I am using OpenSearch version 7.10.2 and encountering an issue with the match_phrase_prefix query on certain asset numbers. My asset_number field is analyzed using a lowercase analyzer. Here is the detailed setup:

Index Settings:

{
  "settings": {
    "analysis": {
      "normalizer": {
        "lowercase_normalizer": {
          "type": "custom",
          "filter": ["lowercase"]
        }
      }
    }
  },
  "mappings": {
    "_routing": {
      "required": true
    },
    "properties": {
      "asset_number": {
        "type": "text",
        "fields": {
          "keyword": {
            "type": "keyword",
            "normalizer": "lowercase_normalizer"
          }
        }
      }
    }
  }
}

Problem Description
I have an index with asset_number values and am trying to run a match_phrase_prefix query to fetch assets starting with specific prefixes. The query works for some prefixes but not for others. Here are the results

• Working: PB1234, PE1234, PG1234, PI1234, PJ1234, PK1234, PL1234, PM1234, PN1234, AH1234, BH1234
• Not working: PC1234, PD1234, PF1234, PH1234, AC1234, BC1234, AD1234, BD1234, CA1234, CB1234, CC1234, CD1234, CE1234, DA1234, DB1234, DC1234, FA1234

The query is supposed to match asset numbers that start with specific prefixes like “PC”, “PD”, etc., but it fails for many of them.

Query Used

{
  "query": {
    "bool": {
      "must": [
        {
          "match_phrase_prefix": {
            "asset_number": "pc"
          }
        }
      ]
    }
  }
}

Any suggestions on what might be causing the query to fail for specific prefixes or how to fix this issue?

Additional Information:

• OpenSearch version: 7.10.2
• Lucene version: 9.4.2
• Query works for some prefixes but not others even though they are in the same index.