How to fix QueryShardException: No mapping found for [field] in order to sort on in Elasticsearch

Check if the field you're trying to sort on actually exists in your index mapping:

GET /your-index/_mapping

Look for the field name in the response. If it's not present, the field hasn't been mapped yet. If you're sorting across multiple indices, check each one:

GET /index1,index2/_mapping/field/your-field-name

This will show you whether the field exists in each index and how it's mapped.

If the field doesn't exist in some indices but you still want the query to work, use the unmapped_type parameter to tell Elasticsearch how to treat unmapped fields:

GET /your-index/_search
{
  "query": {
    "match_all": {}
  },
  "sort": [
    {
      "your-field": {
        "order": "desc",
        "unmapped_type": "long"
      }
    }
  ]
}

For keyword fields, use "unmapped_type": "keyword". For dates, use "unmapped_type": "date". This tells Elasticsearch to treat documents in indices without this field as if they have no value for it, rather than failing the query.

If the field should exist but doesn't, add it to your index mapping. For a new index:

PUT /your-index
{
  "mappings": {
    "properties": {
      "your-field": {
        "type": "keyword"
      }
    }
  }
}

For existing indices, you can add new fields to the mapping:

PUT /your-index/_mapping
{
  "properties": {
    "your-field": {
      "type": "keyword"
    }
  }
}

Choose the appropriate type: keyword for exact-value sorting, long/integer for numbers, date for timestamps. Never use text for fields you need to sort on.

If you're trying to sort on a text field, you cannot sort on the analyzed text directly. Most text fields have a .keyword subfield created automatically for sorting and aggregations:

GET /your-index/_search
{
  "query": {
    "match_all": {}
  },
  "sort": [
    {
      "your-field.keyword": {
        "order": "asc"
      }
    }
  ]
}

Check your mapping to confirm the .keyword subfield exists. If it doesn't, you can add it:

PUT /your-index/_mapping
{
  "properties": {
    "your-field": {
      "type": "text",
      "fields": {
        "keyword": {
          "type": "keyword",
          "ignore_above": 256
        }
      }
    }
  }
}

If you've corrected the mapping but documents were already indexed with incorrect types, you need to reindex:

POST /_reindex
{
  "source": {
    "index": "your-old-index"
  },
  "dest": {
    "index": "your-new-index"
  }
}

Then use an index alias to switch from the old to new index without downtime:

POST /_aliases
{
  "actions": [
    { "remove": { "index": "your-old-index", "alias": "your-alias" } },
    { "add": { "index": "your-new-index", "alias": "your-alias" } }
  ]
}

This ensures all documents are properly indexed according to the corrected mapping.