[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"blog-strapi-article:zh:parallel-browser-proxy-session-test-isolation":3,"blog-article:\u002Fblog\u002Fparallel-browser-proxy-session-test-isolation":1198,"blog-related:\u002Fblog\u002Fparallel-browser-proxy-session-test-isolation":1199,"blog-translations:\u002Fblog\u002Fparallel-browser-proxy-session-test-isolation":1200},{"available":4,"article":5,"relatedArticles":249,"translations":1033},true,{"source":6,"id":7,"documentId":8,"translationKey":8,"locale":9,"slug":10,"path":11,"_path":11,"title":12,"description":13,"excerpt":13,"image":14,"publishedAt":15,"updatedAt":16,"author":17,"tags":18,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":23,"bodyHtml":248},"strapi",3928,"fwsk35qf6e3yt8ck5lc2kgmy","zh","parallel-browser-proxy-session-test-isolation","\u002Fblog\u002Fparallel-browser-proxy-session-test-isolation","并行运行浏览器测试时，如何避免代理IP 会话与测试数据互相干扰","并行跑浏览器测试时，用例之间最容易互相借走代理会话、cookie 和测试数据，结果通过得莫名其妙。文章给出一套可复现的检查方式：三维并发矩阵、每条请求带不可复用的标记、一次故意制造的串扰对照，再做清理与重跑，用来确认隔离是否真的生效。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_cover_614dedacff.webp","2026-11-04T06:47:41.196Z","2026-11-04T06:47:41.178Z","Global Proxies Team",[],false,0,3,"blocks",[24,30,36,53,57,61,65,69,83,87,99,103,107,125,129,133,137,141,145,149,153,171,175,186,190,201,205,209,213,217,224,228,234,238,244],{"type":25,"children":26},"paragraph",[27],{"text":28,"type":29},"并发测试最难排查的情况，是结果通过了但用例实际共享了状态。浏览器上下文、服务端账号状态、代理会话和日志归属是不同隔离层，必须分别核对。本篇用并发矩阵、请求标记和分组负对照验证隔离，不把共享出口本身当成会话串扰证据。","text",{"type":31,"level":32,"children":33},"heading",2,[34],{"text":35,"type":29},"并行测试为什么会出现会话串扰",{"type":37,"format":38,"children":39},"list","unordered",[40,45,49],{"type":41,"children":42},"list-item",[43],{"text":44,"type":29},"数据串：A 用例写入的 cookie、localStorage 或登录态，出现在 B 用例的请求里。",{"type":41,"children":46},[47],{"text":48,"type":29},"出口串：B 用例的请求从 A 用例那条代理会话的出口 IP 发出，地区或身份标签随之错位。",{"type":41,"children":50},[51],{"text":52,"type":29},"记录串：埋点、日志或测试数据归属到错误的 run id、用例 id，事后无法回溯。",{"type":25,"children":54},[55],{"text":56,"type":29},"数据串和出口串会让断言\"恰好通过\"，最难发现；记录串通常先暴露，因为它不依赖业务逻辑。判定顺序建议是先看记录归属，再核对出口，最后才查数据本身。",{"type":31,"level":32,"children":58},[59],{"text":60,"type":29},"如何把用例、浏览器上下文和代理会话分开",{"type":25,"children":62},[63],{"text":64,"type":29},"并发一提高，组合就会失控。用例、浏览器上下文和代理IP会话三个维度写进同一张矩阵，每个格子只允许一种合法组合——任意一格填了两种组合，就等于把串扰写进了测试。",{"type":31,"level":21,"children":66},[67],{"text":68,"type":29},"每个测试用例应该绑定哪些资源",{"type":37,"format":38,"children":70},[71,75,79],{"type":41,"children":72},[73],{"text":74,"type":29},"用例 × 浏览器上下文：通常为每个用例创建独立 context。`storageState` 文件是创建时载入的初始状态快照；多个 context 读取同一只读快照，不会因此自动共享之后的 Cookie 更新。避免并发写同一状态文件。",{"type":41,"children":76},[77],{"text":78,"type":29},"上下文 × 代理会话：一个上下文绑定一条会话，用例结束时释放，不跨用例借用。这是本套测试的约束，不是平台默认——连接池、持久化上下文和代理服务的粘性策略都可能打破它。",{"type":41,"children":80},[81],{"text":82,"type":29},"工作进程 × 出口：同一 worker 内允许轮换出口，但轮换点必须写进日志；验收时同时记录上下文 ID、会话 ID 与连接\u002F出口变化，才能证明轮换没有落到别的用例上。",{"type":31,"level":21,"children":84},[85],{"text":86,"type":29},"为什么只隔离浏览器上下文还不够",{"type":25,"children":88},[89,91,97],{"text":90,"type":29},"Playwright 的独立 `BrowserContext` 默认隔离本地 Cookie、缓存和存储状态；`storageState` 用于初始化上下文。代理既可按浏览器启动配置，也可在支持的 API 中按 context 配置。具体选项受 Playwright 版本及浏览器类型影响，应核对",{"text":92,"type":29,"marks":93},"BrowserContext API",[94],{"href":95,"type":96},"https:\u002F\u002Fplaywright.dev\u002Fdocs\u002Fapi\u002Fclass-browser#browser-new-context-option-proxy","link",{"text":98,"type":29},"并用最小脚本实测，记录每个 context 的代理分配及实际请求出口。",{"type":31,"level":21,"children":100},[101],{"text":102,"type":29},"代理池回收错位会怎样影响下一条用例",{"type":25,"children":104},[105],{"text":106,"type":29},"上下文池和会话池回收节奏不同会增加错绑风险，但不代表编号必然错位。判定时用 run\u002Fcase\u002Fcontext、供应商会话标识、分配时间和实际连接记录共同核对；相同出口 IP 可能是合法共享，不能单独作为错绑证据。",{"type":37,"format":38,"children":108},[109,113,117,121],{"type":41,"children":110},[111],{"text":112,"type":29},"隔离层：浏览器上下文 ｜ 隔离手段：每用例独立 context；状态文件只作初始快照 ｜ 串扰证据：B 收到仅在 A 的 context 内写入的唯一 Cookie ｜ 判定：串扰",{"type":41,"children":114},[115],{"text":116,"type":29},"隔离层：代理会话 ｜ 隔离手段：按测试设计绑定并记录会话分配 ｜ 串扰证据：会话标识、分配记录和实际连接显示 B 被错绑到 A 的会话 ｜ 判定：串扰；仅出口 IP 相同不足以判定",{"type":41,"children":118},[119],{"text":120,"type":29},"隔离层：日志归属 ｜ 隔离手段：请求头携带 run \u002F case 标记 ｜ 串扰证据：标记与用例不匹配 ｜ 判定：串扰",{"type":41,"children":122},[123],{"text":124,"type":29},"隔离层：服务端测试数据 ｜ 隔离手段：按 run id 前缀写入并清理 ｜ 串扰证据：查出其他 run 的记录 ｜ 判定：串扰",{"type":25,"children":126},[127],{"text":128,"type":29},"表里每一层都要留下可检索的证据，只有结论没有证据的条目，等于没隔离。",{"type":25,"children":130},[131],{"text":132,"type":29},"并发量、连接池和会话复用应在测试计划中分别记录；配置参数按实际代理产品文档确认。",{"type":31,"level":32,"children":134},[135],{"text":136,"type":29},"如何给每条请求加上唯一标记",{"type":25,"children":138},[139],{"text":140,"type":29},"标记要同时满足唯一和可穿透。启动时生成 run id（时间戳加随机串），每个请求带 `X-Test-Run: \u003Crun>-\u003Ccase>-\u003Cseq>`，回显服务把头原样写回日志。断言只认自己前缀的数据，见到别的前缀就直接判失败。",{"type":25,"children":142},[143],{"text":144,"type":29},"不可复用是关键：固定字符串会让上一轮的残留看起来像本轮通过。每轮重新生成 run id，测试数据按前缀写入，清理也按前缀删除，避免把并发中的其他轮次一起删掉。",{"type":31,"level":32,"children":146},[147],{"text":148,"type":29},"如何关闭浏览器侧的隐式共享",{"type":25,"children":150},[151],{"text":152,"type":29},"显式配置容易对齐，隐式共享才是漏洞来源：",{"type":37,"format":38,"children":154},[155,159,163,167],{"type":41,"children":156},[157],{"text":158,"type":29},"持久化上下文：`launchPersistentContext` 复用 user data 目录，cookie 与缓存天然共享，不适合并发用例。",{"type":41,"children":160},[161],{"text":162,"type":29},"Service Worker 与 HTTP 缓存：同源资源被缓存后，表现为\"没发请求却拿到了结果\"。",{"type":41,"children":164},[165],{"text":166,"type":29},"共享 storageState 文件：多个 context 读取同一只读文件只会得到相同初始状态；风险在于并发写同一文件或共享服务端账号状态。",{"type":41,"children":168},[169],{"text":170,"type":29},"DNS 与连接复用：进程内连接池复用，出口切换后仍可能走旧连接。",{"type":25,"children":172},[173],{"text":174,"type":29},"处理办法不是逐个打补丁，而是规定并发用例不落盘、不共享目录，把持久化留给串行场景。抓包或代理日志里若出现用例 B 复用了用例 A 的连接，先查上下文是否共用，再查连接池。",{"type":25,"children":176},[177,179,184],{"text":178,"type":29},"浏览器实例准备独立代理参数时，可对照",{"text":180,"type":29,"marks":181},"代理设置帮助",[182],{"href":183,"type":96},"https:\u002F\u002Fhelp.quanqiudaili.com\u002Fdocs\u002Fproxy-Settings",{"text":185,"type":29},"。",{"type":31,"level":32,"children":187},[188],{"text":189,"type":29},"如何用一次可控串扰验证隔离是否生效",{"type":25,"children":191},[192,194,199],{"text":193,"type":29},"检测手段本身也要验证。在自建环境分开做三组负对照，每次只注入一种错误：①故意共用同一 context，在 A 写入唯一 Cookie，确认 B 能观察到该状态且隔离断言失败；②故意把 B 绑定到错误会话，用供应商会话标识和实际连接记录确认错绑断言失败；③故意篡改 case 标记，确认日志归属断言失败。共享初始 `storageState` 或出口 IP 相同，不保证发生 Cookie 交叉或标记错乱。对照结束后恢复配置，以新 run id 重跑。Playwright 的",{"text":195,"type":29,"marks":196},"认证状态文档",[197],{"href":198,"type":96},"https:\u002F\u002Fplaywright.dev\u002Fdocs\u002Fauth",{"text":200,"type":29},"说明了状态快照的使用方式。",{"type":31,"level":32,"children":202},[203],{"text":204,"type":29},"测试结束后如何清理并确认隔离结果",{"type":25,"children":206},[207],{"text":208,"type":29},"每轮收尾按固定顺序：关闭上下文与浏览器实例、释放代理会话、删除本轮前缀的测试数据、归档日志。顺序反了会出现\"浏览器还开着、会话已释放\"，重连请求落到新会话上。",{"type":25,"children":210},[211],{"text":212,"type":29},"验收看四件事：全量重跑结果与首次一致；日志中没有跨 run 标记；出口 IP 与矩阵记录逐条对得上；测试数据清理后计数归零。四项都过，才能说这一轮并发是干净的。",{"type":31,"level":32,"children":214},[215],{"text":216,"type":29},"FAQ",{"type":25,"children":218},[219],{"text":220,"type":29,"marks":221},"Q：串扰只在并发高时出现，低并发是不是不用管？",[222],{"type":223},"bold",{"type":25,"children":225},[226],{"text":227,"type":29},"A：低并发不代表不存在状态或账号共享问题。独立 context 可隔离本地状态；仍应检查共享服务端账号、并发写状态文件和实际复用 context 等风险。",{"type":25,"children":229},[230],{"text":231,"type":29,"marks":232},"Q：代理IP会话由池自动分配，还能做到一会话一上下文吗？",[233],{"type":223},{"type":25,"children":235},[236],{"text":237,"type":29},"A：可以。把会话获取放在上下文创建处，随上下文销毁释放；池只负责供给，不负责跨用例复用。",{"type":25,"children":239},[240],{"text":241,"type":29,"marks":242},"Q：用回显服务判断出口，和看代理后台记录哪个更可靠？",[243],{"type":223},{"type":25,"children":245},[246],{"text":247,"type":29},"A：两者互补。回显服务证明请求实际从哪里出去，后台记录证明会话被谁使用，对不上时先查会话分配环节。","",[250,489,760],{"source":6,"id":251,"documentId":252,"translationKey":252,"locale":9,"slug":253,"path":254,"_path":254,"title":255,"description":256,"excerpt":256,"image":257,"publishedAt":258,"updatedAt":259,"author":17,"tags":260,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":261,"bodyHtml":248},3937,"eg98712cpvp1nwo9plj025g4","websocket-reconnection-session-recovery-testing","\u002Fblog\u002Fwebsocket-reconnection-session-recovery-testing","WebSocket 断线重连：代理IP 出口切换后如何验证会话恢复","客户端显示已重连，用户却丢了消息或看到重复下单，是实时应用常见的假恢复。文章按客户端正常关闭、网络中断与服务端重启三类断线分别制造中断，比较恢复耗时、丢失与重复条数，再用消息序号、去重窗口和业务状态核对，确认恢复的是业务而不只是连接。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_Web_Socket_cover_2052233c88.webp","2026-11-04T06:50:15.056Z","2026-11-04T06:50:15.027Z",[],[262,266,270,280,284,288,292,296,300,310,314,336,340,344,358,362,366,370,388,392,396,400,404,408,412,416,430,434,438,446,456,459,465,469,475,479,485],{"type":25,"children":263},[264],{"text":265,"type":29},"日志里出现\"重连成功\"，并不代表业务恢复完成：消息可能缺失、重复处理或状态未同步。应按受控断线场景检查重连与业务状态，并根据协议契约验证恢复位置。本篇不预设断线一定发生在代理出口，也不把单一关闭码当作根因。",{"type":31,"level":32,"children":267},[268],{"text":269,"type":29},"怎样判断连接恢复是否已经变成业务恢复",{"type":37,"format":38,"children":271},[272,276],{"type":41,"children":273},[274],{"text":275,"type":29},"连接恢复：新的 WebSocket 完成握手，收到 101，可以收发消息。",{"type":41,"children":277},[278],{"text":279,"type":29},"业务恢复：断线期间产生的消息被补齐或明确丢弃，恢复后不重复、不乱序，业务状态与服务端一致。",{"type":25,"children":281},[282],{"text":283,"type":29},"连接恢复是必要条件，不是充分条件。客户端重连后如果没有带上恢复凭据，服务端可能按游标、时间窗或自身策略补发部分或全部消息，也可能直接拒绝恢复；策略不同，出问题的形态也不同：全量重放会重复，拒绝恢复会缺失。具体按哪种策略，以协议文档和测试结果为准。",{"type":31,"level":32,"children":285},[286],{"text":287,"type":29},"如何分别制造三种可控断线",{"type":25,"children":289},[290],{"text":291,"type":29},"不同断线方式会触发不同的关闭码与恢复路径，必须分开测：",{"type":25,"children":293},[294],{"text":295,"type":29},"1. 客户端主动关闭：调用 `close(1000)`，按应用预定策略验证停止或重新连接及状态处理；不自动重连不必然表示故障。\n2. 网络层中断：在自建测试网络中丢弃连接、发送 RST 或停止客户端进程等，记录故障点和方向。若客户端报告 1006，将其视为本地未收到有效 Close 帧的异常状态，不预设所有中断都会触发该码。\n3. 服务端主动断开：服务端关闭连接或重启，验证重连后的鉴权与恢复流程。",{"type":25,"children":297},[298],{"text":299,"type":29},"在自建测试环境中制造中断，不对正常业务链路注入故障。先做少量试跑检查测试流程，再根据业务 SLA、风险和观测波动确定正式覆盖次数。断线时机可覆盖空闲期、消息传输中和鉴权刚完成时，并记录 TCP RST、丢包、进程退出、心跳超时等实际证据。",{"type":25,"children":301},[302,304,309],{"text":303,"type":29},"网络层中断的定位可参考",{"text":305,"type":29,"marks":306},"代理 IP 故障排查中心",[307],{"href":308,"type":96},"https:\u002F\u002Fhelp.quanqiudaili.com\u002Fdocs\u002FSmart-check",{"text":185,"type":29},{"type":31,"level":32,"children":311},[312],{"text":313,"type":29},"重连策略应该观察哪些指标",{"type":37,"format":38,"children":315},[316,320,324,328,332],{"type":41,"children":317},[318],{"text":319,"type":29},"首次重连延迟：从检测到断线到发出重连请求的时间。",{"type":41,"children":321},[322],{"text":323,"type":29},"重连间隔序列：是否指数退避，是否带随机抖动。",{"type":41,"children":325},[326],{"text":327,"type":29},"重连次数上限：达到上限后是放弃还是降级。",{"type":41,"children":329},[330],{"text":331,"type":29},"鉴权是否重做：令牌过期时是否重新获取。",{"type":41,"children":333},[334],{"text":335,"type":29},"是否携带恢复凭据：resume token、最后一条消息的序号或事件 ID。",{"type":25,"children":337},[338],{"text":339,"type":29},"退避没有抖动时，大量客户端会在同一时刻重连，形成二次冲击。验证时观察间隔序列是否递增且带随机量。",{"type":31,"level":32,"children":341},[342],{"text":343,"type":29},"常见的三种会话恢复机制怎么验证",{"type":37,"format":38,"children":345},[346,350,354],{"type":41,"children":347},[348],{"text":349,"type":29},"服务端保留游标：客户端带上最后确认的序号，服务端从下一条开始推送。",{"type":41,"children":351},[352],{"text":353,"type":29},"客户端拉取历史：重连后主动请求断线时间窗内的消息，再切回实时流。",{"type":41,"children":355},[356],{"text":357,"type":29},"无状态重放：服务端按时间窗重放，客户端负责去重。",{"type":25,"children":359},[360],{"text":361,"type":29},"序号的单调范围可能限于某个分区、订阅或过滤条件；先确认序号作用域、客户端确认点、乱序窗口和过滤规则。三种机制按产品契约分别验证：游标是否被接受、历史拉取与实时流是否衔接、重放窗口是否被正确去重。缺少客户端 token 不代表服务端没有已保存的恢复位置。",{"type":31,"level":32,"children":363},[364],{"text":365,"type":29},"如何检查补发消息是否重复或乱序",{"type":25,"children":367},[368],{"text":369,"type":29},"恢复阶段最容易出现\"补多了\"和\"补乱了\"。校验方法：",{"type":37,"format":38,"children":371},[372,376,380,384],{"type":41,"children":373},[374],{"text":375,"type":29},"每条业务消息带唯一 ID 与单调递增序号。",{"type":41,"children":377},[378],{"text":379,"type":29},"客户端记录已处理 ID 的窗口，重复 ID 直接丢弃。",{"type":41,"children":381},[382],{"text":383,"type":29},"收到消息时按已确认的序号作用域检查连续性；出现缺口后，按协议选择缓冲等待、补拉历史或进入降级\u002F重新同步，不规定所有系统都必须立即停止消费。",{"type":41,"children":385},[386],{"text":387,"type":29},"分别记录重复接收、去重后重复产生的业务副作用、最终缺失条数，作为恢复质量指标。",{"type":25,"children":389},[390],{"text":391,"type":29},"序号不能只靠时间戳，同一毫秒内的多条消息会无法排序。服务端生成序号时保证单调，客户端只做校验与去重。",{"type":31,"level":32,"children":393},[394],{"text":395,"type":29},"如何确认重连后业务状态仍然一致",{"type":25,"children":397},[398],{"text":399,"type":29},"连接层和消息层都正常，业务仍可能不一致。核对三项：",{"type":25,"children":401},[402],{"text":403,"type":29},"1. 关键状态字段：账户余额、订单状态、会话归属是否与服务端一致。\n2. 未确认操作：断线前发出但未收到确认的请求，重连后是否重发、是否被服务端去重。\n3. 幂等性：重发的写操作是否只生效一次（去重机制见上一节）。",{"type":25,"children":405},[406],{"text":407,"type":29},"业务核对要读服务端的权威状态，不能只信客户端本地缓存。本地缓存与服务端不一致时，以服务端为准并记录差异。",{"type":31,"level":32,"children":409},[410],{"text":411,"type":29},"如何记录恢复时间并完成验收",{"type":25,"children":413},[414],{"text":415,"type":29},"恢复时间要拆成两个数：断线到新连接可用，断线到业务补齐完成。前者反映重连策略，后者反映恢复机制。只报前者会低估用户实际等待时间。",{"type":37,"format":38,"children":417},[418,422,426],{"type":41,"children":419},[420],{"text":421,"type":29},"断线类型：客户端 close(1000) ｜ 恢复耗时：1.2 s ｜ 丢失条数：0 ｜ 重复条数：0 ｜ 业务状态一致：是 ｜ 证据：序号连续、状态字段一致",{"type":41,"children":423},[424],{"text":425,"type":29},"断线类型：网络中断（1006） ｜ 恢复耗时：3.5 s ｜ 丢失条数：0 ｜ 重复条数：2 ｜ 业务状态一致：是 ｜ 证据：去重窗口命中、补齐完成",{"type":41,"children":427},[428],{"text":429,"type":29},"断线类型：服务端重启 ｜ 恢复耗时：8.0 s ｜ 丢失条数：0 ｜ 重复条数：0 ｜ 业务状态一致：是 ｜ 证据：重新鉴权、游标续传",{"type":25,"children":431},[432],{"text":433,"type":29},"表中数值仅为示例，实际按测试记录填写；网络中断示例中的 1006 仅表示该次客户端本地观察到异常关闭状态，不代表确定的根因。重复接收不必然是缺陷，需进一步核对去重后的业务副作用。",{"type":25,"children":435},[436],{"text":437,"type":29},"验收表包含断线类型、恢复耗时、最终缺失条数、重复接收数、重复业务副作用、状态一致性和证据位置。覆盖次数与阈值按业务 SLA、风险和测量波动事先确定；3 次或 5 次只能作为试跑示例，不能当成通用充分门槛。正常 `close(1000)` 应按产品预定策略验证停止或恢复，不把不自动重连直接判错。任一关键证据缺失，结论应标注“证据不足”或“观察到重连成功”，不能写“恢复完成”。",{"type":25,"children":439},[440,442,445],{"text":441,"type":29},"测试工具的接入参数可对照",{"text":180,"type":29,"marks":443},[444],{"href":183,"type":96},{"text":185,"type":29},{"type":25,"children":447},[448,450,455],{"text":449,"type":29},"WebSocket 关闭状态和 Close 帧语义见 ",{"text":451,"type":29,"marks":452},"RFC 6455",[453],{"href":454,"type":96},"https:\u002F\u002Fwww.rfc-editor.org\u002Frfc\u002Frfc6455",{"text":185,"type":29},{"type":31,"level":32,"children":457},[458],{"text":216,"type":29},{"type":25,"children":460},[461],{"text":462,"type":29,"marks":463},"Q：重连成功但消息少了一段，问题出在哪？",[464],{"type":223},{"type":25,"children":466},[467],{"text":468,"type":29},"A：先对照客户端、服务端和网络事件确定可观测到的故障位置，再按协议检查恢复位置。恢复状态可能来自客户端凭据，也可能由服务端保存；没有客户端序号不能单独证明服务端无法续传。",{"type":25,"children":470},[471],{"text":472,"type":29,"marks":473},"Q：close code 1006 一定能复现吗？",[474],{"type":223},{"type":25,"children":476},[477],{"text":478,"type":29},"A：不保证。1006 是本地报告的异常关闭状态，表示没有收到有效 Close 帧；不同客户端实现和故障方式下的观测结果可能不同，需记录具体故障注入和客户端事件。",{"type":25,"children":480},[481],{"text":482,"type":29,"marks":483},"Q：重复消息一定要在客户端去重吗？",[484],{"type":223},{"type":25,"children":486},[487],{"text":488,"type":29},"A：取决于协议契约。服务端游标续传、时间窗重放和客户端去重的责任不同，应分别检查收到重复与重复业务副作用，不能只统计客户端收到的消息数。",{"source":6,"id":490,"documentId":491,"translationKey":491,"locale":9,"slug":492,"path":493,"_path":493,"title":494,"description":495,"excerpt":495,"image":496,"publishedAt":497,"updatedAt":498,"author":17,"tags":499,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":500,"bodyHtml":248},3936,"wiumgl5vcc6qclqx4xe26751","websocket-proxy-test-report-metrics","\u002Fblog\u002Fwebsocket-proxy-test-report-metrics","WebSocket 代理IP 测试报告应该记录哪些指标","一份只能写「连接成功、延迟正常」的 WebSocket 测试报告，问题不在结论而在字段。把报告拆成环境、连接、业务三层，逐项给出字段、单位、采集方式与用途，配合握手分段、断连码分布、重连次数与 RTT 分位数，形成可直接落表、可复核的模板。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_Web_Socket_cover_bc45bc1e13.webp","2026-11-04T06:50:00.175Z","2026-11-04T06:50:00.160Z",[],[501,505,509,523,527,531,535,553,557,570,574,596,600,604,626,630,634,638,672,676,680,684,702,706,710,714,718,727,730,736,740,746,750,756],{"type":25,"children":502},[503],{"text":504,"type":29},"一份只能写\"连接成功、延迟正常\"的 WebSocket 报告，问题在于缺少可复核字段。本文按环境、连接、业务三层列出报告项，并注明采集工具与可观测边界。浏览器标准 WebSocket 事件通常不能直接提供全部握手分段、原始响应头或每一跳耗时；拿不到的字段应标为 N\u002FA 并说明原因。",{"type":31,"level":32,"children":506},[507],{"text":508,"type":29},"WebSocket 代理测试报告应该分哪三层记录",{"type":37,"format":38,"children":510},[511,515,519],{"type":41,"children":512},[513],{"text":514,"type":29},"环境层：这次测试在什么条件下发生。地区、出口 IP 与 ASN、代理协议、客户端版本、测试时间窗。",{"type":41,"children":516},[517],{"text":518,"type":29},"连接层：连接本身发生了什么。握手各段耗时、会话寿命、关闭码、异常关闭原因。",{"type":41,"children":520},[521],{"text":522,"type":29},"业务层：连接之上传了什么。消息计数、往返时延、丢失与重复、顺序连续性、恢复耗时。",{"type":25,"children":524},[525],{"text":526,"type":29},"三层分开写的好处是定位快：业务层异常先看连接层是否断过，连接层异常再看环境层是否出口漂移。混在一起写，读者只能看到\"整体不好\"。",{"type":31,"level":32,"children":528},[529],{"text":530,"type":29},"握手和建连阶段要记录哪些字段",{"type":25,"children":532},[533],{"text":534,"type":29},"握手要分四段记录，不能只写总和：",{"type":37,"format":38,"children":536},[537,541,545,549],{"type":41,"children":538},[539],{"text":540,"type":29},"DNS 解析耗时（ms）。",{"type":41,"children":542},[543],{"text":544,"type":29},"TCP 建连耗时（ms）。",{"type":41,"children":546},[547],{"text":548,"type":29},"TLS 握手耗时（ms，wss 才有）。",{"type":41,"children":550},[551],{"text":552,"type":29},"Upgrade 往返耗时（ms）：请求发出到收到 `101 Switching Protocols`。",{"type":25,"children":554},[555],{"text":556,"type":29},"同时记录校验结果：`Sec-WebSocket-Accept` 是否与服务端计算值一致、协商的子协议与扩展是否与预期相同、响应头是否被中间设备改写。握手失败要记录失败阶段与错误码，而不是只记\"连接失败\"。",{"type":25,"children":558},[559,561,564,566,569],{"text":560,"type":29},"握手前的代理参数可对照",{"text":180,"type":29,"marks":562},[563],{"href":183,"type":96},{"text":565,"type":29},"；Upgrade 握手字段定义见 ",{"text":451,"type":29,"marks":567},[568],{"href":454,"type":96},{"text":185,"type":29},{"type":31,"level":32,"children":571},[572],{"text":573,"type":29},"连接持续多久、为什么断开要怎么记录",{"type":37,"format":38,"children":575},[576,580,584,588,592],{"type":41,"children":577},[578],{"text":579,"type":29},"会话观察时长（s）：从收到 101 到关闭或观察窗口结束；窗口结束仍在线单独计数，不当作自然断开。",{"type":41,"children":581},[582],{"text":583,"type":29},"关闭发起方：只有两端事件或网络日志能佐证时才填写；无法确认则记未知。",{"type":41,"children":585},[586],{"text":587,"type":29},"本地关闭事件码与实际 Close 帧分开记录。1006 是本地报告的异常关闭状态，不是线上发送的 Close 帧；正常码也不能单独证明业务正常。",{"type":41,"children":589},[590],{"text":591,"type":29},"异常关闭率：明确分子、分母和观察窗口；握手失败、已关闭连接与窗口结束仍在线的连接分别统计。",{"type":41,"children":593},[594],{"text":595,"type":29},"心跳配置：间隔与超时阈值（s），以及心跳是否被中间设备吞掉。",{"type":25,"children":597},[598],{"text":599,"type":29},"关闭事件码分布可帮助筛选样本，但不能仅凭 code 判定发起方或根因。1006 表示本地未观察到有效 Close 帧；需结合客户端、服务端或网络侧日志继续定位。",{"type":31,"level":32,"children":601},[602],{"text":603,"type":29},"重连过程应该记录哪些指标",{"type":37,"format":38,"children":605},[606,610,614,618,622],{"type":41,"children":607},[608],{"text":609,"type":29},"首次重连延迟（ms\u002Fs）：从检测到断线到发起重连。",{"type":41,"children":611},[612],{"text":613,"type":29},"重连次数与退避序列：是否按指数退避，是否有抖动。",{"type":41,"children":615},[616],{"text":617,"type":29},"重连成功率：成功建立新连接的比例。",{"type":41,"children":619},[620],{"text":621,"type":29},"恢复耗时：从断线到业务可用的时间。",{"type":41,"children":623},[624],{"text":625,"type":29},"是否携带恢复凭据：如 resume token、last sequence。",{"type":25,"children":627},[628],{"text":629,"type":29},"重连次数多不代表质量差，可能是服务端主动轮转；但没有退避、密集重连会把故障放大。报告应同时给次数与间隔分布。",{"type":31,"level":32,"children":631},[632],{"text":633,"type":29},"如何汇总三层字段并补充消息细节",{"type":25,"children":635},[636],{"text":637,"type":29},"下表把三层字段汇总成可直接落表的形式。",{"type":37,"format":38,"children":639},[640,644,648,652,656,660,664,668],{"type":41,"children":641},[642],{"text":643,"type":29},"字段组：握手 ｜ 字段：DNS \u002F TCP \u002F TLS \u002F Upgrade ｜ 单位：ms ｜ 采集方式：受控网络日志、协议工具或服务端\u002F代理日志；浏览器标准事件通常不可提供全部分段 ｜ 用途：定位可观测范围内的建连耗时",{"type":41,"children":645},[646],{"text":647,"type":29},"字段组：连接 ｜ 字段：观察时长、结束状态、仍在线数 ｜ 单位：s \u002F 次 ｜ 采集方式：连接事件与观察窗口 ｜ 用途：区分自然断开与测试截止",{"type":41,"children":649},[650],{"text":651,"type":29},"字段组：连接 ｜ 字段：本地关闭事件码、Close 帧收发、可证实的发起方 ｜ 单位：次数 ｜ 采集方式：客户端事件及可用两端日志；不可观测时记 N\u002FA ｜ 用途：作为定位线索，不单独归因",{"type":41,"children":653},[654],{"text":655,"type":29},"字段组：重连 ｜ 字段：重连次数与间隔 ｜ 单位：次 \u002F s ｜ 采集方式：重连逻辑埋点 ｜ 用途：评估退避是否合理",{"type":41,"children":657},[658],{"text":659,"type":29},"字段组：消息 ｜ 字段：收发计数 ｜ 单位：条 ｜ 采集方式：消息回调 ｜ 用途：核对是否丢消息",{"type":41,"children":661},[662],{"text":663,"type":29},"字段组：消息 ｜ 字段：RTT p50 \u002F p95 ｜ 单位：ms ｜ 采集方式：发送与回执时间戳 ｜ 用途：评估体感延迟",{"type":41,"children":665},[666],{"text":667,"type":29},"字段组：消息 ｜ 字段：序号缺口与重复数 ｜ 单位：条 ｜ 采集方式：序号连续性校验 ｜ 用途：判断顺序与去重",{"type":41,"children":669},[670],{"text":671,"type":29},"字段组：代理IP ｜ 字段：类型、出口和会话分配 ｜ 单位：枚举 \u002F 地址 \u002F ID ｜ 采集方式：供应商控制台、分配记录及受控回显 ｜ 用途：说明本次测试路径，不能单独证明物理路由",{"type":25,"children":673},[674],{"text":675,"type":29},"表外还要补两项：消息大小分布与发送频率，否则 RTT 无法横向比较。大消息和小消息的往返时间差异明显，混在一起算分位数会失真。",{"type":31,"level":32,"children":677},[678],{"text":679,"type":29},"地区和代理链路还要记录哪些信息",{"type":25,"children":681},[682],{"text":683,"type":29},"代理IP字段记录实际分配的产品类型、出口和会话变化；无法确认的属性应标为未知。",{"type":37,"format":38,"children":685},[686,690,694,698],{"type":41,"children":687},[688],{"text":689,"type":29},"地区：客户端所在地区，与出口地区分开写。",{"type":41,"children":691},[692],{"text":693,"type":29},"出口 IP 与 ASN：每条连接记录一次，用于验证地区标签。",{"type":41,"children":695},[696],{"text":697,"type":29},"代理协议与会话类型：HTTP、SOCKS5，短会话或长会话；品类按实际分配写动态或静态住宅 IP、动态或静态机房 IP，不写笼统的\"住宅或机房\"，并保留出口 IP 与分配时间。",{"type":41,"children":699},[700],{"text":701,"type":29},"链路跳数：若代理有多跳，记录跳数与各跳耗时。",{"type":25,"children":703},[704],{"text":705,"type":29},"出口 IP 是测试记录的重要字段，但需与服务端回显、代理分配记录和 DNS\u002F调度信息结合；它本身不能证明物理路径或目标服务调度。",{"type":31,"level":32,"children":707},[708],{"text":709,"type":29},"报告模板应该怎样组织和统计",{"type":25,"children":711},[712],{"text":713,"type":29},"模板按\"环境 → 连接 → 业务\"顺序排列，每部分先给汇总表，再列异常样本。统计时报告 p50、p95 和样本数，单次最快值不能代表整体表现。\n失败与超时单独计数，不混入成功样本的分位数。\n按配置地区、出口和时间窗分组；跨窗口比较时同时标明环境差异。",{"type":25,"children":715},[716],{"text":717,"type":29},"异常样本保留 request id、连接 ID、出口 IP、关闭事件、时间戳及可用证据。报告结论写明客户端版本、服务端环境、测试时长、采集工具和不可观测字段；详细握手数据需受控网络日志、协议工具或服务端\u002F代理日志支持。",{"type":25,"children":719},[720,722,725],{"text":721,"type":29},"连接异常样本可按",{"text":305,"type":29,"marks":723},[724],{"href":308,"type":96},{"text":726,"type":29},"的字段顺序补充。",{"type":31,"level":32,"children":728},[729],{"text":216,"type":29},{"type":25,"children":731},[732],{"text":733,"type":29,"marks":734},"Q：报告里必须记录出口 IP 吗？",[735],{"type":223},{"type":25,"children":737},[738],{"text":739,"type":29},"A：建议记录，并同时记代理IP类型。它是验证地区标签与判断出口漂移的重要直接证据之一，但要与服务端回显、代理记录一起看，缺失后地区结论只能靠推断。",{"type":25,"children":741},[742],{"text":743,"type":29,"marks":744},"Q：close code 只记 1006 够不够？",[745],{"type":223},{"type":25,"children":747},[748],{"text":749,"type":29},"A：不够。关闭码是协议层线索，不能单独判断发起方或根因。应同时记录本地事件、实际 Close 帧收发和可取得的两端日志，并明确异常关闭率的分子、分母和仍在线连接的处理方式。",{"type":25,"children":751},[752],{"text":753,"type":29,"marks":754},"Q：时延只报平均值可以吗？",[755],{"type":223},{"type":25,"children":757},[758],{"text":759,"type":29},"A：不建议。平均值会掩盖尾部延迟，应同时报 p50 与 p95，并注明样本量。",{"source":6,"id":761,"documentId":762,"translationKey":762,"locale":9,"slug":763,"path":764,"_path":764,"title":765,"description":766,"excerpt":766,"image":767,"publishedAt":768,"updatedAt":769,"author":17,"tags":770,"featured":19,"sortOrder":20,"readingTime":771,"bodyType":22,"bodyBlocks":772,"bodyHtml":248},3935,"apg9xmkn8f0bozwgyxntuhtj","python-test-proxy-environment-variable-isolation","\u002Fblog\u002Fpython-test-proxy-environment-variable-isolation","Python 自动化测试：代理IP 环境变量污染如何验证","同一份测试在开发机能过、到 CI 就失败，常常是残留的代理变量在捣乱。文章列出这些变量及其读取顺序，用 fixture 固定环境、请求前断言、回显出口 IP 证明链路，再给出清理检查与故意失败自检。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_Python_cover_58829f8376.webp","2026-11-04T06:49:42.792Z","2026-11-04T06:49:42.779Z",[],4,[773,777,781,791,795,799,803,807,825,829,833,837,841,845,853,857,861,865,869,873,883,887,891,917,921,925,929,943,947,951,955,969,973,988,992,996,1000,1003,1009,1013,1019,1023,1029],{"type":25,"children":774},[775],{"text":776,"type":29},"同一份测试，在开发机上通过、在 CI 上失败，或者反过来——十有八九是环境里的代理变量在起作用。requests、urllib、httpx 等常见实现通常会读取环境变量，具体行为随库和版本而变；CI 镜像又常预置 `HTTP_PROXY`。判断\"有没有被污染\"不能靠肉眼扫代码，要靠在测试内部固定环境并断言出口。这些变量指向的代理IP，往往和用例约定的不是同一条。NO_PROXY 规则本身的行为验证属于另一个话题，本篇讲的是测试框架里的环境隔离。",{"type":31,"level":32,"children":778},[779],{"text":780,"type":29},"代理环境变量污染通常会表现为什么",{"type":37,"format":38,"children":782},[783,787],{"type":41,"children":784},[785],{"text":786,"type":29},"意外走代理：本应直连的请求，被环境变量送进了代理，出口 IP 与预期不符。",{"type":41,"children":788},[789],{"text":790,"type":29},"意外绕过代理：本应走代理的请求，被 `NO_PROXY` 或空值放行，直接连到了目标。",{"type":25,"children":792},[793],{"text":794,"type":29},"两种表现都会让断言\"通过得莫名其妙\"。前者常出现在本地缓存或 mock 场景，后者常出现在 CI 里变量拼写错误或作用域不对的场景。判定污染不看代码写了什么，看实际请求从哪里出去。",{"type":31,"level":32,"children":796},[797],{"text":798,"type":29},"哪些代理环境变量会生效，优先级如何判断",{"type":25,"children":800},[801],{"text":802,"type":29},"变量不是设了就生效，读取顺序、大小写和库的默认值都会改变结果；变量只描述地址和协议，不描述代理IP的身份，身份得另做出口校验。把\"默认\"当成待实测的假设：按库和版本分别验证，并补一个故意设错变量的负向用例。",{"type":31,"level":21,"children":804},[805],{"text":806,"type":29},"为什么要同时检查大小写变量和 ALL_PROXY",{"type":37,"format":38,"children":808},[809,813,817,821],{"type":41,"children":810},[811],{"text":812,"type":29},"变量：`HTTP_PROXY` \u002F `http_proxy` ｜ 作用范围：http 请求 ｜ Python 客户端是否读取：requests、urllib 默认读取 ｜ 测试中如何处理：fixture 内显式设置或清空",{"type":41,"children":814},[815],{"text":816,"type":29},"变量：`HTTPS_PROXY` \u002F `https_proxy` ｜ 作用范围：https 请求 ｜ Python 客户端是否读取：同上 ｜ 测试中如何处理：同上",{"type":41,"children":818},[819],{"text":820,"type":29},"变量：`ALL_PROXY` \u002F `all_proxy` ｜ 作用范围：未单独指定协议时 ｜ Python 客户端是否读取：requests 读取 ｜ 测试中如何处理：容易遗漏，需一并清理",{"type":41,"children":822},[823],{"text":824,"type":29},"变量：`NO_PROXY` \u002F `no_proxy` ｜ 作用范围：排除列表 ｜ Python 客户端是否读取：requests、urllib 读取 ｜ 测试中如何处理：边界用例单独覆盖",{"type":25,"children":826},[827],{"text":828,"type":29},"`ALL_PROXY` 最容易被漏：未单独指定协议时它接管全部请求，清理时不能只清 HTTP_PROXY 与 HTTPS_PROXY。",{"type":31,"level":21,"children":830},[831],{"text":832,"type":29},"环境变量大小写相同时应该相信哪一个",{"type":25,"children":834},[835],{"text":836,"type":29},"同名不同大小写时，取值顺序可能相反。requests 通过 `urllib.request.getproxies()` 取值，常见实现小写优先，但版本和平台不同结果会变，也有库只认大写。测试里不要依赖，统一由 fixture 显式设定，并在请求前断言生效值，再用实际出口校验。",{"type":31,"level":21,"children":838},[839],{"text":840,"type":29},"为什么没写代理的 requests 仍可能走代理",{"type":25,"children":842},[843],{"text":844,"type":29},"Python 的 `requests` 常见版本默认 `trust_env=True`，会读取环境配置；该开关还会影响 `.netrc` 等其他来源。需要隔离环境代理时，优先为测试创建专用 `Session` 并设置 `trust_env=False`，再显式配置测试所需的认证与 CA。这样也会关闭该 Session 的其他环境来源，需按实际需求补配。不要把只对 `http`、`https` 设置 `None` 当成覆盖所有代理环境变量的通用方法；应按 requests 版本、目标地址和实际出口验证。",{"type":25,"children":846},[847,849,852],{"text":848,"type":29},"代理环境变量与请求库的关系可对照",{"text":180,"type":29,"marks":850},[851],{"href":183,"type":96},{"text":185,"type":29},{"type":31,"level":32,"children":854},[855],{"text":856,"type":29},"如何用 fixture 固定环境并提前断言",{"type":25,"children":858},[859],{"text":860,"type":29},"在测试层做三件事：",{"type":25,"children":862},[863],{"text":864,"type":29},"1. autouse fixture 在用例开始前清空所有代理变量（大小写各一份），用例结束后恢复。\n2. 需要代理的用例通过参数显式注入，不依赖外部环境。\n3. 请求发出前断言：`os.environ` 中的值、`session.proxies`、`session.trust_env` 与预期一致。",{"type":25,"children":866},[867],{"text":868,"type":29},"```python\nimport pytest",{"type":25,"children":870},[871],{"text":872,"type":29},"@pytest.fixture(autouse=True)\ndef clean_proxy_env(monkeypatch):\nfor key in (\"HTTP_PROXY\", \"HTTPS_PROXY\", \"ALL_PROXY\",\n\"NO_PROXY\", \"http_proxy\", \"https_proxy\", \"all_proxy\",\n\"no_proxy\"):\nmonkeypatch.delenv(key, raising=False)\n```",{"type":25,"children":874},[875,877,882],{"text":876,"type":29},"需要测试 `NO_PROXY` 时，在对应参数化用例中单独设置目标值。`monkeypatch` 会在用例结束时恢复原环境；可加一个检查，确认所有大小写变量恢复到用例前状态。详情见 ",{"text":878,"type":29,"marks":879},"pytest monkeypatch 文档",[880],{"href":881,"type":96},"https:\u002F\u002Fdocs.pytest.org\u002Fen\u002Fstable\u002Fhow-to\u002Fmonkeypatch.html",{"text":185,"type":29},{"type":31,"level":32,"children":884},[885],{"text":886,"type":29},"NO_PROXY 哪些边界情况必须单独测试",{"type":25,"children":888},[889],{"text":890,"type":29},"`NO_PROXY` 的匹配规则容易想当然，至少要覆盖：",{"type":37,"format":38,"children":892},[893,897,901,905,909,913],{"type":41,"children":894},[895],{"text":896,"type":29},"精确域名与子域名：`example.com` 是否覆盖 `api.example.com`。",{"type":41,"children":898},[899],{"text":900,"type":29},"端口写法：`example.com:443` 与不带端口的差异。",{"type":41,"children":902},[903],{"text":904,"type":29},"IP 与 CIDR：单 IP、网段写法是否被支持。",{"type":41,"children":906},[907],{"text":908,"type":29},"本机地址：`localhost`、`127.0.0.1`、`::1`。",{"type":41,"children":910},[911],{"text":912,"type":29},"通配：`*` 是否让所有请求直连。",{"type":41,"children":914},[915],{"text":916,"type":29},"空值：变量存在但为空字符串时的行为。",{"type":25,"children":918},[919],{"text":920,"type":29},"每个边界写成一个用例，断言\"是否走了代理\"，而不是\"变量是否被读取\"。变量被读取但匹配失败，同样属于污染。",{"type":31,"level":32,"children":922},[923],{"text":924,"type":29},"如何证明请求真的走了预期出口",{"type":25,"children":926},[927],{"text":928,"type":29},"环境变量只是配置，出口才是结果。回显请求必须由同一个被测客户端、协议和目标匹配规则发出；另用受控目标的访问日志或代理到达记录核对真正的业务请求。不同目标可能命中不同的 `NO_PROXY` 规则，单独请求另一个回显域名不能证明被测请求走了同一路径。直连出口应现场测量，并注明 NAT 或 CI 共享出口的影响：",{"type":37,"format":38,"children":930},[931,935,939],{"type":41,"children":932},[933],{"text":934,"type":29},"期望走代理：出口应属于代理侧地址段。",{"type":41,"children":936},[937],{"text":938,"type":29},"期望直连：出口应属于本机公网地址段。",{"type":41,"children":940},[941],{"text":942,"type":29},"期望绕过：目标应命中 `NO_PROXY`，出口为直连地址。",{"type":25,"children":944},[945],{"text":946,"type":29},"回显服务用自建或受控端点。分别覆盖预期走代理、命中绕过规则、故意设置错误配置和恢复后的请求，并记录目标、时间、客户端版本与代理到达证据。代理 URL 只记录脱敏后的协议、主机、端口和配置来源；删除 userinfo、令牌及敏感查询参数，避免 CI 日志泄露凭据。",{"type":31,"level":32,"children":948},[949],{"text":950,"type":29},"如何隔离开发机和 CI 继承的代理环境",{"type":25,"children":952},[953],{"text":954,"type":29},"本地 shell 里的代理变量会通过 `pytest` 进程继承进来，CI 的容器镜像也可能预置。隔离手段：",{"type":37,"format":38,"children":956},[957,961,965],{"type":41,"children":958},[959],{"text":960,"type":29},"在 `pytest.ini` 或 `conftest.py` 里统一清理，不依赖开发者本地配置。",{"type":41,"children":962},[963],{"text":964,"type":29},"CI 任务显式声明环境，不继承 runner 的全局变量。",{"type":41,"children":966},[967],{"text":968,"type":29},"需要真实代理的集成测试单独标记，默认不跑。",{"type":25,"children":970},[971],{"text":972,"type":29},"区分\"单元测试\"与\"集成测试\"的边界很重要：前者不应依赖任何外部出口，后者才需要真实代理。",{"type":25,"children":974},[975,977,982,984,987],{"text":976,"type":29},"requests 的环境设置行为见 ",{"text":978,"type":29,"marks":979},"Requests 高级用法文档",[980],{"href":981,"type":96},"https:\u002F\u002Frequests.readthedocs.io\u002Fen\u002Fstable\u002Fuser\u002Fadvanced\u002F",{"text":983,"type":29},"；测试环境变量恢复可参考 ",{"text":878,"type":29,"marks":985},[986],{"href":881,"type":96},{"text":185,"type":29},{"type":31,"level":32,"children":989},[990],{"text":991,"type":29},"如何通过清理检查和故意失败验证测试有效",{"type":25,"children":993},[994],{"text":995,"type":29},"最后一步是验证测试本身有效：把预期出口故意写错，测试必须失败。如果仍然通过，说明断言没有真正落到出口上，前面所有隔离都只是形式。",{"type":25,"children":997},[998],{"text":999,"type":29},"清理检查同样要做：用例结束后确认 `os.environ` 已恢复，没有把变量泄漏给后续用例。多个用例串行时，泄漏会表现为\"只有第一个用例正常\"。",{"type":31,"level":32,"children":1001},[1002],{"text":216,"type":29},{"type":25,"children":1004},[1005],{"text":1006,"type":29,"marks":1007},"Q：代码里没写代理，还会被环境变量影响吗？",[1008],{"type":223},{"type":25,"children":1010},[1011],{"text":1012,"type":29},"A：会。requests 等常见实现默认读取环境变量，`trust_env` 为 True 时自动生效；具体行为以实际版本为准，测试里要显式清理或关闭，并回显出口验证。",{"type":25,"children":1014},[1015],{"text":1016,"type":29,"marks":1017},"Q：`NO_PROXY` 写了域名，为什么子域名还是走了代理？",[1018],{"type":223},{"type":25,"children":1020},[1021],{"text":1022,"type":29},"A：不同实现对子域名的匹配规则不同，需要按库的实际行为写用例验证，不能假设后缀匹配一定生效。",{"type":25,"children":1024},[1025],{"text":1026,"type":29,"marks":1027},"Q：CI 上通过、本地失败，先查什么？",[1028],{"type":223},{"type":25,"children":1030},[1031],{"text":1032,"type":29},"A：先比对变量是否存在、脱敏后的协议\u002F主机\u002F端口和配置来源，再用同一被测请求的目标日志或代理到达记录确认路径。不要打印完整代理 URL，其中可能含账号、密码或令牌。",[1034],{"source":6,"id":7,"documentId":8,"translationKey":8,"locale":9,"slug":10,"path":11,"_path":11,"title":12,"description":13,"excerpt":13,"image":14,"publishedAt":15,"updatedAt":16,"author":17,"tags":1035,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":1036,"bodyHtml":248},[],[1037,1040,1043,1054,1057,1060,1063,1066,1077,1080,1087,1090,1093,1107,1110,1113,1116,1119,1122,1125,1128,1142,1145,1152,1155,1162,1165,1168,1171,1174,1179,1182,1187,1190,1195],{"type":25,"children":1038},[1039],{"text":28,"type":29},{"type":31,"level":32,"children":1041},[1042],{"text":35,"type":29},{"type":37,"format":38,"children":1044},[1045,1048,1051],{"type":41,"children":1046},[1047],{"text":44,"type":29},{"type":41,"children":1049},[1050],{"text":48,"type":29},{"type":41,"children":1052},[1053],{"text":52,"type":29},{"type":25,"children":1055},[1056],{"text":56,"type":29},{"type":31,"level":32,"children":1058},[1059],{"text":60,"type":29},{"type":25,"children":1061},[1062],{"text":64,"type":29},{"type":31,"level":21,"children":1064},[1065],{"text":68,"type":29},{"type":37,"format":38,"children":1067},[1068,1071,1074],{"type":41,"children":1069},[1070],{"text":74,"type":29},{"type":41,"children":1072},[1073],{"text":78,"type":29},{"type":41,"children":1075},[1076],{"text":82,"type":29},{"type":31,"level":21,"children":1078},[1079],{"text":86,"type":29},{"type":25,"children":1081},[1082,1083,1086],{"text":90,"type":29},{"text":92,"type":29,"marks":1084},[1085],{"href":95,"type":96},{"text":98,"type":29},{"type":31,"level":21,"children":1088},[1089],{"text":102,"type":29},{"type":25,"children":1091},[1092],{"text":106,"type":29},{"type":37,"format":38,"children":1094},[1095,1098,1101,1104],{"type":41,"children":1096},[1097],{"text":112,"type":29},{"type":41,"children":1099},[1100],{"text":116,"type":29},{"type":41,"children":1102},[1103],{"text":120,"type":29},{"type":41,"children":1105},[1106],{"text":124,"type":29},{"type":25,"children":1108},[1109],{"text":128,"type":29},{"type":25,"children":1111},[1112],{"text":132,"type":29},{"type":31,"level":32,"children":1114},[1115],{"text":136,"type":29},{"type":25,"children":1117},[1118],{"text":140,"type":29},{"type":25,"children":1120},[1121],{"text":144,"type":29},{"type":31,"level":32,"children":1123},[1124],{"text":148,"type":29},{"type":25,"children":1126},[1127],{"text":152,"type":29},{"type":37,"format":38,"children":1129},[1130,1133,1136,1139],{"type":41,"children":1131},[1132],{"text":158,"type":29},{"type":41,"children":1134},[1135],{"text":162,"type":29},{"type":41,"children":1137},[1138],{"text":166,"type":29},{"type":41,"children":1140},[1141],{"text":170,"type":29},{"type":25,"children":1143},[1144],{"text":174,"type":29},{"type":25,"children":1146},[1147,1148,1151],{"text":178,"type":29},{"text":180,"type":29,"marks":1149},[1150],{"href":183,"type":96},{"text":185,"type":29},{"type":31,"level":32,"children":1153},[1154],{"text":189,"type":29},{"type":25,"children":1156},[1157,1158,1161],{"text":193,"type":29},{"text":195,"type":29,"marks":1159},[1160],{"href":198,"type":96},{"text":200,"type":29},{"type":31,"level":32,"children":1163},[1164],{"text":204,"type":29},{"type":25,"children":1166},[1167],{"text":208,"type":29},{"type":25,"children":1169},[1170],{"text":212,"type":29},{"type":31,"level":32,"children":1172},[1173],{"text":216,"type":29},{"type":25,"children":1175},[1176],{"text":220,"type":29,"marks":1177},[1178],{"type":223},{"type":25,"children":1180},[1181],{"text":227,"type":29},{"type":25,"children":1183},[1184],{"text":231,"type":29,"marks":1185},[1186],{"type":223},{"type":25,"children":1188},[1189],{"text":237,"type":29},{"type":25,"children":1191},[1192],{"text":241,"type":29,"marks":1193},[1194],{"type":223},{"type":25,"children":1196},[1197],{"text":247,"type":29},null,[],[]]