[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"blog-strapi-article:zh:websocket-quality-multi-region-comparison":3,"blog-article:\u002Fblog\u002Fwebsocket-quality-multi-region-comparison":1261,"blog-related:\u002Fblog\u002Fwebsocket-quality-multi-region-comparison":1262,"blog-translations:\u002Fblog\u002Fwebsocket-quality-multi-region-comparison":1263},{"available":4,"article":5,"relatedArticles":284,"translations":1068},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":283},"strapi",3931,"g1ic9kqs54nyrhx5e64bm66a","zh","websocket-quality-multi-region-comparison","\u002Fblog\u002Fwebsocket-quality-multi-region-comparison","WebSocket 多地区会话质量对比：代理IP 怎么测","同一款 WebSocket 应用换到不同网络地区，质量差异常被「感觉卡」掩盖。文章把会话质量拆成握手各段、连接寿命、断连码分布、消息 RTT 与重连次数六项可测指标，把地区作为主要自变量、其余变量逐项记录并尽量匹配后分段采集，再用样本量和分位数比较，结论只到证据支持处。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_Web_Socket_cover_51e3f60377.webp","2026-11-04T06:48:34.161Z","2026-11-04T06:48:34.147Z","Global Proxies Team",[],false,0,3,"blocks",[24,30,36,65,69,73,77,99,103,107,111,115,119,123,127,131,135,149,153,165,169,173,191,195,199,203,207,221,225,236,240,244,248,252,259,263,269,273,279],{"type":25,"children":26},"paragraph",[27],{"text":28,"type":29},"不同地区的 WebSocket 体验差异，常常被一次\"握手 80ms\"的测试概化掉。实际上握手快不等于会话稳，会话稳也不等于消息及时。这篇讲的是横向对照方法：怎么把\"质量\"拆成可测项、怎么固定变量、怎么用样本而不是个案下结论。字段如何记录、断线后如何验证恢复，分别是另外两篇的主题，这里只关心地区之间的可比性。地区只是表象，决定会话质量的是这次连接落到了哪一类代理IP、哪个出口。","text",{"type":31,"level":32,"children":33},"heading",2,[34],{"text":35,"type":29},"比较不同地区时应该测哪些会话指标",{"type":37,"format":38,"children":39},"list","unordered",[40,45,49,53,57,61],{"type":41,"children":42},"list-item",[43],{"text":44,"type":29},"建连成功率：发起连接中成功完成 Upgrade 的比例。",{"type":41,"children":46},[47],{"text":48,"type":29},"握手耗时：TCP 建连、TLS 握手、Upgrade 请求到收到 101 三段之和。",{"type":41,"children":50},[51],{"text":52,"type":29},"首消息时延：101 之后收到首条业务消息的时间。",{"type":41,"children":54},[55],{"text":56,"type":29},"稳态消息 RTT：会话中收发消息的往返时间，取 p50 与 p95。",{"type":41,"children":58},[59],{"text":60,"type":29},"会话观察时长、结束状态及异常关闭情况；观察窗口结束仍在线的连接单独记录。",{"type":41,"children":62},[63],{"text":64,"type":29},"恢复耗时：断线后重新建立可用会话并补齐状态所需时间。",{"type":25,"children":66},[67],{"text":68,"type":29},"六项分别对应\"能不能连上、连得多快、消息多快、能撑多久、断了多久能回来\"。只报其中一两项，结论容易被单一指标带偏：某地区可能握手较快，但消息延迟尾部或关闭情况不同。",{"type":31,"level":32,"children":70},[71],{"text":72,"type":29},"如何固定变量，确保比较结果只反映地区差异",{"type":25,"children":74},[75],{"text":76,"type":29},"地区是主要自变量，其余变量要逐项记录并尽量匹配：",{"type":37,"format":38,"children":78},[79,83,87,91,95],{"type":41,"children":80},[81],{"text":82,"type":29},"同一客户端版本、同一操作系统与网络类型。",{"type":41,"children":84},[85],{"text":86,"type":29},"同一服务端环境与同一业务接口，消息大小和发送频率固定。",{"type":41,"children":88},[89],{"text":90,"type":29},"同一并发数、同一测试时长、同一采样窗口。",{"type":41,"children":92},[93],{"text":94,"type":29},"同一鉴权方式与等价权限；只读场景且确认服务支持并发时可共用账号，涉及会话或状态变化时，各地区使用独立的等价测试账号并记录账号策略。",{"type":41,"children":96},[97],{"text":98,"type":29},"代理IP品类与出口 ASN 尽量匹配；无法匹配时逐行写明，结论按品类分开写。",{"type":25,"children":100},[101],{"text":102,"type":29},"时间也要控制：不同时段的服务端负载不同。尽量在相同时间窗轮转测试，随机安排地区顺序；跨天比较时单独记录服务端版本和负载线索，避免把调度变化误认为地区差异。",{"type":31,"level":32,"children":104},[105],{"text":106,"type":29},"如何按连接生命周期采集数据",{"type":25,"children":108},[109],{"text":110,"type":29},"分段只为归因，不必每段单独出报告。把一次会话切成四段分别记录，才能在结果异常时定位：",{"type":31,"level":21,"children":112},[113],{"text":114,"type":29},"建连过程应该拆成哪四段",{"type":25,"children":116},[117],{"text":118,"type":29},"1. TCP 建连（SYN 到 SYN-ACK）。\n2. TLS 握手（若使用 wss）。\n3. Upgrade（请求发出到收到 101，含 `Sec-WebSocket-Accept` 校验）。\n4. 稳态（首条消息、消息往返、心跳、关闭）。",{"type":25,"children":120},[121],{"text":122,"type":29},"Upgrade 段要额外记 `Sec-WebSocket-Accept` 校验是否通过——101 到了不等于握手成立。",{"type":31,"level":21,"children":124},[125],{"text":126,"type":29},"不同阶段变慢分别说明什么",{"type":25,"children":128},[129],{"text":130,"type":29},"分段结果的用途是提出假设，不是直接下结论：握手三段都慢，要结合代理日志与 DNS 解析看是不是链路；只有 Upgrade 慢，要对照服务端时间戳与鉴权日志；稳态 RTT 抖动大，要查中间设备是否缓冲或限速。某一段变慢只能提示去哪里找，归因要靠多源证据和重复样本。",{"type":31,"level":21,"children":132},[133],{"text":134,"type":29},"地区对照应该看分布而不是单次结果",{"type":37,"format":38,"children":136},[137,141,145],{"type":41,"children":138},[139],{"text":140,"type":29},"配置地区：A（示例配置） ｜ IP库识别地区\u002F版本：库识别结果\u002F版本待填 ｜ 目标调度信息：按目标服务日志填写 ｜ 出口 IP \u002F ASN：示例出口 \u002F ASN ｜ 握手各段：按工具记录 ｜ 观察时长\u002F结束状态：观察窗口结束仍在线 ｜ 消息 RTT p50 \u002F p95：示例数据 ｜ 重连次数：试跑",{"type":41,"children":142},[143],{"text":144,"type":29},"配置地区：B（示例配置） ｜ IP库识别地区\u002F版本：库识别结果\u002F版本待填 ｜ 目标调度信息：按目标服务日志填写 ｜ 出口 IP \u002F ASN：示例出口 \u002F ASN ｜ 握手各段：按工具记录 ｜ 观察时长\u002F结束状态：自然断开，原因待证据确认 ｜ 消息 RTT p50 \u002F p95：示例数据 ｜ 重连次数：试跑",{"type":41,"children":146},[147],{"text":148,"type":29},"配置地区：C（示例配置） ｜ IP库识别地区\u002F版本：库识别结果\u002F版本待填 ｜ 目标调度信息：按目标服务日志填写 ｜ 出口 IP \u002F ASN：示例出口 \u002F ASN ｜ 握手各段：按工具记录 ｜ 观察时长\u002F结束状态：测试主动停止 ｜ 消息 RTT p50 \u002F p95：示例数据 ｜ 重连次数：试跑",{"type":25,"children":150},[151],{"text":152,"type":29},"示例结构不构成地区归因结论。分别记录配置地区、IP 地理库识别地区及数据库版本、目标服务实际调度信息；三者可能不同。出口 IP\u002FASN 只能描述观测到的地址与网络登记信息，不能单独证明物理路径或住宅\u002F机房属性。观察窗口结束仍在线不等于自然寿命达到窗口长度；主动停止和自然断开要分开统计。",{"type":25,"children":154},[155,157,163],{"text":156,"type":29},"连接分段指标的采集口径可参考",{"text":158,"type":29,"marks":159},"代理 IP 可用率、延迟和稳定性怎么测试",[160],{"href":161,"type":162},"https:\u002F\u002Fwww.quanqiudaili.com\u002Fblog\u002Fproxy-ip-availability-latency-stability-test","link",{"text":164,"type":29},"。",{"type":31,"level":32,"children":166},[167],{"text":168,"type":29},"为什么要用样本量和分位数比较",{"type":25,"children":170},[171],{"text":172,"type":29},"单次最快值没有比较价值，它只说明\"有一次运气好\"。最低要求：",{"type":37,"format":38,"children":174},[175,179,183,187],{"type":41,"children":176},[177],{"text":178,"type":29},"先做试跑确认采集链路，再根据要检测的差异、观测波动、成功率精度和时段覆盖确定正式样本量。",{"type":41,"children":180},[181],{"text":182,"type":29},"报告独立会话数及每个会话的消息数；同一会话内的消息不能简单视作相互独立样本。",{"type":41,"children":184},[185],{"text":186,"type":29},"对 p95 等尾部分位数说明样本量和不确定性，不以达到固定示例数直接判定充分。",{"type":41,"children":188},[189],{"text":190,"type":29},"失败样本单独统计，不混入成功样本的时间分布。",{"type":25,"children":192},[193],{"text":194,"type":29},"报告里给出 p50、p95 和最大值，并注明样本量与时间窗。p95 比平均值更能反映体感：平均值被大量快速请求拉低，慢请求集中在尾部。",{"type":31,"level":32,"children":196},[197],{"text":198,"type":29},"如何记录出口身份并识别服务端调度差异",{"type":25,"children":200},[201],{"text":202,"type":29},"同一地区换个代理IP类型，会话质量也会变，所以每行记录都要单独标注。",{"type":25,"children":204},[205],{"text":206,"type":29},"三个因素会混在一起：",{"type":37,"format":38,"children":208},[209,213,217],{"type":41,"children":210},[211],{"text":212,"type":29},"地区：客户端到服务端的物理路径。",{"type":41,"children":214},[215],{"text":216,"type":29},"出口身份：记录可确认的代理产品类型和出口网络信息；IP\u002FASN 本身不足以判定住宅、机房或移动属性。",{"type":41,"children":218},[219],{"text":220,"type":29},"服务端调度：DNS 解析到不同边缘机房，或按地域路由到不同集群。",{"type":25,"children":222},[223],{"text":224,"type":29},"出口 IP\u002FASN 只描述观测地址和网络登记信息，不能单独证明物理路径或住宅\u002F机房属性。分别记录配置地区、IP 地理库识别结果及版本、目标服务返回的机房或调度标识（若有）；这些信息可以帮助解释差异，但不能完全排除路径差异。若某地区始终解析到较远的机房，结论应写成\"该地区到该机房的质量\"，而不是\"该地区质量差\"。",{"type":25,"children":226},[227,229,234],{"text":228,"type":29},"WebSocket 建连阶段的规范定义可参考 ",{"text":230,"type":29,"marks":231},"RFC 6455",[232],{"href":233,"type":162},"https:\u002F\u002Fwww.rfc-editor.org\u002Frfc\u002Frfc6455",{"text":235,"type":29},"；详细指标采集应按所用客户端工具的能力记录。",{"type":31,"level":32,"children":237},[238],{"text":239,"type":29},"如何写清结论边界并持续跟踪",{"type":25,"children":241},[242],{"text":243,"type":29},"结论边界要写清楚：本结论只适用于当前时间窗、当前客户端版本、当前服务端环境，不能外推为\"某地区永久更好\"。地区网络质量会随时间变化，跨周比较需要重新采样。",{"type":25,"children":245},[246],{"text":247,"type":29},"若需要长期跟踪，把每轮的表格存档，按地区汇总 p50、p95、异常关闭率三个数，观察趋势而不是单点。趋势稳定后的差异，才值得作为选型依据。",{"type":31,"level":32,"children":249},[250],{"text":251,"type":29},"FAQ",{"type":25,"children":253},[254],{"text":255,"type":29,"marks":256},"Q：只测握手时间，能不能代表会话质量？",[257],{"type":258},"bold",{"type":25,"children":260},[261],{"text":262,"type":29},"A：不能。握手只覆盖建连阶段；还要报告消息延迟、观察时长、自然断开与窗口结束仍在线等状态，并按可用证据判断异常关闭。",{"type":25,"children":264},[265],{"text":266,"type":29,"marks":267},"Q：不同地区测出的差异，一定是网络造成的吗？",[268],{"type":258},{"type":25,"children":270},[271],{"text":272,"type":29},"A：不一定。配置地区、IP 库识别结果、代理出口和服务端调度都可能不同，应分别记录并限定结论到本次出口、目标和时间窗；ASN 不能单独证明物理路径。",{"type":25,"children":274},[275],{"text":276,"type":29,"marks":277},"Q：样本量多少才够下结论？",[278],{"type":258},{"type":25,"children":280},[281],{"text":282,"type":29},"A：没有适用于所有场景的固定次数。按预期差异、波动、成功率精度和时段覆盖确定样本量，并同时报告独立会话数、每会话消息数及尾部分位数的不确定性。","",[285,524,795],{"source":6,"id":286,"documentId":287,"translationKey":287,"locale":9,"slug":288,"path":289,"_path":289,"title":290,"description":291,"excerpt":291,"image":292,"publishedAt":293,"updatedAt":294,"author":17,"tags":295,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":296,"bodyHtml":283},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",[],[297,301,305,315,319,323,327,331,335,345,349,371,375,379,393,397,401,405,423,427,431,435,439,443,447,451,465,469,473,483,491,494,500,504,510,514,520],{"type":25,"children":298},[299],{"text":300,"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":302},[303],{"text":304,"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":306},[307,311],{"type":41,"children":308},[309],{"text":310,"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":312},[313],{"text":314,"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":316},[317],{"text":318,"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":320},[321],{"text":322,"type":29},"如何分别制造三种可控断线",{"type":25,"children":324},[325],{"text":326,"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":328},[329],{"text":330,"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":332},[333],{"text":334,"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":336},[337,339,344],{"text":338,"type":29},"网络层中断的定位可参考",{"text":340,"type":29,"marks":341},"代理 IP 故障排查中心",[342],{"href":343,"type":162},"https:\u002F\u002Fhelp.quanqiudaili.com\u002Fdocs\u002FSmart-check",{"text":164,"type":29},{"type":31,"level":32,"children":346},[347],{"text":348,"type":29},"重连策略应该观察哪些指标",{"type":37,"format":38,"children":350},[351,355,359,363,367],{"type":41,"children":352},[353],{"text":354,"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":356},[357],{"text":358,"type":29},"重连间隔序列：是否指数退避，是否带随机抖动。",{"type":41,"children":360},[361],{"text":362,"type":29},"重连次数上限：达到上限后是放弃还是降级。",{"type":41,"children":364},[365],{"text":366,"type":29},"鉴权是否重做：令牌过期时是否重新获取。",{"type":41,"children":368},[369],{"text":370,"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":372},[373],{"text":374,"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":376},[377],{"text":378,"type":29},"常见的三种会话恢复机制怎么验证",{"type":37,"format":38,"children":380},[381,385,389],{"type":41,"children":382},[383],{"text":384,"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":386},[387],{"text":388,"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":390},[391],{"text":392,"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":394},[395],{"text":396,"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":398},[399],{"text":400,"type":29},"如何检查补发消息是否重复或乱序",{"type":25,"children":402},[403],{"text":404,"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":406},[407,411,415,419],{"type":41,"children":408},[409],{"text":410,"type":29},"每条业务消息带唯一 ID 与单调递增序号。",{"type":41,"children":412},[413],{"text":414,"type":29},"客户端记录已处理 ID 的窗口，重复 ID 直接丢弃。",{"type":41,"children":416},[417],{"text":418,"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":420},[421],{"text":422,"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":424},[425],{"text":426,"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":428},[429],{"text":430,"type":29},"如何确认重连后业务状态仍然一致",{"type":25,"children":432},[433],{"text":434,"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":436},[437],{"text":438,"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":440},[441],{"text":442,"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":444},[445],{"text":446,"type":29},"如何记录恢复时间并完成验收",{"type":25,"children":448},[449],{"text":450,"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":452},[453,457,461],{"type":41,"children":454},[455],{"text":456,"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":458},[459],{"text":460,"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":462},[463],{"text":464,"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":466},[467],{"text":468,"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":470},[471],{"text":472,"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":474},[475,477,482],{"text":476,"type":29},"测试工具的接入参数可对照",{"text":478,"type":29,"marks":479},"代理设置帮助",[480],{"href":481,"type":162},"https:\u002F\u002Fhelp.quanqiudaili.com\u002Fdocs\u002Fproxy-Settings",{"text":164,"type":29},{"type":25,"children":484},[485,487,490],{"text":486,"type":29},"WebSocket 关闭状态和 Close 帧语义见 ",{"text":230,"type":29,"marks":488},[489],{"href":233,"type":162},{"text":164,"type":29},{"type":31,"level":32,"children":492},[493],{"text":251,"type":29},{"type":25,"children":495},[496],{"text":497,"type":29,"marks":498},"Q：重连成功但消息少了一段，问题出在哪？",[499],{"type":258},{"type":25,"children":501},[502],{"text":503,"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":505},[506],{"text":507,"type":29,"marks":508},"Q：close code 1006 一定能复现吗？",[509],{"type":258},{"type":25,"children":511},[512],{"text":513,"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":515},[516],{"text":517,"type":29,"marks":518},"Q：重复消息一定要在客户端去重吗？",[519],{"type":258},{"type":25,"children":521},[522],{"text":523,"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":525,"documentId":526,"translationKey":526,"locale":9,"slug":527,"path":528,"_path":528,"title":529,"description":530,"excerpt":530,"image":531,"publishedAt":532,"updatedAt":533,"author":17,"tags":534,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":535,"bodyHtml":283},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",[],[536,540,544,558,562,566,570,588,592,605,609,631,635,639,661,665,669,673,707,711,715,719,737,741,745,749,753,762,765,771,775,781,785,791],{"type":25,"children":537},[538],{"text":539,"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":541},[542],{"text":543,"type":29},"WebSocket 代理测试报告应该分哪三层记录",{"type":37,"format":38,"children":545},[546,550,554],{"type":41,"children":547},[548],{"text":549,"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":551},[552],{"text":553,"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":555},[556],{"text":557,"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":559},[560],{"text":561,"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":563},[564],{"text":565,"type":29},"握手和建连阶段要记录哪些字段",{"type":25,"children":567},[568],{"text":569,"type":29},"握手要分四段记录，不能只写总和：",{"type":37,"format":38,"children":571},[572,576,580,584],{"type":41,"children":573},[574],{"text":575,"type":29},"DNS 解析耗时（ms）。",{"type":41,"children":577},[578],{"text":579,"type":29},"TCP 建连耗时（ms）。",{"type":41,"children":581},[582],{"text":583,"type":29},"TLS 握手耗时（ms，wss 才有）。",{"type":41,"children":585},[586],{"text":587,"type":29},"Upgrade 往返耗时（ms）：请求发出到收到 `101 Switching Protocols`。",{"type":25,"children":589},[590],{"text":591,"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":593},[594,596,599,601,604],{"text":595,"type":29},"握手前的代理参数可对照",{"text":478,"type":29,"marks":597},[598],{"href":481,"type":162},{"text":600,"type":29},"；Upgrade 握手字段定义见 ",{"text":230,"type":29,"marks":602},[603],{"href":233,"type":162},{"text":164,"type":29},{"type":31,"level":32,"children":606},[607],{"text":608,"type":29},"连接持续多久、为什么断开要怎么记录",{"type":37,"format":38,"children":610},[611,615,619,623,627],{"type":41,"children":612},[613],{"text":614,"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":616},[617],{"text":618,"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":620},[621],{"text":622,"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":624},[625],{"text":626,"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":628},[629],{"text":630,"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":632},[633],{"text":634,"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":636},[637],{"text":638,"type":29},"重连过程应该记录哪些指标",{"type":37,"format":38,"children":640},[641,645,649,653,657],{"type":41,"children":642},[643],{"text":644,"type":29},"首次重连延迟（ms\u002Fs）：从检测到断线到发起重连。",{"type":41,"children":646},[647],{"text":648,"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":650},[651],{"text":652,"type":29},"重连成功率：成功建立新连接的比例。",{"type":41,"children":654},[655],{"text":656,"type":29},"恢复耗时：从断线到业务可用的时间。",{"type":41,"children":658},[659],{"text":660,"type":29},"是否携带恢复凭据：如 resume token、last sequence。",{"type":25,"children":662},[663],{"text":664,"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":666},[667],{"text":668,"type":29},"如何汇总三层字段并补充消息细节",{"type":25,"children":670},[671],{"text":672,"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":674},[675,679,683,687,691,695,699,703],{"type":41,"children":676},[677],{"text":678,"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":680},[681],{"text":682,"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":684},[685],{"text":686,"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":688},[689],{"text":690,"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":692},[693],{"text":694,"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":696},[697],{"text":698,"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":700},[701],{"text":702,"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":704},[705],{"text":706,"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":708},[709],{"text":710,"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":712},[713],{"text":714,"type":29},"地区和代理链路还要记录哪些信息",{"type":25,"children":716},[717],{"text":718,"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":720},[721,725,729,733],{"type":41,"children":722},[723],{"text":724,"type":29},"地区：客户端所在地区，与出口地区分开写。",{"type":41,"children":726},[727],{"text":728,"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":730},[731],{"text":732,"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":734},[735],{"text":736,"type":29},"链路跳数：若代理有多跳，记录跳数与各跳耗时。",{"type":25,"children":738},[739],{"text":740,"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":742},[743],{"text":744,"type":29},"报告模板应该怎样组织和统计",{"type":25,"children":746},[747],{"text":748,"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":750},[751],{"text":752,"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":754},[755,757,760],{"text":756,"type":29},"连接异常样本可按",{"text":340,"type":29,"marks":758},[759],{"href":343,"type":162},{"text":761,"type":29},"的字段顺序补充。",{"type":31,"level":32,"children":763},[764],{"text":251,"type":29},{"type":25,"children":766},[767],{"text":768,"type":29,"marks":769},"Q：报告里必须记录出口 IP 吗？",[770],{"type":258},{"type":25,"children":772},[773],{"text":774,"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":776},[777],{"text":778,"type":29,"marks":779},"Q：close code 只记 1006 够不够？",[780],{"type":258},{"type":25,"children":782},[783],{"text":784,"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":786},[787],{"text":788,"type":29,"marks":789},"Q：时延只报平均值可以吗？",[790],{"type":258},{"type":25,"children":792},[793],{"text":794,"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":796,"documentId":797,"translationKey":797,"locale":9,"slug":798,"path":799,"_path":799,"title":800,"description":801,"excerpt":801,"image":802,"publishedAt":803,"updatedAt":804,"author":17,"tags":805,"featured":19,"sortOrder":20,"readingTime":806,"bodyType":22,"bodyBlocks":807,"bodyHtml":283},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,[808,812,816,826,830,834,838,842,860,864,868,872,876,880,888,892,896,900,904,908,918,922,926,952,956,960,964,978,982,986,990,1004,1008,1023,1027,1031,1035,1038,1044,1048,1054,1058,1064],{"type":25,"children":809},[810],{"text":811,"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":813},[814],{"text":815,"type":29},"代理环境变量污染通常会表现为什么",{"type":37,"format":38,"children":817},[818,822],{"type":41,"children":819},[820],{"text":821,"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":823},[824],{"text":825,"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":827},[828],{"text":829,"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":831},[832],{"text":833,"type":29},"哪些代理环境变量会生效，优先级如何判断",{"type":25,"children":835},[836],{"text":837,"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":839},[840],{"text":841,"type":29},"为什么要同时检查大小写变量和 ALL_PROXY",{"type":37,"format":38,"children":843},[844,848,852,856],{"type":41,"children":845},[846],{"text":847,"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":849},[850],{"text":851,"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":853},[854],{"text":855,"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":857},[858],{"text":859,"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":861},[862],{"text":863,"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":865},[866],{"text":867,"type":29},"环境变量大小写相同时应该相信哪一个",{"type":25,"children":869},[870],{"text":871,"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":873},[874],{"text":875,"type":29},"为什么没写代理的 requests 仍可能走代理",{"type":25,"children":877},[878],{"text":879,"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":881},[882,884,887],{"text":883,"type":29},"代理环境变量与请求库的关系可对照",{"text":478,"type":29,"marks":885},[886],{"href":481,"type":162},{"text":164,"type":29},{"type":31,"level":32,"children":889},[890],{"text":891,"type":29},"如何用 fixture 固定环境并提前断言",{"type":25,"children":893},[894],{"text":895,"type":29},"在测试层做三件事：",{"type":25,"children":897},[898],{"text":899,"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":901},[902],{"text":903,"type":29},"```python\nimport pytest",{"type":25,"children":905},[906],{"text":907,"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":909},[910,912,917],{"text":911,"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":913,"type":29,"marks":914},"pytest monkeypatch 文档",[915],{"href":916,"type":162},"https:\u002F\u002Fdocs.pytest.org\u002Fen\u002Fstable\u002Fhow-to\u002Fmonkeypatch.html",{"text":164,"type":29},{"type":31,"level":32,"children":919},[920],{"text":921,"type":29},"NO_PROXY 哪些边界情况必须单独测试",{"type":25,"children":923},[924],{"text":925,"type":29},"`NO_PROXY` 的匹配规则容易想当然，至少要覆盖：",{"type":37,"format":38,"children":927},[928,932,936,940,944,948],{"type":41,"children":929},[930],{"text":931,"type":29},"精确域名与子域名：`example.com` 是否覆盖 `api.example.com`。",{"type":41,"children":933},[934],{"text":935,"type":29},"端口写法：`example.com:443` 与不带端口的差异。",{"type":41,"children":937},[938],{"text":939,"type":29},"IP 与 CIDR：单 IP、网段写法是否被支持。",{"type":41,"children":941},[942],{"text":943,"type":29},"本机地址：`localhost`、`127.0.0.1`、`::1`。",{"type":41,"children":945},[946],{"text":947,"type":29},"通配：`*` 是否让所有请求直连。",{"type":41,"children":949},[950],{"text":951,"type":29},"空值：变量存在但为空字符串时的行为。",{"type":25,"children":953},[954],{"text":955,"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":957},[958],{"text":959,"type":29},"如何证明请求真的走了预期出口",{"type":25,"children":961},[962],{"text":963,"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":965},[966,970,974],{"type":41,"children":967},[968],{"text":969,"type":29},"期望走代理：出口应属于代理侧地址段。",{"type":41,"children":971},[972],{"text":973,"type":29},"期望直连：出口应属于本机公网地址段。",{"type":41,"children":975},[976],{"text":977,"type":29},"期望绕过：目标应命中 `NO_PROXY`，出口为直连地址。",{"type":25,"children":979},[980],{"text":981,"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":983},[984],{"text":985,"type":29},"如何隔离开发机和 CI 继承的代理环境",{"type":25,"children":987},[988],{"text":989,"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":991},[992,996,1000],{"type":41,"children":993},[994],{"text":995,"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":997},[998],{"text":999,"type":29},"CI 任务显式声明环境，不继承 runner 的全局变量。",{"type":41,"children":1001},[1002],{"text":1003,"type":29},"需要真实代理的集成测试单独标记，默认不跑。",{"type":25,"children":1005},[1006],{"text":1007,"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":1009},[1010,1012,1017,1019,1022],{"text":1011,"type":29},"requests 的环境设置行为见 ",{"text":1013,"type":29,"marks":1014},"Requests 高级用法文档",[1015],{"href":1016,"type":162},"https:\u002F\u002Frequests.readthedocs.io\u002Fen\u002Fstable\u002Fuser\u002Fadvanced\u002F",{"text":1018,"type":29},"；测试环境变量恢复可参考 ",{"text":913,"type":29,"marks":1020},[1021],{"href":916,"type":162},{"text":164,"type":29},{"type":31,"level":32,"children":1024},[1025],{"text":1026,"type":29},"如何通过清理检查和故意失败验证测试有效",{"type":25,"children":1028},[1029],{"text":1030,"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":1032},[1033],{"text":1034,"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":1036},[1037],{"text":251,"type":29},{"type":25,"children":1039},[1040],{"text":1041,"type":29,"marks":1042},"Q：代码里没写代理，还会被环境变量影响吗？",[1043],{"type":258},{"type":25,"children":1045},[1046],{"text":1047,"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":1049},[1050],{"text":1051,"type":29,"marks":1052},"Q：`NO_PROXY` 写了域名，为什么子域名还是走了代理？",[1053],{"type":258},{"type":25,"children":1055},[1056],{"text":1057,"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":1059},[1060],{"text":1061,"type":29,"marks":1062},"Q：CI 上通过、本地失败，先查什么？",[1063],{"type":258},{"type":25,"children":1065},[1066],{"text":1067,"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。",[1069],{"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":1070,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":1071,"bodyHtml":283},[],[1072,1075,1078,1098,1101,1104,1107,1124,1127,1130,1133,1136,1139,1142,1145,1148,1151,1162,1165,1172,1175,1178,1192,1195,1198,1201,1204,1215,1218,1225,1228,1231,1234,1237,1242,1245,1250,1253,1258],{"type":25,"children":1073},[1074],{"text":28,"type":29},{"type":31,"level":32,"children":1076},[1077],{"text":35,"type":29},{"type":37,"format":38,"children":1079},[1080,1083,1086,1089,1092,1095],{"type":41,"children":1081},[1082],{"text":44,"type":29},{"type":41,"children":1084},[1085],{"text":48,"type":29},{"type":41,"children":1087},[1088],{"text":52,"type":29},{"type":41,"children":1090},[1091],{"text":56,"type":29},{"type":41,"children":1093},[1094],{"text":60,"type":29},{"type":41,"children":1096},[1097],{"text":64,"type":29},{"type":25,"children":1099},[1100],{"text":68,"type":29},{"type":31,"level":32,"children":1102},[1103],{"text":72,"type":29},{"type":25,"children":1105},[1106],{"text":76,"type":29},{"type":37,"format":38,"children":1108},[1109,1112,1115,1118,1121],{"type":41,"children":1110},[1111],{"text":82,"type":29},{"type":41,"children":1113},[1114],{"text":86,"type":29},{"type":41,"children":1116},[1117],{"text":90,"type":29},{"type":41,"children":1119},[1120],{"text":94,"type":29},{"type":41,"children":1122},[1123],{"text":98,"type":29},{"type":25,"children":1125},[1126],{"text":102,"type":29},{"type":31,"level":32,"children":1128},[1129],{"text":106,"type":29},{"type":25,"children":1131},[1132],{"text":110,"type":29},{"type":31,"level":21,"children":1134},[1135],{"text":114,"type":29},{"type":25,"children":1137},[1138],{"text":118,"type":29},{"type":25,"children":1140},[1141],{"text":122,"type":29},{"type":31,"level":21,"children":1143},[1144],{"text":126,"type":29},{"type":25,"children":1146},[1147],{"text":130,"type":29},{"type":31,"level":21,"children":1149},[1150],{"text":134,"type":29},{"type":37,"format":38,"children":1152},[1153,1156,1159],{"type":41,"children":1154},[1155],{"text":140,"type":29},{"type":41,"children":1157},[1158],{"text":144,"type":29},{"type":41,"children":1160},[1161],{"text":148,"type":29},{"type":25,"children":1163},[1164],{"text":152,"type":29},{"type":25,"children":1166},[1167,1168,1171],{"text":156,"type":29},{"text":158,"type":29,"marks":1169},[1170],{"href":161,"type":162},{"text":164,"type":29},{"type":31,"level":32,"children":1173},[1174],{"text":168,"type":29},{"type":25,"children":1176},[1177],{"text":172,"type":29},{"type":37,"format":38,"children":1179},[1180,1183,1186,1189],{"type":41,"children":1181},[1182],{"text":178,"type":29},{"type":41,"children":1184},[1185],{"text":182,"type":29},{"type":41,"children":1187},[1188],{"text":186,"type":29},{"type":41,"children":1190},[1191],{"text":190,"type":29},{"type":25,"children":1193},[1194],{"text":194,"type":29},{"type":31,"level":32,"children":1196},[1197],{"text":198,"type":29},{"type":25,"children":1199},[1200],{"text":202,"type":29},{"type":25,"children":1202},[1203],{"text":206,"type":29},{"type":37,"format":38,"children":1205},[1206,1209,1212],{"type":41,"children":1207},[1208],{"text":212,"type":29},{"type":41,"children":1210},[1211],{"text":216,"type":29},{"type":41,"children":1213},[1214],{"text":220,"type":29},{"type":25,"children":1216},[1217],{"text":224,"type":29},{"type":25,"children":1219},[1220,1221,1224],{"text":228,"type":29},{"text":230,"type":29,"marks":1222},[1223],{"href":233,"type":162},{"text":235,"type":29},{"type":31,"level":32,"children":1226},[1227],{"text":239,"type":29},{"type":25,"children":1229},[1230],{"text":243,"type":29},{"type":25,"children":1232},[1233],{"text":247,"type":29},{"type":31,"level":32,"children":1235},[1236],{"text":251,"type":29},{"type":25,"children":1238},[1239],{"text":255,"type":29,"marks":1240},[1241],{"type":258},{"type":25,"children":1243},[1244],{"text":262,"type":29},{"type":25,"children":1246},[1247],{"text":266,"type":29,"marks":1248},[1249],{"type":258},{"type":25,"children":1251},[1252],{"text":272,"type":29},{"type":25,"children":1254},[1255],{"text":276,"type":29,"marks":1256},[1257],{"type":258},{"type":25,"children":1259},[1260],{"text":282,"type":29},null,[],[]]