Issue in Syncing the data between nodes

[baijugoradia (Baiju)]

We have 2 nodes node_a and node_b they both are on different servers where node_a is the main node here all the data needs to be where as node_b is a temp node where the data is being dump which needs to be synced to node_a
currently we are using reindex method to sync the nodes, but any doing this we have seen there is data loss in the node_a
For e.g.
Node_a has the following document

{
  "_index": "node_a",
  "_type": "fields",
  "_id": "AW9_qd-qSvQmY6fL_xPQ",
  "_score": 1,
  "_source": {
    "name": "BOB",
    "age": 25,
    "Location": "US",
    "other": "123"
  }
}

And node_b has

{
    "_index": "node_b",
    "_type": "fields",
    "_id": "AW9_qd-qSvQmY6fL_xPQ",
    "_score": 1,
    "_source": {
      "name": "BOB",
      "gender": "male",
    }
  }

After reindexing the expected output we desire is

{
    "_index": "node_a",
    "_type": "fields",
    "_id": "AW9_qd-qSvQmY6fL_xPQ",
    "_score": 1,
    "_source": {
      "name": "BOB",
      "age": 25,
      "Location": "US",
      "other": "123",
 "gender": "male",
    }
  }

Where only the missing data gets updates in the above case it's the gender, but instead the whole document is getting replaced.

Is there a way where I can achieve this by re-indexing method where only the missing data is updated int the document.

[tamara (Tamara Rosini)]

It is not possible with the reindex api, you can only know with the reindex api if there is a conflict between the docs (different versions), in order to return the conflicts you need to use "version_type": "external" property when you reindex, otherwise the version is not considered. https://www.elastic.co/guide/en/elasticsearch/reference/current/docs-reindex.html#docs-reindex-api-request

There is a github issue that discuss the "Allow reindex to do update/upsert operations" in here: https://github.com/elastic/elasticsearch/issues/17997