apache/seatunnel · error · JobDefineCheckException

Transform/Sink don't support processing data with two…

Error message

Transform/Sink don't support processing data with two different structures.

What it means

checkProducedTypeEquals validates that all input actions feeding one transform or sink produce identical SeaTunnelDataTypes. A transform/sink merges multiple upstream inputs only when their row schemas match; two different structures cannot be combined, so parsing fails with this JobDefineCheckException.

Solutions

1. Ensure all inputs share the same schema, or split the sink into one sink per structure.
2. Insert a transform (e.g. Sql/FieldMapper/Replace) to normalize each branch to a common schema before merging.
3. Restrict a multi-table source with table_names/table_path regex so only same-structure tables feed the sink.
4. Verify upstream table schemas haven't drifted; refresh cached catalogs if needed.

Example fix

// before: one sink fed by multi-table source with differing schemas
source { Jdbc { table_path = "db.*" } }
sink { Console {} }
// after: group by compatible table sets
source { Jdbc { table_path = "db.orders_.*" } }
sink { Console {} }

// ensure all inputs to a sink share one schema
CatalogTable first = inputs.get(0).getCatalogTable();
for (Action a : inputs) {
  if (!first.getTableSchema().equals(getProducedType(a).getTableSchema())) {
    throw new IllegalArgumentException("Divergent input schemas feeding one sink");
  }
}

try {
  engine.submitJob(config);
} catch (JobDefineCheckException e) {
  if (e.getMessage().contains("two different structures")) {
    // split sinks or add normalizing transforms
  }
}

When it happens

Trigger: parseTransform or parseSink receives a Set of input actions (multi-table source or multiple upstream branches) whose produced CatalogTable types differ (different columns or types), detected during DAG construction.

Common situations: Connecting a multi-table source (table_path matching several tables with different schemas) directly to one sink; merging two branches of different schemas into one transform; upstream schema drift after a table was altered.

Related errors

• Sink don't support simultaneous writing of data from…
• All candidate sink tables were skipped during job parsing.
• Can't find column in table.
• COMMON-02
• COMMON-02

AI-assisted analysis of apache/seatunnel@cf67b549a7 (2026-09-10). Data as JSON: /api/errors/bec6a863b2cd6f1f. Report an issue: GitHub.

Appendix: source

Thrown at seatunnel-engine/seatunnel-engine-core/src/main/java/org/apache/seatunnel/engine/core/parse/MultipleTableJobConfigParser.java:605

            } catch (UnsupportedOperationException e) {
                // TODO remove it when all connector use `getProducedCatalogTables`
                return ((SourceAction<?, ?, ?>) action).getSource().getProducedType();
            }
        } else if (action instanceof TransformAction) {
            return ((TransformAction) action)
                    .getTransform()
                    .getProducedCatalogTable()
                    .getSeaTunnelRowType();
        }
        throw new UnsupportedOperationException();
    }

    public static void checkProducedTypeEquals(Set<Action> inputActions) {
        SeaTunnelDataType<?> expectedType = getProducedType(new ArrayList<>(inputActions).get(0));
        for (Action action : inputActions) {
            SeaTunnelDataType<?> producedType = getProducedType(action);
            if (!expectedType.equals(producedType)) {
                throw new JobDefineCheckException(
                        "Transform/Sink don't support processing data with two different structures.");
            }
        }
    }

    @Deprecated
    private static <T> T findLast(LinkedHashMap<?, T> map) {
        int size = map.size();
        int i = 1;
        for (T value : map.values()) {
            if (i == size) {
                return value;
            }
            i++;
        }
        // never execution
        return null;
    }

View on GitHub (pinned to cf67b549a7)