apache/beam · error

passert.Hash( ) = ( , ), want ( , )

Error message

passert.Hash(%v) = (%v,%v), want (%v,%v)

What it means

passert.Hash asserts both the size and the base64-encoded MD5 hash of a PCollection's concatenated element bytes. errFn.ProcessElement fails with 'passert.Hash(%v) = (%v,%v), want (%v,%v)' when either the observed length or hash differs from the expectation.

Solutions

1. Compare the observed (size, hash) in the message with expected to see whether content or length changed.
2. If the change is intentional, recompute and update the expected hash/size in the test.
3. Make element ordering deterministic (sort before hashing) if order varies.
4. Check for coder changes that alter serialized bytes.

Example fix

// before
passert.Hash(s, col, "c29832...", 42)
// after
// recompute expected hash after a legitimate data/coder change:
passert.Hash(s, col, "9f86d0...", 42)

h := md5.Sum(serializedBytes)
want := base64.StdEncoding.EncodeToString(h[:])
if want != expectedHash {
	t.Fatalf("expected hash stale: have %s want %s", want, expectedHash)
}

if err := beamx.Execute(ctx, p); err != nil {
	if strings.Contains(err.Error(), "passert.Hash") {
		// diff observed (len,hash) vs expected; recompute expectation if change is intentional
	}
}

When it happens

Trigger: The serialized element bytes differ from what the expectation was computed against — element values, ordering, or the coder changed, altering len(col) or the MD5 digest.

Common situations: Coder or serialization format changed between library versions; nondeterministic element order; upstream data or transforms modified content.

Understand the failure class

Background: Checksum mismatch errors: "checksum verification failed", "digest mismatch", "expected vs actual checksum" — what they mean and how to fix them — this error's family across 41 libraries.

Related errors

• missing entries (missing in actual, present in expected)
• observed PCollection has incompatible type
• passert.Count( ) = , want
• passert.Diff input PColections don't have matching types
• PCollections of different lengths, got

AI-assisted analysis of apache/beam@12126d8942 (2026-09-13). Data as JSON: /api/errors/5f4b6ece47e9baf0. Report an issue: GitHub.

Appendix: source

Thrown at sdks/go/pkg/beam/testing/passert/hash.go:63

func (f *hashFn) ProcessElement(_ int, lines func(*string) bool) error {
	var col []string
	var str string
	for lines(&str) {
		col = append(col, str)
	}
	sort.Strings(col)

	md5W := md5.New()
	for _, str := range col {
		if _, err := md5W.Write([]byte(str)); err != nil {
			panic(err) // cannot fail
		}
	}
	hash := base64.StdEncoding.EncodeToString(md5W.Sum(nil))

	if f.Size != len(col) || f.Hash != hash {
		return errors.Errorf("passert.Hash(%v) = (%v,%v), want (%v,%v)", f.Name, len(col), hash, f.Size, f.Hash)
	}
	return nil
}

View on GitHub (pinned to 12126d8942)