Can't do date range query in a small range

Elastic Stack Elasticsearch
1

my index mapping:

{
  "order": 0,
  "template": "test_index*",
  "settings": {
    "index": {
      "refresh_interval": "1s"
    }
  },
  "mappings": {
    "logs": {
      "properties": {
        "sell_date_start": {
          "format": "yyy-MM-dd HH:mm:ss||yyyy-MM-dd||epoch_millis",
          "type": "date"
        }
      },
      "_all": {
        "enabled": false
      }
    }
  },
  "aliases": {}
}

and my test data:

{"sell_date_start":"2016-07-16 10:00:00", "@timestamp":"2016-07-21T06:18:38.659Z"}
{"sell_date_start":"2016-07-17 20:00:00", "@timestamp":"2016-07-21T06:18:50.659Z"}
{"sell_date_start":"2016-07-17 10:00:00", "@timestamp":"2016-07-21T06:18:00.659Z"}
{"sell_date_start":"2016-07-16 15:00:00", "@timestamp":"2016-07-21T06:18:18.659Z"}
{"sell_date_start":"2016-07-16 20:00:00", "@timestamp":"2016-07-21T06:18:28.659Z"}
{"sell_date_start":"2016-07-16 11:00:00", "@timestamp":"2016-07-21T06:19:38.659Z"}

the following query doesn't get any result:

{
  "query": {
    "query_string": {
      "query": "sell_date_start:[1468634400000 TO 1468641600000]"
    }
  }
}
or
{
  "query": {
    "query_string": {
      "query": "sell_date_start:1468634400000"
    }
  }
}

But the following query can:

{
  "query": {
    "query_string": {
      "query": "sell_date_start:[1468634400000 TO 1468670400000]"
    }
  }
}
or
{
  "query": {
    "query_string": {
      "query": "sell_date_start:[1468634400000 TO *]"
    }
  }
}

es version: 2.3.3

Why ?????
Can anyone tell me why??

thanks

2

the doc date format says "In JSON documents, dates are represented as strings. Elasticsearch uses a set of preconfigured formats to recognize and parse these strings into a long value representing milliseconds-since-the-epoch in UTC."