apache/druid · error · org.apache.druid.java.util.common.ISE

Value dictionaries already serialized for column

Error message

Value dictionaries already serialized for column [%s], cannot serialize again

What it means

serializeDictionaries writes the null sentinel plus string/long/double/arrays dictionaries for a variant column and can only run once per serializer instance. If dictionarySerialized is already true, calling it again would corrupt the column, so it throws ISE 'Value dictionaries already serialized ... cannot serialize again'.

Solutions

Ensure serializeDictionaries is called exactly once per serializer instance
Use a fresh VariantColumnSerializer if dictionaries must be rewritten
Make retry wrappers idempotent or recreate the serializer on retry
Track serialization state in the calling merge code

Example fix

// before
if (needsRetry) serializer.serializeDictionaries(...); // second call
// after
if (needsRetry) { serializer = createNewSerializer(name, medium); serializer.serializeDictionaries(...); }

Defensive patterns

Strategy: validation

Validate before calling

if (serializer.isDictionarySerialized()) { throw new IllegalStateException("dictionaries already written for " + name); }

Try / catch

try { serializer.serializeDictionaries(strings, longs, doubles, arrays); } catch (ISE e) { if (e.getMessage().startsWith("Value dictionaries already serialized")) { /* skip second call */ } else { throw e; } }

Prevention

Call serializeDictionaries exactly once per serializer
Make retry logic recreate the serializer instead of re-calling
Keep dictionary and value phases in one code path

When it happens

Trigger: Calling serializeDictionaries twice on the same VariantColumnSerializer, e.g. retry logic that re-invokes the method after a partial failure, or a merge path that processes dictionaries per-part and per-column.

Common situations: Custom segment merge tooling with retriable dictionary serialization; adding dictionary handling code to an existing merge pipeline without tracking state.

Understand the failure class

Background: "Invalid state transition" errors: "status must be X, actually Y", "already rejected/charging/uninstalled", "cannot ... while running" — what they mean when a library rejects your call — this error's family across 31 libraries.

Related errors

AI-assisted analysis of apache/druid@9b90983fd2 (2026-09-07). Data as JSON: /api/errors/02ad053cd18b8b6a. Report an issue: GitHub.

Appendix: source

Thrown at processing/src/main/java/org/apache/druid/segment/nested/VariantColumnSerializer.java:201

  public void open() throws IOException
  {
    if (!dictionarySerialized) {
      throw new IllegalStateException("Dictionary not serialized, cannot open value serializer");
    }
    intermediateValueWriter = new FixedIndexedIntWriter(segmentWriteOutMedium, false);
    intermediateValueWriter.open();
  }

  @Override
  public void serializeDictionaries(
      Iterable<String> strings,
      Iterable<Long> longs,
      Iterable<Double> doubles,
      Iterable<int[]> arrays
  ) throws IOException
  {
    if (dictionarySerialized) {
      throw new ISE("Value dictionaries already serialized for column [%s], cannot serialize again", name);
    }

    // null is always 0
    dictionaryWriter.write(null);
    for (String value : strings) {
      if (value == null) {
        continue;
      }

      dictionaryWriter.write(value);
    }

    for (Long value : longs) {
      if (value == null) {
        continue;
      }
      longDictionaryWriter.write(value);
    }

View on GitHub (pinned to 9b90983fd2)