Elasticsearch 基于scoll+bulk+索引别名实现零停机重建索引

Elasticsearch实战

一个field的设置是不能被修改的，如果要修改一个Field，那么应该重新按照新的mapping，建立一个index，然后将数据批量查询出来，重新用bulk api写入index中

批量查询的时候，建议采用scroll api，并且采用多线程并发的方式来reindex数据，每次scoll就查询指定日期的一段数据，交给一个线程即可

step1

首先我们建两个文档，然后看下mapping

PUT /vehicle/car/1
{
  "name":"benz",
  "create": "2020-01-01"
}

PUT /vehicle/car/2
{
  "name":"bmw",
  "create": "2019-01-01"
}


GET /vehicle/_mapping

{
  "vehicle" : {
    "mappings" : {
      "properties" : {
        "create" : {
          "type" : "date"
        },
        "name" : {
          "type" : "text",
          "fields" : {
            "keyword" : {
              "type" : "keyword",
              "ignore_above" : 256
            }
          }
        }
      }
    }
  }
}
 

step2

一开始，依靠dynamic mapping，插入数据，但是不小心有些数据是2020-01-01这种日期格式的，所以title这种field被自动映射为了date类型，实际上它应该是string类型的，当后期向索引中加入string类型的title值的时候，就会报错，如果此时想修改title的类型，是不可能的

PUT /vehicle/car/3
{
  "name":"Range Rover",
  "create":"china"
}

{
  "error" : {
    "root_cause" : [
      {
        "type" : "mapper_parsing_exception",
        "reason" : "failed to parse field [create] of type [date] in document with id '3'. Preview of field's value: 'china'"
      }
    ],
    "type" : "mapper_parsing_exception",
    "reason" : "failed to parse field [create] of type [date] in document with id '3'. Preview of field's value: 'china'",
    "caused_by" : {
      "type" : "illegal_argument_exception",
      "reason" : "failed to parse date field [china] with format [strict_date_optional_time||epoch_millis]",
      "caused_by" : {
        "type" : "date_time_parse_exception",
        "reason" : "Failed to parse with all enclosed parsers"
      }
    }
  },
  "status" : 400
}
 

PUT /vehicle/_mapping/car?include_type_name=true
{
  "properties":{
    "create":{
      "type":"text"
    }
  }
}

{
  "error" : {
    "root_cause" : [
      {
        "type" : "illegal_argument_exception",
        "reason" : "mapper [create] cannot be changed from type [date] to [text]"
      }
    ],
    "type" : "illegal_argument_exception",
    "reason" : "mapper [create] cannot be changed from type [date] to [text]"
  },
  "status" : 400
}

step3

此时，唯一的办法，就是进行reindex，也就是说，重新建立一个索引，将旧索引的数据查询出来，再导入新索引, 给java应用一个别名，这个别名是指向旧索引的，java应用先用着，java应用先用old_vehicle alias来操作，此时实际指向的是旧的vehicle

PUT /vehicle/_alias/alias_vehicle

{
  "acknowledged" : true
}

查询使用alias

GET /alias_vehicle/car/_search

#! Deprecation: [types removal] Specifying types in search requests is deprecated.
{
  "took" : 1,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 2,
      "relation" : "eq"
    },
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "vehicle",
        "_type" : "car",
        "_id" : "1",
        "_score" : 1.0,
        "_source" : {
          "name" : "benz",
          "create" : "2020-01-01"
        }
      },
      {
        "_index" : "vehicle",
        "_type" : "car",
        "_id" : "2",
        "_score" : 1.0,
        "_source" : {
          "name" : "bmw",
          "create" : "2019-01-01"
        }
      }
    ]
  }
}
 

step4

建立新的索引,并通过scrool分页查询原数据

PUT new_vehicle?include_type_name=true
{
  "mappings": {
    "car":{
      "properties":{
        "name":{
          "type":"text"
        },
        "create":{
          "type":"text"
        }
      }
    }
  }
}

#! Deprecation: [types removal] Using include_type_name in create index requests is deprecated. The parameter will be removed in the next major version.
{
  "acknowledged" : true,
  "shards_acknowledged" : true,
  "index" : "new_vehicle"
}

scroll深度分页查询

GET /vehicle/car/_search?scroll=1m
{
  "query": {
    "match_all": {}
  },
  "sort": [
    {
      "_doc": {
        "order": "asc"
      }
    }
  ], 
  "size": 1
}


GET _search/scroll
{
  "scroll":"1m",
  "scroll_id":"FGluY2x1ZGVfY29udGV4dF91dWlkDXF1ZXJ5QW5kRmV0Y2gBFks2Qi1fV0ptUkJtNlhETTVSU2piSlEAAAAAAAAE5BZSemFEbHd6SlRaQ3E0R29YbXgtcVNR"
}

step5

采用bulk api将scoll查出来的一批数据，批量写入新索引

POST /new_vehicle/car/_bulk
{"index":{"_id":1}}
{"name":"benz","create":"2020-01-01"}
{"index":{"_id":2}}
{"name":"bmw11","create":"2019-01-01"}

step6

将alias_vehicle alias切换到new_vehicle上去，java应用会直接通过index别名使用新的索引中的数据，java应用程序不需要停机，零提交，高可用

POST /_aliases
{
  "actions": [
    {
      "remove": {
        "index": "vehicle",
        "alias": "alias_vehicle"
      }
    },
    {
      "add": {
        "index": "new_vehicle",
        "alias": "alias_vehicle"
      }
    }
  ]
}

#! Deprecation: [types removal] Specifying types in search requests is deprecated.
{
  "took" : 568,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 2,
      "relation" : "eq"
    },
    "max_score" : 1.0,
    "hits" : [
      {
        "_index" : "new_vehicle",
        "_type" : "car",
        "_id" : "1",
        "_score" : 1.0,
        "_source" : {
          "name" : "benz",
          "create" : "2020-01-01"
        }
      },
      {
        "_index" : "new_vehicle",
        "_type" : "car",
        "_id" : "2",
        "_score" : 1.0,
        "_source" : {
          "name" : "bmw11",
          "create" : "2019-01-01"
        }
      }
    ]
  }
}
 

至此已经显示了索引的重建以及数据迁移

欢迎访问我的个人博客：小马博客

欢迎关注公众号《JAVA拾贝》