apache/beam · error

bad length prefix: %+v

Error message

bad length prefix: %+v

What it means

DecodeCoderRef maps length-prefix-type refs (used for bytes/string coders wrapped by Dataflow) back to coders. Such a ref must contain exactly one component; another count means the encoded graph is malformed, so decoding fails with the ref in the message.

Solutions

1. Regenerate the pipeline from source so length-prefix coders are re-encoded correctly.
2. Verify SDK version parity between the tool that produced the graph and the one decoding it.
3. Check any custom graph mutation passes for component-list edits on bytes/string coders.
4. Inspect the printed CoderRef to confirm the malformed component count and locate its producer.

Example fix

// before
ref := &graphx.CoderRef{Type: graphx.LengthPrefix, Components: []*graphx.CoderRef{a, b}}
c, err := graphx.DecodeCoderRef(ref)

// after
ref := &graphx.CoderRef{Type: graphx.LengthPrefix, Components: []*graphx.CoderRef{innerRef}}
c, err := graphx.DecodeCoderRef(ref)

func validLengthPrefixRef(r *graphx.CoderRef) bool {
	return r != nil && r.Type == graphx.LengthPrefix && len(r.Components) == 1
}

func isLengthPrefixRef(r *graphx.CoderRef) bool { return r != nil && r.Type == graphx.LengthPrefix && len(r.Components) == 1 }

c, err := graphx.DecodeCoderRef(ref)
if err != nil {
	if strings.Contains(err.Error(), "bad length prefix") {
		return nil, fmt.Errorf("corrupt length-prefix coder ref %v: %w", ref, err)
	}
	return nil, err
}

When it happens

Trigger: Decoding a CoderRef with Type lengthPrefixType whose Components has length != 1 — during deserialization of a pipeline graph or job description produced/corrupted outside the current encoder.

Common situations: Pipeline artifacts serialized by a different Beam version with a different length-prefix layout; custom graph rewriters duplicating components; truncated or manually edited proto/json pipelines.

Related errors

• bad iterable/stream: %+v
• bad nullable: %+v
• bad pair: %+v
• bad windowed value: %+v
• bad coder kind

AI-assisted analysis of apache/beam@12126d8942 (2026-09-13). Data as JSON: /api/errors/3839396def48b69f. Report an issue: GitHub.

Appendix: source

Thrown at sdks/go/pkg/beam/core/runtime/graphx/dataflow.go:300

		t := typex.New(typex.KVType, key.T, value.T)
		return &coder.Coder{Kind: coder.KV, T: t, Components: []*coder.Coder{key, value}}, nil

	case nullableType:
		if len(c.Components) != 1 {
			return nil, errors.Errorf("bad nullable: %+v", c)
		}

		inner, err := DecodeCoderRef(c.Components[0])
		if err != nil {
			return nil, err
		}

		t := typex.New(typex.NullableType, inner.T)
		return &coder.Coder{Kind: coder.Nullable, T: t, Components: []*coder.Coder{inner}}, nil

	case lengthPrefixType:
		if len(c.Components) != 1 {
			return nil, errors.Errorf("bad length prefix: %+v", c)
		}

		subC := c.Components[0]
		switch subC.Type {
		case stringType: // Needs special handling if wrapped by dataflow.
			return coder.NewString(), nil
		default:
			return decodeDataflowCustomCoder(subC.Type)
		}

	case intervalWindowType:
		return coder.NewIntervalWindowCoder(), nil

	case windowedValueType:
		if len(c.Components) != 2 {
			return nil, errors.Errorf("bad windowed value: %+v", c)
		}

View on GitHub (pinned to 12126d8942)