elasticsearch 6.5 : reindex random sample of documents from remote 2. 4 簇导致 illegal_argument_exception

Question

我正在尝试将远程 elasticsearch 2.4 集群中的随机文档样本重新索引到新的 elasticsearch 6.5 集群中。我正在尝试将此示例 ( https://www.elastic.co/guide/en/elasticsearch/reference/6.5/docs-reindex.html#_extracting_a_random_subset_of_an_index) 改编为远程重建索引:

例子:

POST _reindex
{
  "size": 10,
  "source": {
    "index": "twitter",
    "query": {
      "function_score" : {
        "query" : { "match_all": {} },
        "random_score" : {}
      }
    },
    "sort": "_score"    
  },
  "dest": {
    "index": "random_twitter"
  }
}

适配远程重建索引:

curl -H "Content-Type: application/json" -X POST localhost:9200/_reindex -d '
{
  "size": 100,
  "source": {
    "remote": {
      "host": "http://REMOTE_HOST:9200"
    },
    "index": "myindex",
    "query": {
      "function_score" : {
        "query" : { "match_all": {} },
        "random_score" : {}
      }
    },
    "sort": "_score"
  },
  "dest": {
    "index": "myindex"
  }
}'

这将返回状态代码 400 和以下消息:

{"error":{"root_cause":[{"type":"illegal_argument_exception","reason":"Unsupported sort [{\n  \"_score\" : {\n    \"order\" : \"desc\"\n  }\n}]"}],"type":"illegal_argument_exception","reason":"Unsupported sort [{\n  \"_score\" : {\n    \"order\" : \"desc\"\n  }\n}]"},"status":400}

所以按_score排序好像不行。当我删除 _score 排序时，它工作正常 - 但是，样本并不是真正随机的，如文档中所述。

从远程重建索引时这是不可能的，还是我做错了什么？ ;)

Answer 1

我会尝试指定这样的排序:

"sort": {"_score": "desc" }