Replace document in Elasticsearch index with field condition

Question

If I have indexed a document in Elasticsearch that contains a datetime parameter, or some kind of sequence number, can I update/replace the entire document with a new version if, and only if, the value in my new document is greater than that in the currently indexed document?

Searching has shown me so far how I can affect the values of specific fields through scripting, but I'm not sure if I can use a script or operation as an update criterion, and replace the whole document if it's met.

To be more specific, we have a document object that contains a timestamp of when it was placed on the queue for processing, and since we may have multiple processors pulling things off the queue we would like to ensure that we only index documents newer than the one we already have in the index, discarding any old changes.

Answer 1

Try to use the _update_by_query Api.

Update By Query

Example:

Mappings

PUT my_index
{
  "mappings": {
    "properties": {
      "user": {
        "type": "keyword"
      },
      "timestamp": {
        "type": "keyword"
      }
    }
  }
}

Indexing documents

POST my_index/_doc/1
{
  "user":"user1",
  "timestamp":1234
}

POST my_index/_doc/2
{
  "user":"user2",
  "timestamp":1235
}

Update By Query

Let's update only documents with timestamp greater than 1234 .

POST /my_index/_update_by_query
{
  "script": {
    "source": "ctx._source.user='new user';", ----> updating field user
    "lang": "painless"
  },
  "query": {
    "range": {
      "timestamp": {
        "gt": 1234
      }
    }
  }
}

You can update other fields or insert new ones, just play with "source": "ctx._source.user='new user';ctx._source.timestamp=456";ctx._source.new_field=value"

Results

 {
    "_index": "my_index",
    "_type": "_doc",
    "_id": "2",
    "_score": 1,
    "_source": {
      "user": "new user",
      "timestamp": 1235
    }
  }

Hope this helps