使用 search_after 进行分页

适用于不断下一页这样的查询
我这里用的releaseTimestamp倒序。

{
  "query": {
    "range": {
      "releaseTime": {
        "gte": "2019-09-20 00:00:00",
        "lte": "2019-09-21 00:00:00"
      }
    }
  },
  "search_after":[ 1568995200000],
  "_source": ["title","releaseTime"],
  "sort": [
    {
      "releaseTimestamp": {
        "order": "desc"
      }
    }
  ],
  "size": 5
}

得到3条结果：

 "hits": {
    "total": 3,
    "max_score": null,
    "hits": [
      {
        "_index": "test_news",
        "_type": "_doc",
        "_id": "45a88261c9975aa6e0eb996c369c32de",
        "_score": null,
        "_source": {
          "releaseTime": "2019-09-20 23:59:48",
          "title": "岭南大学与五邑大学将开展居家养老联合研究-新华网"
        },
        "sort": [
          1568995190935
        ]
      },
      {
        "_index": "test_news",
        "_type": "_doc",
        "_id": "3ba4cc4d5ea82ff6da48109c070cb265",
        "_score": null,
        "_source": {
          "releaseTime": "2019-09-20 23:59:48",
          "title": "岭南大学与五邑大学将开展居家养老联合研究"
        },
        "sort": [
          1568995188633
        ]
      },
      {
        "_index": "test_news",
        "_type": "_doc",
        "_id": "45a88261c9975aa6e0eb996c369c32de",
        "_score": null,
        "_source": {
          "releaseTime": "2019-09-20 23:59:48",
          "title": "岭南大学与五邑大学将开展居家养老联合研究-新华网"
        },
        "sort": [
          1568995185201
        ]
      }
    ]
  }

接下来用第二条数据的sort值 放到search_after中再次查询：

{
  "query": {
    "range": {
      "releaseTime": {
        "gte": "2019-09-20 00:00:00",
        "lte": "2019-09-21 00:00:00"
      }
    }
  },
  "search_after":[ 1568995188633],
  "_source": ["title","releaseTime"],
  "sort": [
    {
      "releaseTimestamp": {
        "order": "desc"
      }
    }
  ],
  "size": 5
}

结果会得到上次查询的第二条之后的数据：

 "hits": {
    "total": 3,
    "max_score": null,
    "hits": [
      {
        "_index": "test_news",
        "_type": "_doc",
        "_id": "45a88261c9975aa6e0eb996c369c32de",
        "_score": null,
        "_source": {
          "releaseTime": "2019-09-20 23:59:48",
          "title": "岭南大学与五邑大学将开展居家养老联合研究-新华网"
        },
        "sort": [
          1568995185201
        ]
      }
    ]
  }

如果是多个字段排序的话，search_after值的顺序要与我们排序条件的顺序一致
还有使用search_after时 from设置为0，-1或者直接不加from
使用search_after 进行分页 相比from&size的方式要更加高效，而且在不断有新数据入库的时候仅仅使用from和size分页会有重复的情况
相比使用scroll分页，search_after可以进行实时的查询
不过search_after不适合跳跃式的分页