MemcachedContainer

A single-node Memcached container, ready-checked with a protocol-level version probe instead of the bare listening-port wait.

Default image: memcached:1.6-alpine Guest port: 11211

Why not the default wait strategy

Memcached logs nothing useful on startup, and the port-forwarding layer on either backend can bind the host port before the server inside is actually accepting — a bare TCP-connect wait (even with the standard read-probe) can pass while the first real client connection still gets a dead stream. This module ships a custom WaitStrategy that sends version\r\n over the wire and requires a reply starting with VERSION before considering the container ready — a genuine protocol handshake, not just “something answered.”

This is the worked example referenced throughout the rest of this book whenever a readiness signal needs to be protocol-level rather than a bare port check — see Wait Strategies.

Complete example

use rightsize_modules::MemcachedContainer;
use std::io::{Read, Write};
use std::net::TcpStream;
use std::time::Duration;

#[tokio::test]
async fn memcached_speaks_the_protocol() -> rightsize::Result<()> {
    let guard = MemcachedContainer::new().start().await?;

    let mut stream = TcpStream::connect(guard.address()).unwrap();
    stream.set_read_timeout(Some(Duration::from_secs(2))).unwrap();
    stream.write_all(b"version\r\n").unwrap();
    let mut buf = [0u8; 64];
    let n = stream.read(&mut buf).unwrap();
    assert!(String::from_utf8_lossy(&buf[..n]).starts_with("VERSION"));

    guard.stop().await?;
    Ok(())
}

By the time start() returns Ok, the module’s own wait strategy has already proven this exact protocol exchange succeeds — a test using a real memcached client crate is proving the client library, not the container’s readiness.

Backend notes

No memory-limit override and no known quirks beyond the readiness story above, which applies identically on both backends.