Design and Implementation of a Generic Connection Pool in Go (Conecta)

Author LEE, a veteran with 17 years in IT, introduces the need for a lightweight, universal connection pool for various services such as Redis, MySQL, MongoDB, TCP/UDP, etc., and explains why reinventing the wheel is inefficient.

He demonstrates how typical client SDKs (e.g., github.com/go-sql-driver/mysql ) provide built‑in pools, but custom scenarios require manual management.

After analyzing existing ad‑hoc implementations (map+lock, channel, third‑party wrappers), he proposes a generic pool built on a channel‑like queue, supporting configurable max connections, idle limits, ping checks, and custom callbacks.

Key design includes a NewPool constructor, Get , Put , GetOrCreate methods, a background goroutine that periodically pings connections, and extensible hooks ( WithNewFunc , WithPingFunc , WithCloseFunc , WithCallback , etc.).

func execSQL() error {
    // 打开数据库
    db, err := sql.Open("mysql", "user:password@/dbname")
    if err != nil {
        return err
    }
    defer db.Close()

    // 执行 SQL 语句
    db.Exec("INSERT INTO user (name, age) VALUES (?, ?)", "老李", 18)
    return nil
}

The pool can be created with a simple configuration:

pool := NewPool(10, 30, 10*time.Second, func() (interface{}, error) {
    // 创建一个 TCP 连接
    return net.Dial("tcp", "192.168.0.1:80")
})
defer pool.Close()

conn, err := pool.Get()
if err != nil {
    // handle error
}
if _, err := conn.Write([]byte("hello")); err != nil {
    conn.Close()
    return
}
pool.Put(conn)

A periodic timer checks connection health:

// 创建一个定时器，每 p.config.scanInterval 毫秒触发一次
ticker := time.NewTicker(time.Millisecond * time.Duration(p.config.scanInterval))

If a connection fails ping checks beyond the maximum retry count, it is closed and the user‑defined OnClose callback is invoked:

if retryCount >= p.config.maxRetries {
    if value != nil {
        err := p.config.closeFunc(value)
        p.config.callback.OnClose(value, err)
        element.SetData(nil)
    }
} else {
    if ok := p.config.pingFunc(value, retryCount); ok {
        element.SetValue(0)
        p.config.callback.OnPingSuccess(value)
    } else {
        element.SetValue(int64(retryCount) + 1)
        p.config.callback.OnPingFailure(value)
    }
}

The Conecta library (https://github.com/shengyanli1982/conecta) offers a simple API, lifecycle management, and can be integrated with any queue implementation, aiming to reduce duplicated effort across projects.

Conclusion: Conecta provides a reusable, configurable connection‑pool solution for backend services, improving development efficiency and maintainability.