apache/iceberg · error · UnsupportedOperationException

Received invalid default split request event from subtask %d

Error message

Received invalid default split request event from subtask %d as Iceberg source uses custom split request event

What it means

AbstractIcebergEnumerator.handleSplitRequest overrides the default FLIP-27 enumerator hook. The Iceberg source does not use the framework's default split-request mechanism; readers instead send a custom SplitRequestEvent via handleSourceEvent so they can piggyback finished split ids. Receiving a default split request means a reader (or foreign event source) is not using the Iceberg source protocol, so the enumerator throws UnsupportedOperationException naming the offending subtask id.

Source

Thrown at flink/v2.2/flink/src/main/java/org/apache/iceberg/flink/source/enumerator/AbstractIcebergEnumerator.java:78

        // excessive memory footprint. Some pending splits may not have been discovered yet.
        .setUnassignedSplitsGauge(() -> Long.valueOf(assigner.pendingSplitCount()));
    this.enumeratorContext.metricGroup().gauge("pendingRecords", assigner::pendingRecords);
  }

  @Override
  public void start() {
    assigner.start();
  }

  @Override
  public void close() throws IOException {
    assigner.close();
  }

  @Override
  public void handleSplitRequest(int subtaskId, @Nullable String requesterHostname) {
    // Iceberg source uses custom split request event to piggyback finished split ids.
    throw new UnsupportedOperationException(
        String.format(
            Locale.ROOT,
            "Received invalid default split request event "
                + "from subtask %d as Iceberg source uses custom split request event",
            subtaskId));
  }

  @Override
  public void handleSourceEvent(int subtaskId, SourceEvent sourceEvent) {
    if (sourceEvent instanceof SplitRequestEvent) {
      SplitRequestEvent splitRequestEvent = (SplitRequestEvent) sourceEvent;
      LOG.info("Received request split event from subtask {}", subtaskId);
      assigner.onCompletedSplits(splitRequestEvent.finishedSplitIds());
      readersAwaitingSplit.put(subtaskId, splitRequestEvent.requesterHostname());
      assignSplits();
    } else {
      throw new IllegalArgumentException(
          String.format(

View on GitHub (pinned to 86d9c8fc54)

Solutions

1. Use the stock IcebergSourceReader (IcebergSourceReaderBase subclasses) which sends SplitRequestEvent, not the default request.
2. Remove custom reader code that calls SourceReaderContext.sendSplitRequest(); instead emit SplitRequestEvent with finished split ids.
3. Verify no non-Iceberg readers are connected to this enumerator (check subtask id from the message in JobManager logs).
4. Align reader and enumerator versions — an old reader jar with a new enumerator can fall back to default requests.

Example fix

// before
context.sendSourceEvent(new DefaultSplitRequest()); // wrong: default protocol
// after
context.sendSourceEvent(new SplitRequestEvent(finishedSplitIds, requesterHostname));

// in custom reader code: never call the default request hook
// ensure you emit the Iceberg event instead:
context.sendSourceEvent(new SplitRequestEvent(finishedIds, requesterHostname));

When it happens

Trigger: A SourceReader sends a plain default split request (e.g., a non-Iceberg reader added to the same source, or custom reader code calling context.sendSplitRequest() instead of emitting SplitRequestEvent) while an AbstractIcebergEnumerator is the enumerator.

Common situations: Custom/patched IcebergSourceReader modifications; mixing source reader implementations after a Flink version upgrade; accidentally using the generic SourceReader base class without the Iceberg split-request event.

Understand the failure class

Background: UnsupportedOperationException and "is not supported" errors: when a library deliberately refuses a call — this error's family across 30 libraries.

Related errors

• Altering schema is not supported in the old alterTable API.
• Altering partition keys is not supported yet.
• Creating table with computed columns is not supported yet.
• Creating table with watermark specs is not supported yet.
• Could not set a field in the RowDataWrapper because rowData

AI-assisted analysis of apache/iceberg@86d9c8fc54 (2026-09-12). Data as JSON: /api/errors/3a7d93768dc089ce. Report an issue: GitHub.