Solr MaxGram Changes Not Taking Effect (Managed-Schema Mode)

Solution

Issue / Symptom

Search results appear limited to matching only the first ~18 characters of a filename (or similar text), even after setting maxGramSize to a higher value (for example, 50) in schema.xml and rebuilding the Solr index.

Observed Symptom Text (Verbatim)

"it only sees the first 18 characters. It doesn't care about anything else after that."
"It's always 18 characters."

Root Cause

Solr is running in managed-schema mode by default. In this mode, edits to schema.xml are not applied. Solr reads field type and analyzer definitions from a dynamically generated managed schema (for example, managed-schema.xml). As a result, the active NGramTokenizerFactory configuration (including maxGramSize) remains unchanged when only schema.xml is edited.

Resolution (Update MaxGram via Solr Schema API)

Use the Schema API to update the field type that controls the analyzer (example shown for field type text_en_ext), then reindex.

Step 1: Verify Current Field Type Settings

Replace placeholders for your Solr host/port and core/collection as needed.

curl "http://127.0.0.1:<solr_port>/solr/<core_or_collection>/schema/fieldtypes/text_en_ext"

Step 2: Backup Existing Documents (as used during troubleshooting)

curl "http://127.0.0.1:<solr_port>/solr/<core_or_collection>/select?q=*:*"

Step 3: Update maxGramSize Using the Schema API

Example that replaces the field type definition and sets maxGramSize to 50:

curl -X POST -H 'Content-type:application/json' \
"http://127.0.0.1:<solr_port>/solr/<core_or_collection>/schema" \
--data-binary '{
  "replace-field-type": {
    "name": "text_en_ext",
    "class": "solr.TextField",
    "positionIncrementGap": "100",
    "indexAnalyzer": {
      "charFilters": [{"class": "solr.ICUNormalizer2CharFilterFactory"}],
      "tokenizer": {"class": "solr.NGramTokenizerFactory", "minGramSize": "3", "maxGramSize": "50"},
      "filters": [
        {"class": "solr.ASCIIFoldingFilterFactory"},
        {"class": "solr.LowerCaseFilterFactory"}
      ]
    },
    "queryAnalyzer": {
      "tokenizer": {"class": "solr.KeywordTokenizerFactory"},
      "filters": [{"class": "solr.LowerCaseFilterFactory"}]
    }
  }
}'

Step 4: Verify the Change

curl "http://127.0.0.1:<solr_port>/solr/<core_or_collection>/schema/fieldtypes/text_en_ext"

Confirm the response shows "maxGramSize": "50" (or your intended value).

Step 5: Rebuild / Reindex Search Data

After changing the schema/analyzer, rebuild/reindex the existing data so the new analyzer settings are applied to indexed content. (Without reindexing, existing indexed terms may still reflect the prior analyzer behavior.)

Validation

1. Repeat the same search scenario that previously failed (searching by a string longer than 18 characters).
2. Confirm that adding characters beyond the first 18 now changes the results appropriately (that is, it no longer returns matches based only on the initial 18-character prefix).
3. Re-confirm the field type configuration via the schema endpoint to ensure maxGramSize remains at the configured value.

Engineering / Development Notes

The behavior was reproduced in a test environment and escalated for development review. Engineering analysis confirmed the root cause was managed-schema behavior (direct schema.xml edits not being loaded), and the associated defect record was later cancelled after the engineering analysis/work completed.