返回首页

对比阅读

对比阅读:A Client Site Got Hit by an HTTP/2 Exploit. I Patched It Overnight 与 客户官网被恶意请求触发漏洞,我连夜补上了防护

AEN
Nginx安全独立创业者网站运维·

A Client Site Got Hit by an HTTP/2 Exploit. I Patched It Overnight

That night, a client said the dashboard would not load

It was 10 p.m. on Tuesday, and I was updating the delivery page for a freelance client when Li Ting messaged me: “The site has been throwing up blank pages all day.” My first reaction was that I had broken the styles myself. It turned out Nginx had released a new version fixing an HTTP/2 handling vulnerability. Malicious requests could trigger remote code execution—not a normal page error, but someone using the server to run commands they should not have access to run. I panicked too: the machine was still connected to client data.

What it is, and why I might need to act

Simply put, Nginx is the “receptionist” many sites put at the front, forwarding visitor requests to backend programs. HTTP/2 lets a browser send multiple requests in one connection. Older versions had a vulnerability when handling certain requests, allowing attackers to bypass the normal access path and execute commands on the server. Wiz researcher Mikhail Korobov publicly disclosed the issue in January 2026; security teams generally recommend upgrading Nginx promptly rather than waiting for a hosting platform to handle it automatically.

For sites fully managed by platforms such as Vercel or Cloudflare Pages, the platform may handle the underlying upgrade. But when a client runs Nginx on a cloud server, virtual machine, or their own infrastructure, I still check the version proactively. Tencent Cloud, Alibaba Cloud, and other providers have also published fix notices and operational guidance. Not every setup needs the same response: a purely static showcase or a fully managed site with a confirmed current version can probably wait.

Today’s replication: back up, check the version, then decide

The cash cost is usually ¥0, but cloud server snapshots, temporary bandwidth, or a maintenance window can add up. Tens to hundreds of yuan is common. The check itself usually takes 20 minutes; for the upgrade, configuration review, and rollback rehearsal, I would reserve one to two hours. The technical barrier is knowing whether my site uses Nginx and being able to access the server management console. Someone who does not understand command-line tools can hand these steps to an operations specialist.

The first step is not changing code. It is logging into the cloud server console, finding “Instances” or “Cloud Servers,” opening the instance details, recording the current Nginx version, and confirming whether HTTP/2 is enabled. Next, I check the cloud provider’s security advisory to confirm whether that version includes the fix. Then I create a snapshot before following the official upgrade instructions to the fixed version. After that, I verify the website, admin dashboard, and API endpoints. If the upgrade fails, I roll back to the snapshot. The exact version numbers and commands depend on the system and cloud provider’s advisory. I nearly pasted an upgrade command blindly, but I caught myself: it was for a different operating system.

How I would handle this at each stage

Just starting out: If the site is only a portfolio or landing page, I would first ask the host whether the underlying Nginx installation has been patched, take a server snapshot, and then check the version. I could leave it for today, but I would at least know who hosts the site and who is responsible for it.

Serving one or two clients: I would prioritize customer dashboards, file uploads, and sites with authenticated sessions because those hold real data. I would send the client the check details and rollback point so they know what I did and where I could restore from if needed.

Scaling up: I would add vulnerability checks to the monthly maintenance checklist, standardize server versions, record changes, and schedule alerts. There is no need to panic over every security story. The key is establishing a repeatable “discover—back up—fix—verify” process.

来源: www.wiz.io
BZH
Nginx安全独立创业者网站运维·

客户官网被恶意请求触发漏洞,我连夜补上了防护

今天客户说后台偶尔打不开

周二晚上十点,我正在给自由职业客户改交付页,李婷忽然发来消息:“网站今天跳了好几次空白页。”我第一反应是自己改坏了样式,后来才发现 Nginx 新版本修复了 HTTP/2 处理漏洞。恶意请求可能触发远程代码执行——不是普通页面出错,而是有人借服务器执行不该执行的命令。那一刻我也慌了:这台机器还连着客户资料。

它到底是什么,为什么你可能需要处理

简单说,Nginx 是很多网站放在最前面的“接待员”,负责把访客请求转给后端程序。HTTP/2 是浏览器一次发送多个请求的通信方式。旧版本在处理特定请求时存在漏洞,攻击者可能绕过正常访问路径,在服务器上执行命令。Wiz 研究员 Mikhail Korobov 已在 2026 年 1 月公开披露相关问题;安全团队通常建议尽快升级 Nginx,而不是只等托管平台自动处理。

如果你只用 Vercel、Cloudflare Pages 这类托管平台,平台可能负责底层升级,但客户在云服务器、虚拟机或自己公司的 Nginx 上,仍应主动核对版本。腾讯云、阿里云等服务商也陆续发布修复说明和操作建议。工具不是所有人都需要:纯静态展示页、已经由平台托管且能确认版本已更新的站点,可以先不折腾。

今天复刻:先备份,再查版本,最后决定升级

钱的成本通常是 0 元,但可能产生云服务器快照、临时带宽或维护窗口费用,几十到几百元都算常见。时间上,检查通常 20 分钟;涉及升级、配置检查和回滚演练,建议预留 1—2 小时。技术门槛是:需要知道自己的网站是否由 Nginx 提供服务,并能使用服务器管理界面;完全不懂命令也没关系,可以把下面步骤交给运维人员。

第一步不是改代码,而是登录云服务器控制台,找到“实例”或“云服务器”,进入实例详情,记录当前 Nginx 版本,并确认 HTTP/2 是否启用。第二步,查看云厂商安全公告,核对对应版本是否已修复。第三步,先创建快照,再按官方说明升级到已修复版本,随后检查网站、后台和接口是否正常。若升级失败,就用快照回滚。具体版本号和命令必须以你的系统与云厂商公告为准,我之前差点直接照抄升级命令,幸好看懂那是针对另一种系统。

按你的阶段决定,别一次做太多

刚起步:如果网站只是作品集或落地页,我会先问托管商“底层 Nginx 是否已修复”,保留一份服务器快照,再做版本核对。今天不处理也行,但至少要知道自己的主机商和网站负责人是谁。

已有 1—2 个客户:我会优先排查客户后台、文件上传和带登录状态的网站,因为这些位置放着真实资料。我会把检查过程和回滚点发给客户,让对方知道我做了什么。

正在扩规模:我会把漏洞检查纳入每月维护清单,统一服务器版本、记录变更并安排告警。这里没必要追着每篇安全新闻恐慌,重点是建立“发现—备份—修复—验证”的固定流程。

来源: www.wiz.io