More Like This Queries should error when Routing Set to "required" and no routing is supplied

[kylelyk]

Elasticsearch version: 6.2.4

Plugins installed: []

JVM version: 1.8.0_101

OS version: MacOS (Darwin Kernel Version 15.6.0)

Description of the problem including expected versus actual behavior:
"More Like This" queries do not return any results when "mappings.type._routing" = true.

Steps to reproduce:

PUT test
{
  "mappings": {
    "type": {
      "_routing": {
        "required": true
      },
      "properties": {
        "myField": {
          "type": "text"
        }
      }
    }
  }
}

POST /_bulk
{ "index":  { "_index": "test", "_type": "type","_id":1, "_routing":1 }}
{"myField":"and_foo"}
{ "index":  { "_index": "test", "_type": "type","_id":2, "_routing":2 }}
{"myField":"and_foo" }

The following query returns no results:

GET /_search
{
  "query": {
    "more_like_this": {
      "fields": [
        "myField"
      ],
      "like": [
        {
          "_index": "test",
          "_type": "type",
          "_id": "1"
        }
      ],
      "min_term_freq": 1,
      "min_doc_freq": 1
    }
  }
}

If the _routing setting in the index mapping is removed, it correctly returns 1 result.

[elasticmachine]

Pinging @elastic/es-search-aggs

[colings86]

You should be able to add a routing field next to the _id field in the like array to specify the routing to use to retrieve the item.

[kylelyk]

Adding the routing field makes the query work as expected. Could an error be returned if you attempt to specify a document without a routing id if routing is set to required? Every other API I have used has given an appropriate error if I forget to set the routing id.

[colings86]

Thats a good suggestion @kylelyk, I'm going to reword the title of this issue to reflect the intention to change to that behaviour and mark it as an adoptme

[ashishguptak]

Hello, Can I take up this issue?
any pointers are much appreciated.

[cbismuth]

I'll have a look on this one.

[cbismuth]

I can reproduce this issue on master, see Gist here. As @kylelyk said, its returns no result.

{"took":5,"timed_out":false,"_shards":{"total":2,"successful":2,"skipped":0,"failed":0},"hits":{"total":0,"max_score":null,"hits":[]}}

Besides, Bulk API doesn't seem to allow _routing attribute any more, see error below. I've switched to two POST commands with refresh forced.

{"error":{"root_cause":[{"type":"illegal_argument_exception","reason":"Action/metadata line [2] contains an unknown parameter [_routing]"}],"type":"illegal_argument_exception","reason":"Action/metadata line [2] contains an unknown parameter [_routing]"},"status":400}

I'm now digging a little deeper in issue resolution and let you know.

EDIT

I've updated Gist test script to trigger a missing routing exception.
RoutingMissingException when searching a document by id with missing routing attribute is thrown from the server action layer, see here.

Unlike the GET API, a MoreLikeThisQuery query has to be parsed into search context to check whether or not the routing attribute field has been set (see here).

[cbismuth]

I've opened a pull request to fix this issue, see #33974.

[cbismuth]

Obsolete PR closed (due to bad Git rebase operation) and new one opened in sync with master branch.