Memcached "CLIENT_ERROR bad data chunk" in CI

By Kaveh Alemi·Latchkey

The memcached text protocol declares a byte length before the value, and if the bytes sent do not match, the server replies "CLIENT_ERROR bad data chunk." In CI this usually comes from a byte-vs-character length mismatch (multibyte data) or a mismatched client library.

What this error means

A store command fails with "CLIENT_ERROR bad data chunk" when the value contains multibyte characters or when a raw socket protocol is hand-written.

CLIENT_ERROR bad data chunk

Common causes

Declared length does not match the byte length

The length prefix was computed from character count, not byte count, so multibyte UTF-8 values overrun or underrun the frame.

A hand-rolled or mismatched protocol implementation

Custom socket code or an incompatible client mis-frames the value, breaking the text protocol.

How to fix it

Use a maintained client library

1. Prefer pymemcache or a mature client that frames values correctly.
2. Store bytes, letting the client compute the byte length.
3. Avoid hand-writing the memcached text protocol.

from pymemcache.client.base import Client
c = Client(("localhost", 11211))
c.set(b"key", "caf\u00e9".encode("utf-8"))

Compute length in bytes, not characters

If you must speak the protocol directly, encode first and use the byte length in the command.

data = value.encode("utf-8")
# length is len(data), the byte count, not len(value)

How to prevent it

• Let a real client library handle framing and lengths.
• Encode to bytes and use byte length for the protocol.
• Do not hand-write the memcached text protocol in tests.

Frequently asked questions

Related guides

References

• GitHub Actions documentation