grafana/k6 · error

the protobuf message is too large to be handled by Snappy en

Error message

the protobuf message is too large to be handled by Snappy encoder; size: %d, limit: %d

What it means

The marshaled protobuf WriteRequest is so large that its Snappy-encoded length would overflow uint32, so the batch can never be sent as one remote-write request. The reported size exceeds the ~4GiB encoder limit.

Source

Thrown at internal/output/prometheusrw/remote/client.go:130

	}()

	_, err = io.Copy(io.Discard, resp.Body)
	if err != nil {
		return err
	}

	return validateResponseStatus(resp.StatusCode)
}

func newWriteRequestBody(series []*prompb.TimeSeries) ([]byte, error) {
	b, err := proto.Marshal(&prompb.WriteRequest{
		Timeseries: series,
	})
	if err != nil {
		return nil, fmt.Errorf("encoding series as protobuf write request failed: %w", err)
	}
	if snappy.MaxEncodedLen(len(b)) < 0 {
		return nil, fmt.Errorf("the protobuf message is too large to be handled by Snappy encoder; "+
			"size: %d, limit: %d", len(b), math.MaxUint32)
	}
	return snappy.Encode(nil, b), nil
}

func validateResponseStatus(code int) error {
	if code >= http.StatusOK && code < 300 {
		return nil
	}

	return fmt.Errorf("got status code: %d instead expected a 2xx successful status code", code)
}

View on GitHub (pinned to 01ffac6f24)

Solutions

1. Reduce the number of time series per batch (lower K6_PROMETHEUS_RW_MAX_BATCH_SIZE or the equivalent)
2. Reduce metric cardinality: fewer custom tags/labels in the script
3. Shorten the flush interval so batches are smaller

When it happens

Trigger: Thrown at internal/output/prometheusrw/remote/client.go:130 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of grafana/k6@01ffac6f24 (2026-08-18). Data as JSON: /api/errors/0808c838db81713f. Report an issue: GitHub.