[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"blog-strapi-article:zh:ai-api-client-proxy-log-correlation":3,"blog-article:\u002Fblog\u002Fai-api-client-proxy-log-correlation":1125,"blog-related:\u002Fblog\u002Fai-api-client-proxy-log-correlation":1126,"blog-translations:\u002Fblog\u002Fai-api-client-proxy-log-correlation":1127},{"available":4,"article":5,"relatedArticles":206,"translations":992},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":205},"strapi",3930,"shvlqdilhendu4v1l6eo7yi0","zh","ai-api-client-proxy-log-correlation","\u002Fblog\u002Fai-api-client-proxy-log-correlation","AI API 报错与代理IP 连接日志如何关联","客户端报超时或 502、代理后台却显示连接正常时，可先用应用请求 ID 对齐客户端与源站，再结合代理连接 ID、目标、时间窗和出口映射隧道记录。文章说明普通隧道的可观测边界，并按错误类型列出核对字段。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_AI_API_cover_86d84b587d.webp","2026-11-04T06:48:13.652Z","2026-11-04T06:48:13.622Z","Global Proxies Team",[],false,0,3,"blocks",[24,30,36,40,57,61,65,69,73,77,81,85,89,93,97,101,105,109,131,135,147,151,155,159,170,174,181,185,191,195,201],{"type":25,"children":26},"paragraph",[27],{"text":28,"type":29},"客户端抛出超时或 502，代理后台却显示\"连接正常\"，这类争执通常源于两侧各记各的。关联不是把两个文件按时间排在一起，而是先定义一个共同的定位点，再让两侧都能落到这个点上。下面先给判定标准，再讲客户端、代理侧各自要补什么字段，最后是按错误类型选择关联依据的实测表。在代理IP这一侧，能落到具体请求的字段通常只有时间、连接四元组和出口 IP。","text",{"type":31,"level":32,"children":33},"heading",2,[34],{"text":35,"type":29},"什么情况下才算客户端错误和代理日志关联成功",{"type":25,"children":37},[38],{"text":39,"type":29},"关联成功要同时满足三条：",{"type":41,"format":42,"children":43},"list","unordered",[44,49,53],{"type":45,"children":46},"list-item",[47],{"text":48,"type":29},"可映射：按逻辑请求 ID、连接 ID、时间窗和目标地址建立对应关系。重试、连接复用、分块上传和一条连接承载多个请求都会形成一对多或多对一，逐次识别重试与复用即可，不要求唯一对应。",{"type":45,"children":50},[51],{"text":52,"type":29},"一致：两侧记录的目标主机、端口、出口 IP、时间段能对上。",{"type":45,"children":54},[55],{"text":56,"type":29},"可解释：代理侧的状态码或连接阶段，能解释客户端看到的错误类型。",{"type":25,"children":58},[59],{"text":60,"type":29},"三条缺一条，就只是\"时间相近\"，不是关联。很多排查失败是因为只做了第三条——看到时间接近就下结论，忽略了重试导致的一对多。",{"type":31,"level":32,"children":62},[63],{"text":64,"type":29},"如何让客户端生成并传递关联 ID",{"type":25,"children":66},[67],{"text":68,"type":29},"在应用层为每个逻辑请求生成 request ID，并记入客户端日志。若 API 文档允许透传自定义头，可同时发送给源站；这不会使普通 HTTPS 代理隧道自动看见该头：",{"type":25,"children":70},[71],{"text":72,"type":29},"```http\nX-Request-Id: 8f3c1a2e-\u003Cseq>\n```",{"type":25,"children":74},[75],{"text":76,"type":29},"注意常规 HTTPS 隧道下代理通常看不到应用层自定义头，能记录的是 CONNECT 目标、连接 ID，以及实现允许的部分 TLS 元数据。所以关联 ID 只能解决\"客户端 ↔ 源站\"这一段，代理侧需要另外的连接级标识。把这两段分开设计，才不会指望代理日志里出现应用自定义头。",{"type":31,"level":32,"children":78},[79],{"text":80,"type":29},"代理日志还需要记录哪些连接字段",{"type":25,"children":82},[83],{"text":84,"type":29},"代理侧可记录会话或连接 ID、目标 `host:port`、出口 IP、建连与结束时间、传输字节数和实现提供的结束原因。按实现情况补充 SNI、TLS 阶段等字段，并注明普通 CONNECT\u002FSOCKS 隧道通常无法观察端到端 TLS 握手错误。TLS 具体错误应从客户端或源站取证。",{"type":25,"children":86},[87],{"text":88,"type":29},"如果代理实现能提供 SNI，可用它补充目标域名；不能提供时不要推断。出口 IP、会话分配记录和时间窗可帮助映射连接，但不能单独证明物理地区。一条应用逻辑请求可能产生多次连接尝试；HTTP\u002F2 等连接复用也可能让多个逻辑请求共用一条隧道。",{"type":31,"level":32,"children":90},[91],{"text":92,"type":29},"如何用时间和检索窗口对齐两边日志",{"type":25,"children":94},[95],{"text":96,"type":29},"两侧时钟不同步是关联失败的头号原因。做法是先校准：",{"type":25,"children":98},[99],{"text":100,"type":29},"1. 用同一时间源（NTP）对齐客户端主机与代理侧（代理IP服务）的时间，记录当前偏移量。\n2. 检索时以客户端错误时间为锚点，向前后各取一个窗口（例如 ±5 秒），而不是精确匹配。\n3. 在窗口内再用连接 ID 或出口 IP 收缩，避免把相邻请求的错误算进来。",{"type":25,"children":102},[103],{"text":104,"type":29},"跨时区时统一用 UTC 存储，展示层再转换。日志里同时写本地时间和 UTC 偏移，能省掉大量事后换算。",{"type":31,"level":32,"children":106},[107],{"text":108,"type":29},"不同错误类型应该优先匹配哪些字段",{"type":41,"format":42,"children":110},[111,115,119,123,127],{"type":45,"children":112},[113],{"text":114,"type":29},"错误类型：407 认证失败 ｜ 客户端可见：“Proxy Authentication Required” ｜ 代理侧可查字段：代理认证响应及会话记录（若可用） ｜ 关联优先级：会话\u002F尝试 ID 与时间窗",{"type":45,"children":116},[117],{"text":118,"type":29},"错误类型：502 \u002F 504 ｜ 客户端可见：网关错误、超时 ｜ 代理侧可查字段：客户端错误；代理上游连接信息（实现提供时可用） ｜ 关联优先级：目标、连接 ID 与时间窗",{"type":45,"children":120},[121],{"text":122,"type":29},"错误类型：连接被重置 ｜ 客户端可见：Connection reset ｜ 代理侧可查字段：客户端错误；代理\u002F源站重置证据（可取得时） ｜ 关联优先级：目标、连接 ID 与时间窗",{"type":45,"children":124},[125],{"text":126,"type":29},"错误类型：TLS 握手失败 ｜ 客户端可见：客户端证书或握手错误 ｜ 代理侧可查字段：客户端\u002F源站错误；代理 SNI 或握手字段（实现提供时可用） ｜ 关联优先级：目标、连接 ID 与时间窗",{"type":45,"children":128},[129],{"text":130,"type":29},"错误类型：长时间无响应 ｜ 客户端可见：读超时 ｜ 代理侧可查字段：传输字节数、最后活跃时间 ｜ 关联优先级：连接 ID 加时间窗",{"type":25,"children":132},[133],{"text":134,"type":29},"407 通常可从代理认证响应及会话记录核对；502、504 和连接重置需要结合代理实现、客户端错误和上游连接证据判断。同一错误码在不同实现中的含义可能不同，不能只凭码值定责。",{"type":25,"children":136},[137,139,145],{"text":138,"type":29},"代理连接失败可先查看",{"text":140,"type":29,"marks":141},"智能检测助手",[142],{"href":143,"type":144},"https:\u002F\u002Fhelp.quanqiudaili.com\u002Fdocs\u002FSmart-check","link",{"text":146,"type":29},"；客户端自定义头与源站关联应按所用 HTTP 客户端及 API 文档核实，不能假设普通代理会记录这些头。",{"type":31,"level":32,"children":148},[149],{"text":150,"type":29},"客户端和代理日志对不上时先查什么",{"type":25,"children":152},[153],{"text":154,"type":29},"1. 确认代理日志已启用，覆盖了正确时间、实例和保留期，并确认查询账号有读取权限。\n2. 核对逻辑请求 ID、尝试 ID、客户端连接 ID 与代理连接 ID 的映射；该映射应脱敏保存。\n3. 用客户端连接事件和代理实际建连结果互证请求是否到达、上游是否建立。缺少任一侧记录时，标注“证据不足”，不要直接断定故障在客户端出口或代理上游。\n4. 考虑采样、时钟偏移、连接复用和重试，再按目标与时间窗收缩候选记录。",{"type":25,"children":156},[157],{"text":158,"type":29},"脱敏示例：`logical_id=L7 → attempt_id=A2 → client_conn=C9 → proxy_conn=P4`。每个箭头都应有实际日志或分配记录支持。",{"type":25,"children":160},[161,163,168],{"text":162,"type":29},"CONNECT 隧道的基本语义见 ",{"text":164,"type":29,"marks":165},"RFC 9110 §9.3.6",[166],{"href":167,"type":144},"https:\u002F\u002Fwww.rfc-editor.org\u002Frfc\u002Frfc9110.html#name-connect",{"text":169,"type":29},"；客户端超时和 TLS 错误应查所用 HTTP 库与 API 提供方的官方文档，代理日志不能替代这些端点证据。",{"type":31,"level":32,"children":171},[172],{"text":173,"type":29},"FAQ",{"type":25,"children":175},[176],{"text":177,"type":29,"marks":178},"Q：代理日志里看不到自定义请求头，还能关联吗？",[179],{"type":180},"bold",{"type":25,"children":182},[183],{"text":184,"type":29},"A：可以做候选关联。普通 HTTPS 隧道通常看不到应用自定义头，可用目标地址、时间窗和连接 ID 映射；是否关联成功还需两侧日志和连接映射证据支持。",{"type":25,"children":186},[187],{"text":188,"type":29,"marks":189},"Q：客户端和代理时间差几分钟，怎么处理？",[190],{"type":180},{"type":25,"children":192},[193],{"text":194,"type":29},"A：先测偏移量并校准，再按偏移后的时间检索。偏移无法消除时，用出口 IP 和目标地址先收缩候选，再看连接时长是否吻合。",{"type":25,"children":196},[197],{"text":198,"type":29,"marks":199},"Q：一次客户端请求在代理侧出现多条记录，正常吗？",[200],{"type":180},{"type":25,"children":202},[203],{"text":204,"type":29},"A：可能正常。重试可能产生多条连接尝试，连接复用则可能让多个请求共用一条连接。分块上传本身不必然增加连接，只有发生重连、并行分片等情况时才可能增加。应按逻辑请求、尝试和连接 ID 分层核对。","",[207,448,719],{"source":6,"id":208,"documentId":209,"translationKey":209,"locale":9,"slug":210,"path":211,"_path":211,"title":212,"description":213,"excerpt":213,"image":214,"publishedAt":215,"updatedAt":216,"author":17,"tags":217,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":218,"bodyHtml":205},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",[],[219,223,227,237,241,245,249,253,257,267,271,293,297,301,315,319,323,327,345,349,353,357,361,365,369,373,387,391,395,405,415,418,424,428,434,438,444],{"type":25,"children":220},[221],{"text":222,"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":224},[225],{"text":226,"type":29},"怎样判断连接恢复是否已经变成业务恢复",{"type":41,"format":42,"children":228},[229,233],{"type":45,"children":230},[231],{"text":232,"type":29},"连接恢复：新的 WebSocket 完成握手，收到 101，可以收发消息。",{"type":45,"children":234},[235],{"text":236,"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":238},[239],{"text":240,"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":242},[243],{"text":244,"type":29},"如何分别制造三种可控断线",{"type":25,"children":246},[247],{"text":248,"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":250},[251],{"text":252,"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":254},[255],{"text":256,"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":258},[259,261,265],{"text":260,"type":29},"网络层中断的定位可参考",{"text":262,"type":29,"marks":263},"代理 IP 故障排查中心",[264],{"href":143,"type":144},{"text":266,"type":29},"。",{"type":31,"level":32,"children":268},[269],{"text":270,"type":29},"重连策略应该观察哪些指标",{"type":41,"format":42,"children":272},[273,277,281,285,289],{"type":45,"children":274},[275],{"text":276,"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":45,"children":278},[279],{"text":280,"type":29},"重连间隔序列：是否指数退避，是否带随机抖动。",{"type":45,"children":282},[283],{"text":284,"type":29},"重连次数上限：达到上限后是放弃还是降级。",{"type":45,"children":286},[287],{"text":288,"type":29},"鉴权是否重做：令牌过期时是否重新获取。",{"type":45,"children":290},[291],{"text":292,"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":294},[295],{"text":296,"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":298},[299],{"text":300,"type":29},"常见的三种会话恢复机制怎么验证",{"type":41,"format":42,"children":302},[303,307,311],{"type":45,"children":304},[305],{"text":306,"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":45,"children":308},[309],{"text":310,"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":45,"children":312},[313],{"text":314,"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":316},[317],{"text":318,"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":320},[321],{"text":322,"type":29},"如何检查补发消息是否重复或乱序",{"type":25,"children":324},[325],{"text":326,"type":29},"恢复阶段最容易出现\"补多了\"和\"补乱了\"。校验方法：",{"type":41,"format":42,"children":328},[329,333,337,341],{"type":45,"children":330},[331],{"text":332,"type":29},"每条业务消息带唯一 ID 与单调递增序号。",{"type":45,"children":334},[335],{"text":336,"type":29},"客户端记录已处理 ID 的窗口，重复 ID 直接丢弃。",{"type":45,"children":338},[339],{"text":340,"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":45,"children":342},[343],{"text":344,"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":346},[347],{"text":348,"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":350},[351],{"text":352,"type":29},"如何确认重连后业务状态仍然一致",{"type":25,"children":354},[355],{"text":356,"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":358},[359],{"text":360,"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":362},[363],{"text":364,"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":366},[367],{"text":368,"type":29},"如何记录恢复时间并完成验收",{"type":25,"children":370},[371],{"text":372,"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":41,"format":42,"children":374},[375,379,383],{"type":45,"children":376},[377],{"text":378,"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":45,"children":380},[381],{"text":382,"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":45,"children":384},[385],{"text":386,"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":388},[389],{"text":390,"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":392},[393],{"text":394,"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":396},[397,399,404],{"text":398,"type":29},"测试工具的接入参数可对照",{"text":400,"type":29,"marks":401},"代理设置帮助",[402],{"href":403,"type":144},"https:\u002F\u002Fhelp.quanqiudaili.com\u002Fdocs\u002Fproxy-Settings",{"text":266,"type":29},{"type":25,"children":406},[407,409,414],{"text":408,"type":29},"WebSocket 关闭状态和 Close 帧语义见 ",{"text":410,"type":29,"marks":411},"RFC 6455",[412],{"href":413,"type":144},"https:\u002F\u002Fwww.rfc-editor.org\u002Frfc\u002Frfc6455",{"text":266,"type":29},{"type":31,"level":32,"children":416},[417],{"text":173,"type":29},{"type":25,"children":419},[420],{"text":421,"type":29,"marks":422},"Q：重连成功但消息少了一段，问题出在哪？",[423],{"type":180},{"type":25,"children":425},[426],{"text":427,"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":429},[430],{"text":431,"type":29,"marks":432},"Q：close code 1006 一定能复现吗？",[433],{"type":180},{"type":25,"children":435},[436],{"text":437,"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":439},[440],{"text":441,"type":29,"marks":442},"Q：重复消息一定要在客户端去重吗？",[443],{"type":180},{"type":25,"children":445},[446],{"text":447,"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":449,"documentId":450,"translationKey":450,"locale":9,"slug":451,"path":452,"_path":452,"title":453,"description":454,"excerpt":454,"image":455,"publishedAt":456,"updatedAt":457,"author":17,"tags":458,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":459,"bodyHtml":205},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",[],[460,464,468,482,486,490,494,512,516,529,533,555,559,563,585,589,593,597,631,635,639,643,661,665,669,673,677,686,689,695,699,705,709,715],{"type":25,"children":461},[462],{"text":463,"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":465},[466],{"text":467,"type":29},"WebSocket 代理测试报告应该分哪三层记录",{"type":41,"format":42,"children":469},[470,474,478],{"type":45,"children":471},[472],{"text":473,"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":45,"children":475},[476],{"text":477,"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":45,"children":479},[480],{"text":481,"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":483},[484],{"text":485,"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":487},[488],{"text":489,"type":29},"握手和建连阶段要记录哪些字段",{"type":25,"children":491},[492],{"text":493,"type":29},"握手要分四段记录，不能只写总和：",{"type":41,"format":42,"children":495},[496,500,504,508],{"type":45,"children":497},[498],{"text":499,"type":29},"DNS 解析耗时（ms）。",{"type":45,"children":501},[502],{"text":503,"type":29},"TCP 建连耗时（ms）。",{"type":45,"children":505},[506],{"text":507,"type":29},"TLS 握手耗时（ms，wss 才有）。",{"type":45,"children":509},[510],{"text":511,"type":29},"Upgrade 往返耗时（ms）：请求发出到收到 `101 Switching Protocols`。",{"type":25,"children":513},[514],{"text":515,"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":517},[518,520,523,525,528],{"text":519,"type":29},"握手前的代理参数可对照",{"text":400,"type":29,"marks":521},[522],{"href":403,"type":144},{"text":524,"type":29},"；Upgrade 握手字段定义见 ",{"text":410,"type":29,"marks":526},[527],{"href":413,"type":144},{"text":266,"type":29},{"type":31,"level":32,"children":530},[531],{"text":532,"type":29},"连接持续多久、为什么断开要怎么记录",{"type":41,"format":42,"children":534},[535,539,543,547,551],{"type":45,"children":536},[537],{"text":538,"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":45,"children":540},[541],{"text":542,"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":45,"children":544},[545],{"text":546,"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":45,"children":548},[549],{"text":550,"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":45,"children":552},[553],{"text":554,"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":556},[557],{"text":558,"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":560},[561],{"text":562,"type":29},"重连过程应该记录哪些指标",{"type":41,"format":42,"children":564},[565,569,573,577,581],{"type":45,"children":566},[567],{"text":568,"type":29},"首次重连延迟（ms\u002Fs）：从检测到断线到发起重连。",{"type":45,"children":570},[571],{"text":572,"type":29},"重连次数与退避序列：是否按指数退避，是否有抖动。",{"type":45,"children":574},[575],{"text":576,"type":29},"重连成功率：成功建立新连接的比例。",{"type":45,"children":578},[579],{"text":580,"type":29},"恢复耗时：从断线到业务可用的时间。",{"type":45,"children":582},[583],{"text":584,"type":29},"是否携带恢复凭据：如 resume token、last sequence。",{"type":25,"children":586},[587],{"text":588,"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":590},[591],{"text":592,"type":29},"如何汇总三层字段并补充消息细节",{"type":25,"children":594},[595],{"text":596,"type":29},"下表把三层字段汇总成可直接落表的形式。",{"type":41,"format":42,"children":598},[599,603,607,611,615,619,623,627],{"type":45,"children":600},[601],{"text":602,"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":45,"children":604},[605],{"text":606,"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":45,"children":608},[609],{"text":610,"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":45,"children":612},[613],{"text":614,"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":45,"children":616},[617],{"text":618,"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":45,"children":620},[621],{"text":622,"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":45,"children":624},[625],{"text":626,"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":45,"children":628},[629],{"text":630,"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":632},[633],{"text":634,"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":636},[637],{"text":638,"type":29},"地区和代理链路还要记录哪些信息",{"type":25,"children":640},[641],{"text":642,"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":41,"format":42,"children":644},[645,649,653,657],{"type":45,"children":646},[647],{"text":648,"type":29},"地区：客户端所在地区，与出口地区分开写。",{"type":45,"children":650},[651],{"text":652,"type":29},"出口 IP 与 ASN：每条连接记录一次，用于验证地区标签。",{"type":45,"children":654},[655],{"text":656,"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":45,"children":658},[659],{"text":660,"type":29},"链路跳数：若代理有多跳，记录跳数与各跳耗时。",{"type":25,"children":662},[663],{"text":664,"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":666},[667],{"text":668,"type":29},"报告模板应该怎样组织和统计",{"type":25,"children":670},[671],{"text":672,"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":674},[675],{"text":676,"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":678},[679,681,684],{"text":680,"type":29},"连接异常样本可按",{"text":262,"type":29,"marks":682},[683],{"href":143,"type":144},{"text":685,"type":29},"的字段顺序补充。",{"type":31,"level":32,"children":687},[688],{"text":173,"type":29},{"type":25,"children":690},[691],{"text":692,"type":29,"marks":693},"Q：报告里必须记录出口 IP 吗？",[694],{"type":180},{"type":25,"children":696},[697],{"text":698,"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":700},[701],{"text":702,"type":29,"marks":703},"Q：close code 只记 1006 够不够？",[704],{"type":180},{"type":25,"children":706},[707],{"text":708,"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":710},[711],{"text":712,"type":29,"marks":713},"Q：时延只报平均值可以吗？",[714],{"type":180},{"type":25,"children":716},[717],{"text":718,"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":720,"documentId":721,"translationKey":721,"locale":9,"slug":722,"path":723,"_path":723,"title":724,"description":725,"excerpt":725,"image":726,"publishedAt":727,"updatedAt":728,"author":17,"tags":729,"featured":19,"sortOrder":20,"readingTime":730,"bodyType":22,"bodyBlocks":731,"bodyHtml":205},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,[732,736,740,750,754,758,762,766,784,788,792,796,800,804,812,816,820,824,828,832,842,846,850,876,880,884,888,902,906,910,914,928,932,947,951,955,959,962,968,972,978,982,988],{"type":25,"children":733},[734],{"text":735,"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":737},[738],{"text":739,"type":29},"代理环境变量污染通常会表现为什么",{"type":41,"format":42,"children":741},[742,746],{"type":45,"children":743},[744],{"text":745,"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":45,"children":747},[748],{"text":749,"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":751},[752],{"text":753,"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":755},[756],{"text":757,"type":29},"哪些代理环境变量会生效，优先级如何判断",{"type":25,"children":759},[760],{"text":761,"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":763},[764],{"text":765,"type":29},"为什么要同时检查大小写变量和 ALL_PROXY",{"type":41,"format":42,"children":767},[768,772,776,780],{"type":45,"children":769},[770],{"text":771,"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":45,"children":773},[774],{"text":775,"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":45,"children":777},[778],{"text":779,"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":45,"children":781},[782],{"text":783,"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":785},[786],{"text":787,"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":789},[790],{"text":791,"type":29},"环境变量大小写相同时应该相信哪一个",{"type":25,"children":793},[794],{"text":795,"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":797},[798],{"text":799,"type":29},"为什么没写代理的 requests 仍可能走代理",{"type":25,"children":801},[802],{"text":803,"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":805},[806,808,811],{"text":807,"type":29},"代理环境变量与请求库的关系可对照",{"text":400,"type":29,"marks":809},[810],{"href":403,"type":144},{"text":266,"type":29},{"type":31,"level":32,"children":813},[814],{"text":815,"type":29},"如何用 fixture 固定环境并提前断言",{"type":25,"children":817},[818],{"text":819,"type":29},"在测试层做三件事：",{"type":25,"children":821},[822],{"text":823,"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":825},[826],{"text":827,"type":29},"```python\nimport pytest",{"type":25,"children":829},[830],{"text":831,"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":833},[834,836,841],{"text":835,"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":837,"type":29,"marks":838},"pytest monkeypatch 文档",[839],{"href":840,"type":144},"https:\u002F\u002Fdocs.pytest.org\u002Fen\u002Fstable\u002Fhow-to\u002Fmonkeypatch.html",{"text":266,"type":29},{"type":31,"level":32,"children":843},[844],{"text":845,"type":29},"NO_PROXY 哪些边界情况必须单独测试",{"type":25,"children":847},[848],{"text":849,"type":29},"`NO_PROXY` 的匹配规则容易想当然，至少要覆盖：",{"type":41,"format":42,"children":851},[852,856,860,864,868,872],{"type":45,"children":853},[854],{"text":855,"type":29},"精确域名与子域名：`example.com` 是否覆盖 `api.example.com`。",{"type":45,"children":857},[858],{"text":859,"type":29},"端口写法：`example.com:443` 与不带端口的差异。",{"type":45,"children":861},[862],{"text":863,"type":29},"IP 与 CIDR：单 IP、网段写法是否被支持。",{"type":45,"children":865},[866],{"text":867,"type":29},"本机地址：`localhost`、`127.0.0.1`、`::1`。",{"type":45,"children":869},[870],{"text":871,"type":29},"通配：`*` 是否让所有请求直连。",{"type":45,"children":873},[874],{"text":875,"type":29},"空值：变量存在但为空字符串时的行为。",{"type":25,"children":877},[878],{"text":879,"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":881},[882],{"text":883,"type":29},"如何证明请求真的走了预期出口",{"type":25,"children":885},[886],{"text":887,"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":41,"format":42,"children":889},[890,894,898],{"type":45,"children":891},[892],{"text":893,"type":29},"期望走代理：出口应属于代理侧地址段。",{"type":45,"children":895},[896],{"text":897,"type":29},"期望直连：出口应属于本机公网地址段。",{"type":45,"children":899},[900],{"text":901,"type":29},"期望绕过：目标应命中 `NO_PROXY`，出口为直连地址。",{"type":25,"children":903},[904],{"text":905,"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":907},[908],{"text":909,"type":29},"如何隔离开发机和 CI 继承的代理环境",{"type":25,"children":911},[912],{"text":913,"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":41,"format":42,"children":915},[916,920,924],{"type":45,"children":917},[918],{"text":919,"type":29},"在 `pytest.ini` 或 `conftest.py` 里统一清理，不依赖开发者本地配置。",{"type":45,"children":921},[922],{"text":923,"type":29},"CI 任务显式声明环境，不继承 runner 的全局变量。",{"type":45,"children":925},[926],{"text":927,"type":29},"需要真实代理的集成测试单独标记，默认不跑。",{"type":25,"children":929},[930],{"text":931,"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":933},[934,936,941,943,946],{"text":935,"type":29},"requests 的环境设置行为见 ",{"text":937,"type":29,"marks":938},"Requests 高级用法文档",[939],{"href":940,"type":144},"https:\u002F\u002Frequests.readthedocs.io\u002Fen\u002Fstable\u002Fuser\u002Fadvanced\u002F",{"text":942,"type":29},"；测试环境变量恢复可参考 ",{"text":837,"type":29,"marks":944},[945],{"href":840,"type":144},{"text":266,"type":29},{"type":31,"level":32,"children":948},[949],{"text":950,"type":29},"如何通过清理检查和故意失败验证测试有效",{"type":25,"children":952},[953],{"text":954,"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":956},[957],{"text":958,"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":960},[961],{"text":173,"type":29},{"type":25,"children":963},[964],{"text":965,"type":29,"marks":966},"Q：代码里没写代理，还会被环境变量影响吗？",[967],{"type":180},{"type":25,"children":969},[970],{"text":971,"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":973},[974],{"text":975,"type":29,"marks":976},"Q：`NO_PROXY` 写了域名，为什么子域名还是走了代理？",[977],{"type":180},{"type":25,"children":979},[980],{"text":981,"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":983},[984],{"text":985,"type":29,"marks":986},"Q：CI 上通过、本地失败，先查什么？",[987],{"type":180},{"type":25,"children":989},[990],{"text":991,"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。",[993],{"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":994,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":995,"bodyHtml":205},[],[996,999,1002,1005,1016,1019,1022,1025,1028,1031,1034,1037,1040,1043,1046,1049,1052,1055,1072,1075,1082,1085,1088,1091,1098,1101,1106,1109,1114,1117,1122],{"type":25,"children":997},[998],{"text":28,"type":29},{"type":31,"level":32,"children":1000},[1001],{"text":35,"type":29},{"type":25,"children":1003},[1004],{"text":39,"type":29},{"type":41,"format":42,"children":1006},[1007,1010,1013],{"type":45,"children":1008},[1009],{"text":48,"type":29},{"type":45,"children":1011},[1012],{"text":52,"type":29},{"type":45,"children":1014},[1015],{"text":56,"type":29},{"type":25,"children":1017},[1018],{"text":60,"type":29},{"type":31,"level":32,"children":1020},[1021],{"text":64,"type":29},{"type":25,"children":1023},[1024],{"text":68,"type":29},{"type":25,"children":1026},[1027],{"text":72,"type":29},{"type":25,"children":1029},[1030],{"text":76,"type":29},{"type":31,"level":32,"children":1032},[1033],{"text":80,"type":29},{"type":25,"children":1035},[1036],{"text":84,"type":29},{"type":25,"children":1038},[1039],{"text":88,"type":29},{"type":31,"level":32,"children":1041},[1042],{"text":92,"type":29},{"type":25,"children":1044},[1045],{"text":96,"type":29},{"type":25,"children":1047},[1048],{"text":100,"type":29},{"type":25,"children":1050},[1051],{"text":104,"type":29},{"type":31,"level":32,"children":1053},[1054],{"text":108,"type":29},{"type":41,"format":42,"children":1056},[1057,1060,1063,1066,1069],{"type":45,"children":1058},[1059],{"text":114,"type":29},{"type":45,"children":1061},[1062],{"text":118,"type":29},{"type":45,"children":1064},[1065],{"text":122,"type":29},{"type":45,"children":1067},[1068],{"text":126,"type":29},{"type":45,"children":1070},[1071],{"text":130,"type":29},{"type":25,"children":1073},[1074],{"text":134,"type":29},{"type":25,"children":1076},[1077,1078,1081],{"text":138,"type":29},{"text":140,"type":29,"marks":1079},[1080],{"href":143,"type":144},{"text":146,"type":29},{"type":31,"level":32,"children":1083},[1084],{"text":150,"type":29},{"type":25,"children":1086},[1087],{"text":154,"type":29},{"type":25,"children":1089},[1090],{"text":158,"type":29},{"type":25,"children":1092},[1093,1094,1097],{"text":162,"type":29},{"text":164,"type":29,"marks":1095},[1096],{"href":167,"type":144},{"text":169,"type":29},{"type":31,"level":32,"children":1099},[1100],{"text":173,"type":29},{"type":25,"children":1102},[1103],{"text":177,"type":29,"marks":1104},[1105],{"type":180},{"type":25,"children":1107},[1108],{"text":184,"type":29},{"type":25,"children":1110},[1111],{"text":188,"type":29,"marks":1112},[1113],{"type":180},{"type":25,"children":1115},[1116],{"text":194,"type":29},{"type":25,"children":1118},[1119],{"text":198,"type":29,"marks":1120},[1121],{"type":180},{"type":25,"children":1123},[1124],{"text":204,"type":29},null,[],[]]