apache/beam · error · IllegalArgumentException

Cannot set both startVersion and startTimestamp.

Error message

Cannot set both startVersion and startTimestamp.

What it means

expand() enforces that the read start point is unambiguous: startVersion and startTimestamp are mutually exclusive. Both express different semantics (exact snapshot version vs. time travel by timestamp), so setting both is rejected with IllegalArgumentException.

Source

Thrown at sdks/java/io/delta/src/main/java/org/apache/beam/sdk/io/delta/DeltaIO.java:331

    }

    public ReadChanges withConfig(Map<String, String> config) {
      return toBuilder().setHadoopConfig(config).build();
    }

    @Override
    public PCollection<Row> expand(PBegin input) {
      String path = getTablePath();
      if (path == null) {
        throw new IllegalArgumentException("Table path must be set.");
      }
      if (getStartVersion() == null && getStartTimestamp() == null) {
        // TODO: for unbounded reads, support using current HEAD or the latest snapshot
        // as the default starting point.
        throw new IllegalArgumentException("Either startVersion or startTimestamp must be set.");
      }
      if (getStartVersion() != null && getStartTimestamp() != null) {
        throw new IllegalArgumentException("Cannot set both startVersion and startTimestamp.");
      }
      if (getEndVersion() != null && getEndTimestamp() != null) {
        throw new IllegalArgumentException("Cannot set both endVersion and endTimestamp.");
      }

      Configuration conf = new Configuration();
      Map<String, String> hadoopConfig = getHadoopConfig();
      if (hadoopConfig != null) {
        for (Map.Entry<String, String> entry : hadoopConfig.entrySet()) {
          conf.set(entry.getKey(), entry.getValue());
        }
      }
      Engine engine = DefaultEngine.create(conf);
      Table table = Table.forPath(engine, path);

      TableImpl tableImpl = (TableImpl) table;

      long resolvedEndVersion;

View on GitHub (pinned to 12126d8942)

Solutions

1. Keep only one: remove .withStartTimestamp(...) and rely on startVersion
2. Or remove .withStartVersion(...) and keep .withStartTimestamp(...)
3. If both come from config, add explicit validation/mutual-exclusion handling in your pipeline options

Example fix

// before
DeltaIO.read().withTable(path).withStartVersion(5L).withStartTimestamp("2024-01-01T00:00:00");
// after
DeltaIO.read().withTable(path).withStartVersion(5L);

if (startVersion != null && startTimestamp != null) {
  throw new IllegalArgumentException("Choose either startVersion or startTimestamp, not both");
}

When it happens

Trigger: Calling both .withStartVersion(v) and .withStartTimestamp(t) on the same DeltaIO.read() builder before applying the transform.

Common situations: Merging two configurations (one specifying a version, one a timestamp); copy-pasting from examples that use different start styles; a generic time-travel config exposing both fields and the code forwarding both.

Related errors

• Cannot set both endVersion and endTimestamp.
• Either startVersion or startTimestamp must be set.
• Timing number 0b" + timingNumber.toString(2) + " has more th
• No filesystem found for scheme
• Exploded field %s must be an iterable type, got %s.

AI-assisted analysis of apache/beam@12126d8942 (2026-09-13). Data as JSON: /api/errors/dd11b08404220e0e. Report an issue: GitHub.