apache/druid · error · IllegalStateException

Could not transform value for

Error message

Could not transform value for %s reason: %s

What it means

ExpressionTransform.eval evaluates the Druid expression against a row's input bindings and wraps ANY Throwable (syntax error at plan time is separate; here it is evaluation failure) in this ISE naming the transform and the underlying message. It means the transform expression failed while computing a value for this row during ingestion or query-time transforms.

Solutions

1. Inspect the nested cause message and fix the expression (guard nulls, validate inputs) in your transforms spec.
2. Wrap risky sub-expressions in functions that handle nulls (e.g. NVL, TRY expressions where available) or use 'nullIf' patterns to avoid division by zero.
3. Verify referenced columns exist in the input data and check a sample of input records for unexpected values (nulls, empty strings, bad formats).
4. Test the expression standalone with Druid's expression evaluation (e.g. via a small ingestion spec or EXPLAIN /表达式 eval in tests) before full ingestion.

Example fix

// before
"transformSpec": { "transforms": [{ "name": "ratio", "expression": "a / b" }] }
// after
"transformSpec": { "transforms": [{ "name": "ratio", "expression": "a / (b + 0.0 == 0 ? null : b)" }] }

// validate expression before submitting ingestion:
// use Druid's built-in expression parser/validator or EXPLAIN on a query using the same expression,
// and null-guard inputs: expression should tolerate null columns.

try {
    Object value = transform.eval(row);
} catch (ISE e) {
    log.warn("Transform '{}' failed for row: {}", transform.getName(), e.getMessage(), e.getCause());
    metrics.incrementTransformFailures(transform.getName());
    return null; // or route row to a rejection path
}

When it happens

Trigger: Calling ExpressionTransform.eval(row) when the underlying expression evaluation throws — e.g. arithmetic on null/missing columns, division by zero, type errors in functions, or malformed expression logic that only fails for certain row values.

Common situations: Transform expressions referencing missing columns; regex_replace or timestampParse given unexpected row values; divide-by-zero in numeric transforms; NaN/empty string inputs to math functions during ingestion with transforms configured in the ioConfig.

Related errors

• Could not transform dimension value for
• Transform name ' ' cannot be used twice
• Attempt to add row to swapped-out sink for segment
• attempt to get boolean[] null vector from string[] only…
• attempt to get double[] from string[] only scalar binding

AI-assisted analysis of apache/druid@9b90983fd2 (2026-09-07). Data as JSON: /api/errors/413d86e337fa6a59. Report an issue: GitHub.

Appendix: source

Thrown at processing/src/main/java/org/apache/druid/segment/transform/ExpressionTransform.java:116

  static class ExpressionRowFunction implements RowFunction
  {
    private final String name;
    private final Expr expr;

    ExpressionRowFunction(final String name, final Expr expr)
    {
      this.name = name;
      this.expr = expr;
    }

    @Override
    public Object eval(final Row row)
    {
      try {
        return expr.eval(InputBindings.forRow(row)).value();
      }
      catch (Throwable t) {
        throw new ISE(t, "Could not transform value for %s reason: %s", name, t.getMessage());
      }
    }

    @Override
    public List<String> evalDimension(Row row)
    {
      try {
        return Rows.objectToStrings(expr.eval(InputBindings.forRow(row)).value());
      }
      catch (Throwable t) {
        throw new ISE(t, "Could not transform dimension value for %s reason: %s", name, t.getMessage());
      }
    }
  }

  @Override
  public boolean equals(final Object o)
  {

View on GitHub (pinned to 9b90983fd2)