examination/topics/networking/http-keepalive/fill_blank.json

{
  "topic": "http-keepalive",
  "type": "fill_blank",
  "schema_version": "1.0.0",
  "generated": "2026-09-02T21:20:00+08:00",
  "questions": [
    {
      "id": "fb-001",
      "type": "fill_blank",
      "difficulty": 2,
      "tags": [
        "HTTP/1.0",
        "TCP",
        "连接模型"
      ],
      "question": "HTTP/1.0 不支持 Keep-Alive，每次请求都需要建立一个新的 TCP 连接，即____模型。",
      "answer": [
        "一请求一连接",
        "一次请求一个连接",
        "短连接"
      ],
      "answer_rule": "any",
      "explanation": "HTTP/1.0 默认行为是每次 HTTP 请求都经历完整的 TCP 三次握手，请求完成后立即关闭连接。这意味着每次请求都要付出建立和拆除连接的开销，在页面包含大量资源时效率极低。这是后来引入 Keep-Alive 机制的直接动因。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-002",
      "type": "fill_blank",
      "difficulty": 2,
      "tags": [
        "HTTP/1.1",
        "Keep-Alive",
        "默认行为"
      ],
      "question": "与 HTTP/1.0 不同，HTTP/1.1 默认启用____机制，无需显式声明即可复用 TCP 连接。",
      "answer": [
        "Keep-Alive",
        "持久连接",
        "keep-alive"
      ],
      "answer_rule": "any",
      "explanation": "HTTP/1.1 将 Keep-Alive（持久连接）作为默认行为。客户端和服务器建立连接后，可以在这个连接上依次发送和接收多个请求/响应，而不需要每次请求都重新建连。如果需要关闭连接，需要显式发送 Connection: close 头。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-003",
      "type": "fill_blank",
      "difficulty": 2,
      "tags": [
        "Connection头",
        "Keep-Alive",
        "HTTP头"
      ],
      "question": "HTTP 协议通过 ____ 请求头来协商连接复用行为。当值为 Keep-Alive 表示____，当值为 close 表示____。",
      "answer": [
        "Connection，希望复用连接，请求完成后关闭连接",
        "Connection 希望复用连接 请求完成后关闭连接"
      ],
      "answer_rule": "any",
      "explanation": "Connection 是 HTTP 中控制连接行为的首部字段。在 HTTP/1.1 中默认 Keep-Alive，通常不需要显式声明；但在 HTTP/1.0 中需要显式发送 Connection: Keep-Alive 才能启用复用。Connection: close 则用于告诉对端：这是最后一个请求，处理完请关闭连接。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-004",
      "type": "fill_blank",
      "difficulty": 3,
      "tags": [
        "Keep-Alive参数",
        "timeout",
        "max"
      ],
      "question": "Keep-Alive 的 timeout 参数表示____，max 参数表示____，两者任一达到限制服务器即关闭连接。",
      "answer": [
        "空闲超时时间，最大请求次数",
        "空闲超时时间 最大请求数量"
      ],
      "answer_rule": "any",
      "explanation": "timeout 指连接在空闲状态下（没有新请求）的最大存活时间，单位通常为秒；max 指该连接上允许处理的最大请求次数。例如 Keep-Alive: timeout=5, max=100 表示空闲超过 5 秒或已处理 100 个请求后关闭连接。服务器通过这两个参数防止连接被客户端长期占用。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-005",
      "type": "fill_blank",
      "difficulty": 3,
      "tags": [
        "队头阻塞",
        "HTTP/1.1",
        "串行"
      ],
      "question": "HTTP/1.1 Keep-Alive 模式下，同一连接上的请求是严格____的，前一个请求的响应未返回之前不能发送下一个请求，这被称为____。",
      "answer": [
        "串行，队头阻塞",
        "串行 队头阻塞"
      ],
      "answer_rule": "any",
      "explanation": "HTTP/1.1 的管道化（pipelining）虽然理论上允许客户端连续发送多个请求，但浏览器出于兼容性和安全性考虑基本未启用。实际中同一连接上的请求仍然是串行的——必须等上一个响应完全接收后才能发送下一个请求。这种「队头阻塞」（head-of-line blocking）是 HTTP/1.1 最大的性能瓶颈。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-006",
      "type": "fill_blank",
      "difficulty": 3,
      "tags": [
        "浏览器",
        "并发连接",
        "队头阻塞"
      ],
      "question": "为规避 HTTP/1.1 的队头阻塞问题，主流浏览器对同一域名默认开启____个 TCP 连接来并行加载资源。",
      "answer": [
        "6"
      ],
      "answer_rule": "all",
      "explanation": "Chrome、Firefox 等主流浏览器对同一域名同时打开 6 个 TCP 连接（HTTP/1.1），将资源分配到不同连接上并行下载。这是一种工程上的折中：既利用了 Keep-Alive 复用，又通过多连接规避了串行问题。但这会消耗更多连接数和内存，也是推动 HTTP/2 普及的重要原因之一。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-007",
      "type": "fill_blank",
      "difficulty": 4,
      "tags": [
        "HTTP演进",
        "Keep-Alive",
        "HTTP/2",
        "HTTP/3"
      ],
      "question": "HTTP 连接复用的演进路线为：无 Keep-Alive（HTTP/1.0）→____（HTTP/1.1）→____（HTTP/2）→____（HTTP/3，基于 QUIC 解决 TCP 层队头阻塞）。",
      "answer": [
        "Keep-Alive，多路复用，QUIC"
      ],
      "answer_rule": "ordered",
      "explanation": "HTTP/1.0 每请求一连接；HTTP/1.1 引入 Keep-Alive 实现连接复用但有队头阻塞；HTTP/2 通过多路复用在单连接上交错传输多个流，解决了应用层队头阻塞；HTTP/3 基于 QUIC 协议用 UDP 替代 TCP，连传输层的队头阻塞也一并解决。这条演进路线的核心驱动力就是不断提升连接复用效率。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-008",
      "type": "fill_blank",
      "difficulty": 3,
      "tags": [
        "空闲连接",
        "内存",
        "文件描述符"
      ],
      "question": "Keep-Alive 空闲连接虽然不传输数据，但仍占用服务器的____和____资源，大量空闲连接会导致资源耗尽。",
      "answer": [
        "内存，文件描述符",
        "内存 fd",
        "内存 socket"
      ],
      "answer_rule": "any",
      "explanation": "每条 TCP 连接在内核中需要维护 socket 缓冲区（消耗内存）和一个文件描述符（fd）。如果服务器对所有客户端保持大量空闲的 Keep-Alive 连接，内存和 fd 可能被耗尽，导致新连接无法建立。这就是为什么服务器必须配置合理的 keepalive_timeout 和 max connections 来限制空闲连接数。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-009",
      "type": "fill_blank",
      "difficulty": 3,
      "tags": [
        "服务器",
        "连接关闭",
        "RST"
      ],
      "question": "当服务器单方面关闭了一个 Keep-Alive 空闲连接时，客户端可能并不知道连接已断开，下次请求时会收到____错误。",
      "answer": [
        "Connection reset by peer",
        "连接被重置",
        "连接重置"
      ],
      "answer_rule": "any",
      "explanation": "服务器因超时或负载原因关闭了空闲连接后发送 FIN 或 RST 包，但如果客户端此时没有监听，可能不会感知。当下次请求尝试使用这条已关闭的连接时，会收到 Connection reset by peer（对端重置连接）或 Broken pipe 错误。成熟的客户端库会自动重试新建连接来处理此情况。",
      "source": null,
      "related": []
    },
    {
      "id": "fb-010",
      "type": "fill_blank",
      "difficulty": 2,
      "tags": [
        "Keep-Alive",
        "利弊",
        "权衡"
      ],
      "question": "Keep-Alive 的核心优势是____，主要代价是空闲时____，需要根据场景权衡使用。",
      "answer": [
        "减少建连开销、提升吞吐，占用内存和文件描述符",
        "减少握手开销 占用资源"
      ],
      "answer_rule": "any",
      "explanation": "Keep-Alive 通过复用 TCP 连接避免了反复建连和拆连的开销，显著提升高频请求场景的吞吐和延迟。但代价是空闲连接会持续占用服务器的内存和文件描述符资源。因此在高频访问的场景应开启，而对低频或一次性的请求，Keep-Alive 的收益很小，反而白白占用资源。",
      "source": null,
      "related": []
    }
  ]
}