一个文档进行比较

如何使用elasticsearch创建子查询

How to create subquery with elasticsearch

database

elasticsearch

kibana

elasticsearch-dsl

elastic-stack