[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"blog-strapi-article:zh:proxy-egress-switch-regression-testing":3,"blog-article:\u002Fblog\u002Fproxy-egress-switch-regression-testing":619,"blog-related:\u002Fblog\u002Fproxy-egress-switch-regression-testing":620,"blog-translations:\u002Fblog\u002Fproxy-egress-switch-regression-testing":621},{"available":4,"article":5,"relatedArticles":155,"translations":527},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":154},"strapi",3979,"is6kltrtpypl3t5zd4yd4lll","zh","proxy-egress-switch-regression-testing","\u002Fblog\u002Fproxy-egress-switch-regression-testing","应用切换代理 IP 出口后，如何确认请求和业务状态正常","切换出口后同时核对请求侧状态码、耗时与业务侧幂等结果，固定输入和时段做前后对照；接口返回 200 不等于业务状态没有回归。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_cover_b5987beb4d.webp","2026-11-06T07:37:48.321Z","2026-11-06T07:37:48.307Z","Global Proxies Team",[],false,0,2,"blocks",[24,37,42,46,50,54,58,62,66,78,82,86,90,94,98,102,106,117,121,133,137,142,146,150],{"type":25,"children":26},"paragraph",[27,30,35],{"text":28,"type":29},"换出口后的回归，要把“正常”拆成连通、响应结构、会话、业务状态四层。只测连通，会把“能访问但业务错”当成通过。","text",{"text":31,"type":29,"marks":32},"代理IP",[33],{"type":34},"bold",{"text":36,"type":29}," 变化最先影响的是会话粘性和地域策略，这两类问题在状态码上往往看不出来。",{"type":38,"level":21,"children":39},"heading",[40],{"text":41,"type":29},"分别检查连接、响应、登录会话和业务状态",{"type":25,"children":43},[44],{"text":45,"type":29},"连接层检查连接能否建立、是否中途断开；超时或连接重置属于连接问题。407 表示代理认证未通过，5xx 则是服务端返回的错误响应，应记录响应来自代理还是目标服务，不能统称为连接失败。",{"type":25,"children":47},[48],{"text":49,"type":29},"响应层按接口的字段契约比较，字段是否存在、类型是否正确、允许的取值差异（地区价格、实验分组）都要写进契约；数量不一定要求与基线完全一致，要求“数量不变”要给出业务理由。",{"type":25,"children":51},[52],{"text":53,"type":29},"会话层要按系统设计判断：有些业务要求维持原会话，另一些允许切换后重新鉴权。出现挑战页或验证码时，先记录现象并与基线比较，不要仅凭一次结果断定根因。",{"type":25,"children":55},[56],{"text":57,"type":29},"业务层核对订单、余额或状态机结果；重复提交或状态错乱都属于回归。\n四层要分开断言，不能用一个“请求成功”概括。",{"type":38,"level":21,"children":59},[60],{"text":61,"type":29},"使用相同任务和测试数据对比切换前后结果",{"type":25,"children":63},[64],{"text":65,"type":29},"控制变量至少包括：同一测试账号、同一组请求、同一时段、同一客户端版本。用可重置的测试资源，不要拿真实付费账号或生产数据做样本。",{"type":25,"children":67},[68,70,76],{"text":69,"type":29},"开始前可通过",{"text":71,"type":29,"marks":72},"代理设置帮助",[73],{"href":74,"type":75},"https:\u002F\u002Fhelp.quanqiudaili.com\u002Fdocs\u002Fproxy-Settings","link",{"text":77,"type":29},"核对出口及接入方式。",{"type":38,"level":21,"children":79},[80],{"text":81,"type":29},"检查出口切换后 Cookie、Token 和验证状态",{"type":25,"children":83},[84],{"text":85,"type":29},"服务端按 IP 绑定会话时，换出口会掉登录或触发验证。要验三件事：切换后 cookie 是否仍有效、token 是否被吊销、是否出现挑战页。出现任意一项，先归因到会话策略，不要直接换回原出口掩盖问题。测试范围限自有或已获授权的服务与账号；挑战页只用于兼容性与故障归因，不尝试绕过验证机制。",{"type":38,"level":21,"children":87},[88],{"text":89,"type":29},"用幂等键和状态机检查重复请求与业务状态",{"type":25,"children":91},[92],{"text":93,"type":29},"重试加上换出口，可能造成重复下单、重复扣款。先注明被测接口是否支持幂等：支持的话写清幂等键的作用域、保留期和结果查询方式，再在可重置资源上覆盖同键重放、并发提交、超时后结果未知、换出口重试四种情况；不支持的话，测的是客户端如何查询结果、何时停止重试，不能要求“加个键就一定只生效一次”。有副作用的接口，只允许在可重置的测试资源上跑。",{"type":25,"children":95},[96],{"text":97,"type":29},"差异归因时，先排除服务端地域策略和客户端缓存：地域定价、AB 实验、CDN 缓存都会让结果看起来“像代理问题”。隔离办法是固定测试账号、请求参数、缓存键与实验分组，在授权环境里对照直连、代理及同一地区样本；控制不了的变量列为未知，不直接归因到代理。",{"type":38,"level":21,"children":99},[100],{"text":101,"type":29},"记录四层回归结果并确认验收条件",{"type":25,"children":103},[104],{"text":105,"type":29},"记录时包括测试任务和数据、四层判据结果、会话是否保持、幂等是否生效，以及尚未解释的差异。连接、响应、会话和业务状态都通过，且关键差异有证据支持，才能判定没有回归。\n回归清单四项：连接与响应按契约通过、会话策略符合预期（保持或正确重新鉴权）、幂等与重试行为符合接口声明、关键差异有证据支持。任一项不满足就不能判为“无回归”。",{"type":25,"children":107},[108,110,115],{"text":109,"type":29},"若切换后请求仍异常，可按",{"text":111,"type":29,"marks":112},"全球代理使用流程",[113],{"href":114,"type":75},"https:\u002F\u002Fhelp.quanqiudaili.com\u002Fdocs\u002FProcess",{"text":116,"type":29},"回查获取代理、接入设置和验证步骤。",{"type":25,"children":118},[119],{"text":120,"type":29},"幂等方法的官方定义可参考：",{"type":122,"format":123,"children":124},"list","unordered",[125],{"type":126,"children":127},"list-item",[128],{"text":129,"type":29,"marks":130},"RFC 9110《HTTP Semantics》第 9.2 节（幂等方法）",[131],{"href":132,"type":75},"https:\u002F\u002Fwww.rfc-editor.org\u002Frfc\u002Frfc9110.html",{"type":38,"level":21,"children":134},[135],{"text":136,"type":29},"常见问题",{"type":38,"level":138,"children":139},3,[140],{"text":141,"type":29},"换出口后能正常访问，为什么还要测业务状态？",{"type":25,"children":143},[144],{"text":145,"type":29},"能访问只证明连通，会话和业务数据可能已经异常。",{"type":38,"level":138,"children":147},[148],{"text":149,"type":29},"怎样判断差异是代理造成的？",{"type":25,"children":151},[152],{"text":153,"type":29},"做同任务、同数据、同时段的对照，排除地域策略和缓存后再归因。","",[156,289,410],{"source":6,"id":157,"documentId":158,"translationKey":158,"locale":9,"slug":159,"path":160,"_path":160,"title":161,"description":162,"excerpt":162,"image":163,"publishedAt":164,"updatedAt":165,"author":17,"tags":166,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":167,"bodyHtml":154},3987,"ehyrf4ygtsshhwkd0o87xat6","tls-certificate-compatibility-proxy-testing","\u002Fblog\u002Ftls-certificate-compatibility-proxy-testing","应用发布前，如何检查代理环境下的 TLS 证书兼容性","先分清直连、隧道转发和代理中间人三类链路，再逐级核对证书链、信任库位置与 SAN；本机正常而容器失败的，多为 CA 没进镜像。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_TLS_cover_ad09ec52a5.webp","2026-11-06T07:43:24.293Z","2026-11-06T07:43:24.278Z",[],[168,177,181,185,194,198,202,206,210,214,218,222,226,230,234,238,242,246,257,261,270,273,277,281,285],{"type":25,"children":169},[170,172,175],{"text":171,"type":29},"代理环境下的 TLS 问题，先要分清证书由谁签发：直连看目标证书，隧道转发看目标证书，代理中间人看代理自签 CA。",{"text":31,"type":29,"marks":173},[174],{"type":34},{"text":176,"type":29}," 本身不改变 TLS 握手，但企业代理或抓包工具会替换证书链，发布前必须把这条链路完整验一遍。",{"type":38,"level":21,"children":178},[179],{"text":180,"type":29},"先确认代理是否会替换目标站的 TLS 证书",{"type":25,"children":182},[183],{"text":184,"type":29},"直连或 HTTP CONNECT 隧道通常保留目标站点的证书，应检查证书链、有效期和 SAN（证书允许使用的域名）。如果企业代理会解密并重新签发 TLS 流量，客户端看到的是企业代理签发的站点证书；此时要确认该证书能连到受信任的企业 CA。先判断代理是否替换了证书，再确定检查对象。",{"type":25,"children":186},[187,189,192],{"text":188,"type":29},"核对代理协议与客户端接入方式时，可参考",{"text":71,"type":29,"marks":190},[191],{"href":74,"type":75},{"text":193,"type":29},"。",{"type":38,"level":21,"children":195},[196],{"text":197,"type":29},"使用 OpenSSL 查看代理连接返回的证书链",{"type":25,"children":199},[200],{"text":201,"type":29},"```bash\nopenssl s_client -proxy proxy.example.com:8080 \\\n-connect target.example.com:443 \\\n-servername target.example.com -showcerts \\\n-verify_hostname target.example.com -verify_return_error \\\n-CAfile \u002Fpath\u002Fto\u002Fca-bundle.pem\n```",{"type":25,"children":203},[204],{"text":205,"type":29},"查看输出中的 `issuer`、有效期和 SAN，并重点看两件事：证书链能否构建到受信锚、主机名是否匹配。服务端通常只提供叶证书和中间证书，不要求它回传根证书；根证书应该来自你信任的本地信任库（`-CAfile` 或 `-CApath`），而不是从连接里“捡”到一个就信。验收看退出码和验证输出（`Verify return code`），不要只看“连接成功”；最后仍在真实应用运行时或容器里做一次正负对照。",{"type":25,"children":207},[208],{"text":209,"type":29},"`-proxy` 参数需要 OpenSSL 1.1.0 或更高版本，且默认只支持 HTTP CONNECT；SOCKS5 场景改用支持该协议的工具（如 curl 的 `--proxy socks5h:\u002F\u002F`）或对应的客户端验证。运行 `openssl version` 可先检查本机版本。",{"type":38,"level":21,"children":211},[212],{"text":213,"type":29},"检查系统、运行时和容器使用的证书信任库",{"type":25,"children":215},[216],{"text":217,"type":29},"信任库位置取决于应用使用的运行时：Linux 系统证书通常位于 `\u002Fetc\u002Fssl\u002Fcerts`，Java 使用自己的 `cacerts`，Node 可通过 `NODE_EXTRA_CA_CERTS` 指定额外 CA，Python 的 `requests` 常使用 `certifi`。容器还要单独检查镜像内的证书。本机正常而容器失败时，先比较两边实际加载的 CA。",{"type":38,"level":21,"children":219},[220],{"text":221,"type":29},"根据 TLS 错误定位证书链和主机名问题",{"type":25,"children":223},[224],{"text":225,"type":29},"出现 `unable to get local issuer certificate` 时，先检查证书链是否缺少中间 CA；出现 `self signed certificate` 时，确认它是否为预期的企业 CA 或测试证书。`hostname mismatch` 通常表示证书 SAN 不包含目标域名；`protocol version` 或 `cipher mismatch` 则需要核对客户端与服务端支持的 TLS 版本和加密套件。",{"type":38,"level":21,"children":227},[228],{"text":229,"type":29},"发布前检查证书有效期、信任链和轮换结果",{"type":38,"level":138,"children":231},[232],{"text":233,"type":29},"在预发布环境验证证书更新和轮换",{"type":25,"children":235},[236],{"text":237,"type":29},"发布前记录证书到期时间，证书轮换后复测一次；灰度环境的信任库要和生产一致，否则灰度通过、生产失败。",{"type":38,"level":138,"children":239},[240],{"text":241,"type":29},"记录证书签发者、有效期和信任库位置",{"type":25,"children":243},[244],{"text":245,"type":29},"发布前记录目标域名和端口、证书签发者与到期时间、信任库位置，以及代理是否会重新签发证书。\n检查清单四项：链能构建到受信锚、CA 在信任库、主机名匹配 SAN、证书未过期。任一项不满足就不能发布；发现证书替换影响生产流量时，立即回滚配置。",{"type":25,"children":247},[248,250,255],{"text":249,"type":29},"证书链异常仍无法归因时，可从",{"text":251,"type":29,"marks":252},"全球代理帮助中心",[253],{"href":254,"type":75},"https:\u002F\u002Fhelp.quanqiudaili.com\u002F",{"text":256,"type":29},"查找对应接入文档。",{"type":25,"children":258},[259],{"text":260,"type":29},"OpenSSL 参数以手册为准：",{"type":122,"format":123,"children":262},[263],{"type":126,"children":264},[265],{"text":266,"type":29,"marks":267},"`openssl s_client` 手册（`-proxy`、`-verify_hostname`、`-verify_return_error`）",[268],{"href":269,"type":75},"https:\u002F\u002Fdocs.openssl.org\u002F3.5\u002Fman1\u002Fopenssl-s_client\u002F",{"type":38,"level":21,"children":271},[272],{"text":136,"type":29},{"type":38,"level":138,"children":274},[275],{"text":276,"type":29},"代理环境下证书报错一定是代理问题吗？",{"type":25,"children":278},[279],{"text":280,"type":29},"不一定。先看证书链和信任库，主机名不符或 CA 缺失更常见。",{"type":38,"level":138,"children":282},[283],{"text":284,"type":29},"为什么本机正常、容器里失败？",{"type":25,"children":286},[287],{"text":288,"type":29},"容器镜像没有导入同一个 CA，信任库不一致。",{"source":6,"id":290,"documentId":291,"translationKey":291,"locale":9,"slug":292,"path":293,"_path":293,"title":294,"description":295,"excerpt":295,"image":296,"publishedAt":297,"updatedAt":298,"author":17,"tags":299,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":300,"bodyHtml":154},3986,"g5ee4zs0pa8hyuc307nk5nwf","proxy-error-handling-mock-testing","\u002Fblog\u002Fproxy-error-handling-mock-testing","如何在不依赖真实代理服务时测试应用的代理错误处理","用可注入的代理桩和故障矩阵覆盖超时、拒绝、认证失败，在受控环境验证重试、退避与降级；不接真实代理，也不记录任何真实凭据。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_cover_5ce09c130c.webp","2026-11-06T07:42:03.722Z","2026-11-06T07:42:03.708Z",[],[301,310,314,318,322,326,330,334,338,342,346,355,359,363,372,376,391,394,398,402,406],{"type":25,"children":302},[303,305,308],{"text":304,"type":29},"代理错误处理最适合用本地 mock 验证。407、超时、连接重置、TLS 失败、慢响应、半关闭六类故障都能稳定复现，测试不再依赖服务商侧的偶发故障。",{"text":31,"type":29,"marks":306},[307],{"type":34},{"text":309,"type":29}," 的真实链路不可控，mock 的价值就是把“偶发”变成“可重复”，但结论只对应用逻辑成立。",{"type":38,"level":21,"children":311},[312],{"text":313,"type":29},"代理错误测试需要覆盖的六种故障",{"type":25,"children":315},[316],{"text":317,"type":29},"可先覆盖六种常见故障：让 mock 返回带 `Proxy-Authenticate` 的 407，检查客户端是否重试或报告认证错误；延迟超过客户端超时时，检查超时处理；连接重置时，检查中断后的动作；TLS 负例一次只改一个变量（未受信 CA、可信 CA 但 SAN 不对、证书过期），并保持校验开启，看客户端实际报什么错；分段延迟发送时，记录首字节和完整响应耗时；测“响应不完整”时，用声明了 `Content-Length` 却提前 EOF、或缺少 chunked 终止块的响应做负例，完整的 close-delimited 响应只作正常对照——单纯关闭写方向并不等于响应不完整。之后再按应用实际支持的错误分支补充测试。\n矩阵按应用实际支持的错误分支建立，上表是起步集合，可以补代理 DNS 失败、认证挑战格式异常和重试上限三类用例。矩阵未覆盖的分支，不会被这一组测试验证。",{"type":38,"level":21,"children":319},[320],{"text":321,"type":29},"如何搭建可重复触发故障的 mock 代理",{"type":25,"children":323},[324],{"text":325,"type":29},"mitmproxy、toxiproxy 或自建 TCP 服务都可以用于 mock，但注入点要和故障所在层对应。407 需要能处理 CONNECT 并返回 `Proxy-Authenticate` 的受控代理；TCP 延迟和连接重置可在传输层注入；SOCKS 认证错误则要按 SOCKS 协议构造，不能用 HTTP 的 407 代替。记录工具、客户端版本、触发步骤和清理方法，其他人才能复现。",{"type":38,"level":21,"children":327},[328],{"text":329,"type":29},"除了错误码，还要验证应用如何处理故障",{"type":25,"children":331},[332],{"text":333,"type":29},"收到 407 只说明应用识别到了代理认证错误。还要检查重试次数和间隔、熔断或降级是否符合预期，以及日志是否隐藏凭据。407 和 401 要分开判断：前者由代理要求认证，后者由目标服务要求认证。",{"type":25,"children":335},[336],{"text":337,"type":29},"407 和 401 必须分开断言：`Proxy-Authenticate` 是代理要求认证，`WWW-Authenticate` 是目标服务要求认证。混在一起，会把代理认证失败记成应用登录失败。",{"type":38,"level":21,"children":339},[340],{"text":341,"type":29},"避免 mock 测试漏掉真实故障分支",{"type":25,"children":343},[344],{"text":345,"type":29},"要触发客户端超时，mock 延迟必须超过客户端配置的超时值；要测试请求中断后的处理，应在连接建立后再重置。mock 结果只说明应用对模拟故障的处理方式，不能直接推断真实代理链路表现。",{"type":25,"children":347},[348,350,353],{"text":349,"type":29},"mock 通过后若要转入真实环境验证，可按",{"text":111,"type":29,"marks":351},[352],{"href":114,"type":75},{"text":354,"type":29},"依次核对产品、参数和接入步骤。",{"type":38,"level":21,"children":356},[357],{"text":358,"type":29},"记录 mock 故障参数和应用处理结果",{"type":25,"children":360},[361],{"text":362,"type":29},"每个用例记录故障类型、注入参数、期望动作、实际结果，以及日志是否隐藏凭据。\n用例边界是应用逻辑，不覆盖真实服务商的 SLA 和地区行为。出现真实凭据进入 mock 日志、或测试影响到共享环境时，立即停止并清理。",{"type":25,"children":364},[365,367,370],{"text":366,"type":29},"开始真实环境验证前，可用",{"text":71,"type":29,"marks":368},[369],{"href":74,"type":75},{"text":371,"type":29},"核对协议、地址和认证参数。",{"type":25,"children":373},[374],{"text":375,"type":29},"故障注入工具与协议定义：",{"type":122,"format":123,"children":377},[378,385],{"type":126,"children":379},[380],{"text":381,"type":29,"marks":382},"toxiproxy（TCP 延迟、重置、带宽限制）",[383],{"href":384,"type":75},"https:\u002F\u002Fgithub.com\u002FShopify\u002Ftoxiproxy",{"type":126,"children":386},[387],{"text":388,"type":29,"marks":389},"RFC 9110《HTTP Semantics》（407 与 `Proxy-Authenticate`）",[390],{"href":132,"type":75},{"type":38,"level":21,"children":392},[393],{"text":136,"type":29},{"type":38,"level":138,"children":395},[396],{"text":397,"type":29},"mock 代理能替代真实代理测试吗？",{"type":25,"children":399},[400],{"text":401,"type":29},"不能。它验证应用逻辑，不验证真实链路的稳定性和地区行为。",{"type":38,"level":138,"children":403},[404],{"text":405,"type":29},"407 和 401 有什么区别？",{"type":25,"children":407},[408],{"text":409,"type":29},"407 是代理要求认证，401 是目标服务要求认证。",{"source":6,"id":411,"documentId":412,"translationKey":412,"locale":9,"slug":413,"path":414,"_path":414,"title":415,"description":416,"excerpt":416,"image":417,"publishedAt":418,"updatedAt":419,"author":17,"tags":420,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":421,"bodyHtml":154},3985,"bpamp8c4gnpz3p9kslssa97y","ci-runner-proxy-build-impact-testing","\u002Fblog\u002Fci-runner-proxy-build-impact-testing","如何在 CI Runner 中验证代理不会影响构建步骤","在 Runner 内固定镜像、构建输入与测试时段，用出口 IP、DNS、TLS 和依赖清单作为证据，区分代理影响与构建环境波动；产物哈希只在构建可复现时才作对照。","https:\u002F\u002Fcms.quanqiudaili.com\u002Fuploads\u002Flarge_CI_Runner_cover_023a325e36.webp","2026-11-06T07:41:47.066Z","2026-11-06T07:41:47.049Z",[],[422,426,430,434,438,442,451,455,459,463,467,471,475,479,488,492,508,511,515,519,523],{"type":25,"children":423},[424],{"text":425,"type":29},"判断代理是否影响 CI，不能只看最终构建是否成功。应按实际流程检查代码拉取、依赖安装、单元测试、构建和产物上传。代理配置可能来自环境变量、包管理器或容器后台服务；配置不一致时，本地命令与流水线可能表现不同。",{"type":38,"level":21,"children":427},[428],{"text":429,"type":29},"如何判断代理没有影响 CI 构建",{"type":25,"children":431},[432],{"text":433,"type":29},"按顺序检查代码是否拉取到相同 commit、依赖校验是否通过、单元测试结果是否与基线一致。构建可复现时比较产物哈希；否则比较关键文件和依赖清单。上传环节要确认远端文件实际存在，并在支持时核对摘要或完整性校验。各阶段记录退出码和耗时；确需记录出口 IP 时，应使用受控日志并隐藏代理凭据。",{"type":38,"level":21,"children":435},[436],{"text":437,"type":29},"检查 CI 环境变量、包管理器和容器的代理配置",{"type":25,"children":439},[440],{"text":441,"type":29},"先检查构建进程是否读取 `HTTP_PROXY` 和 `HTTPS_PROXY`，再确认 `NO_PROXY` 中哪些地址会绕过代理。npm、pip 和 Maven 也可能分别读取 `.npmrc`、`pip.conf` 或 `settings.xml`；容器拉取镜像则由容器后台服务的配置决定。不要只凭宿主机环境变量判断代理已经生效。",{"type":25,"children":443},[444,446,449],{"text":445,"type":29},"配置入口和协议参数可对照",{"text":71,"type":29,"marks":447},[448],{"href":74,"type":75},{"text":450,"type":29},"逐项核实。",{"type":38,"level":21,"children":452},[453],{"text":454,"type":29},"对比启用代理和不启用代理时的构建结果",{"type":25,"children":456},[457],{"text":458,"type":29},"在相同 commit、Runner 镜像、依赖来源和资源规格下，分别运行启用代理与不启用代理的任务。冷缓存和暖缓存应分开比较，并使用隔离的缓存目录；每组可重复运行并保留原始结果。除代理开关外，其他条件尽量保持一致，便于判断差异来源。",{"type":38,"level":21,"children":460},[461],{"text":462,"type":29},"区分正常的耗时变化与构建产物差异",{"type":25,"children":464},[465],{"text":466,"type":29},"代理会让下载耗时发生变化，但不能预设“代理一定更慢”。判断是否影响构建，用各阶段耗时分布和事前设定的预算，不用单次耗时下结论；产物对照的前提是构建可复现、非确定性元数据已规范化；做不到时改比依赖清单、测试结果、关键文件和构建日志，不能仅凭整体哈希变化判定代理造成回归。",{"type":25,"children":468},[469],{"text":470,"type":29},"两类误判最常见：`NO_PROXY` 漏配，内部制品库被送进代理，出现 407 或超时；缓存掩盖失败，上一次的缓存让本次跳过了真实下载，清缓存复跑才暴露问题。",{"type":38,"level":21,"children":472},[473],{"text":474,"type":29},"记录 CI 代理测试结果并设定验收条件",{"type":25,"children":476},[477],{"text":478,"type":29},"记录 Runner 镜像、commit、实际生效的代理配置和各阶段结果，并注明缓存命中情况。被缓存跳过的步骤应标为未覆盖，必要时单独补跑。所有相关阶段通过后，再依据构建是否可复现判断产物差异。若影响其他流水线、日志出现凭据或耗时超过预算，应停止并回滚配置。",{"type":25,"children":480},[481,483,486],{"text":482,"type":29},"CI 中出现连接失败时，可从",{"text":251,"type":29,"marks":484},[485],{"href":254,"type":75},{"text":487,"type":29},"查找官方接入说明。",{"type":25,"children":489},[490],{"text":491,"type":29},"相关机制说明：",{"type":122,"format":123,"children":493},[494,501],{"type":126,"children":495},[496],{"text":497,"type":29,"marks":498},"`npm ci` 的安装与校验行为",[499],{"href":500,"type":75},"https:\u002F\u002Fdocs.npmjs.com\u002Fcli\u002Fv10\u002Fcommands\u002Fnpm-ci",{"type":126,"children":502},[503],{"text":504,"type":29,"marks":505},"GitHub Actions 依赖缓存",[506],{"href":507,"type":75},"https:\u002F\u002Fdocs.github.com\u002Factions\u002Fusing-workflows\u002Fcaching-dependencies-to-speed-up-workflows",{"type":38,"level":21,"children":509},[510],{"text":136,"type":29},{"type":38,"level":138,"children":512},[513],{"text":514,"type":29},"CI 里设了 HTTP_PROXY 为什么还是不走代理？",{"type":25,"children":516},[517],{"text":518,"type":29},"可能被 NO_PROXY 覆盖，或包管理器有独立配置，以实际生效的配置为准。",{"type":38,"level":138,"children":520},[521],{"text":522,"type":29},"构建成功就说明代理没影响吗？",{"type":25,"children":524},[525],{"text":526,"type":29},"不一定。缓存可能掩盖下载失败，需要清缓存复跑并比对产物。",[528],{"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":529,"featured":19,"sortOrder":20,"readingTime":21,"bodyType":22,"bodyBlocks":530,"bodyHtml":154},[],[531,538,541,544,547,550,553,556,559,566,569,572,575,578,581,584,587,594,597,604,607,610,613,616],{"type":25,"children":532},[533,534,537],{"text":28,"type":29},{"text":31,"type":29,"marks":535},[536],{"type":34},{"text":36,"type":29},{"type":38,"level":21,"children":539},[540],{"text":41,"type":29},{"type":25,"children":542},[543],{"text":45,"type":29},{"type":25,"children":545},[546],{"text":49,"type":29},{"type":25,"children":548},[549],{"text":53,"type":29},{"type":25,"children":551},[552],{"text":57,"type":29},{"type":38,"level":21,"children":554},[555],{"text":61,"type":29},{"type":25,"children":557},[558],{"text":65,"type":29},{"type":25,"children":560},[561,562,565],{"text":69,"type":29},{"text":71,"type":29,"marks":563},[564],{"href":74,"type":75},{"text":77,"type":29},{"type":38,"level":21,"children":567},[568],{"text":81,"type":29},{"type":25,"children":570},[571],{"text":85,"type":29},{"type":38,"level":21,"children":573},[574],{"text":89,"type":29},{"type":25,"children":576},[577],{"text":93,"type":29},{"type":25,"children":579},[580],{"text":97,"type":29},{"type":38,"level":21,"children":582},[583],{"text":101,"type":29},{"type":25,"children":585},[586],{"text":105,"type":29},{"type":25,"children":588},[589,590,593],{"text":109,"type":29},{"text":111,"type":29,"marks":591},[592],{"href":114,"type":75},{"text":116,"type":29},{"type":25,"children":595},[596],{"text":120,"type":29},{"type":122,"format":123,"children":598},[599],{"type":126,"children":600},[601],{"text":129,"type":29,"marks":602},[603],{"href":132,"type":75},{"type":38,"level":21,"children":605},[606],{"text":136,"type":29},{"type":38,"level":138,"children":608},[609],{"text":141,"type":29},{"type":25,"children":611},[612],{"text":145,"type":29},{"type":38,"level":138,"children":614},[615],{"text":149,"type":29},{"type":25,"children":617},[618],{"text":153,"type":29},null,[],[]]