Match Phrase Prefix Query

Question

I'm having an issue where elasticsearch is returning different results across my environments (staging and production).

The version of elasticsearch I'm using are the same for both environments.

version: { created: 1000199}

The two environments both have identical mapping and index settings.

I have an item indexed with a title field of "test". I am trying to perform a match_phrase_prefix query. Yet, in my staging environment when I just search for "te" the result comes back as expected. In production, I have to extend my search query to "tes" (3 letters) to retrieve the document.

The query I am issuing is a follows:

{
  "query": {
    "match_phrase_prefix": {
      "title": "test"
    }
  },
  "post_filter": {
    "term": {
      "uniqueId": "my_unique_id"
    }
  }
}

The only difference I can see between the two environments are the amount of documents index.

In my staging environment, I have approx. 150k documents index which equates to about 297M.

In production, I have over 120 million documents indexed which equates close to 120gb.

Is this a scale problem, or is there some elasticsearch setting that I am missing. Like I said the mappings and settings are identical across the environments, so I am a bit stumped here.

ChintanShah25 ChintanShah25 · Accepted Answer · 2015-11-17T02:58:52

I am pretty late, but see my other answer. The problem is max_expansions parameter which is 10 by default.

Your staging environment has less words which start with "te" than production environment and it is able to get test within 10 expansions(alphabetically)

Match Phrase Prefix Query

1 Answers