docs(linkedin): curate and package batch-02 articles for consultant review

docs/linkedin/01-running-production-infrastructure-solo.md

@@ -0,0 +1,56 @@
+# LinkedIn Draft & Consultant Review: Running Production Infrastructure Solo: 162 Containers Across Three Hosts
+
+> **文档性质**：从 `blog.hoelee.com` 改写的 LinkedIn 待发布技术动态。
+> **发布状态**：`待顾问审阅 (未在 LinkedIn 发布)`
+> **源文章**：https://blog.hoelee.com/posts/running-production-infrastructure-solo/
+> **系统标识**：`running-production-infrastructure-solo` | **分类**：`case-studies`
+> **字数**：2155 chars (LinkedIn 上限 3,000 chars)
+
+---
+
+## 一、 核心背景与铁律约束提示（供顾问审阅对照）
+
+1. **求职画像契合（Senior 全栈/DevOps/AI 架构师）**：严禁流露“做生意的老板/小公司创始人”倾向；一切商业实战均转换为“生产系统架构、高可用交付与自动化运维”视角。
+2. **纯洁人设与零禁用词**：绝对禁止 `homelab`, `home lab`, `家用`, `小主机`, `玩具`, `折腾`, `玩票` 等词汇。必须使用 `the infrastructure I manage`, `my self-hosted production nodes`, `production host`。
+3. **第一人称单数**：统一使用单数 `I` / `我`，杜绝虚张声势的复数 `we`。
+4. **数据绝对保真**：所有数字、性能比率、吞吐提升、故障现象严禁凭空捏造。
+5. **OPSEC 安全脱敏**：无内部绝对路径（如 `D:/dev`），无真实私有内网 IP，公网 IP 必须转为文档保留地址。
+
+---
+
+## 二、 LinkedIn 拟发布动态文案
+
+```text
+I run 162 Docker containers across three self-managed production hosts with 78 compose stacks — serving public client applications, mail infrastructure, workflow automation, and CI/CD pipelines.
+
+No multi-person ops team. No managed cloud PaaS swallowing thousands a month.
+
+When you manage an estate of that density solo, you learn quickly that conventional dashboard metrics are dangerous. Two real-world production failures changed how I audit my systems:
+
+1. The silent database dashboard disconnect
+My analytics dashboard (NocoDB) was happily displaying rows and answering HTTP 200. Downstream, the background replication worker had dropped its database socket four hours earlier. The UI was serving cached metadata from memory while live mutations stalled. A green dashboard is not health; transactional read-after-write verification is health.
+
+2. The zero-config daemon lockup
+A credential management container hung on startup after a routine patch. Docker health checks passed because PID 1 stayed alive, but reverse-proxy requests timed out with HTTP 504. The culprit? An empty GPG keyring directory in a newly mounted volume triggered a blocking interactive entropy prompt inside the daemon. Health checks that probe process existence rather than live application handshakes will always deceive you.
+
+My operational rules for running production infrastructure solo:
+• Segregate workloads by isolation boundary: public ingress and edge routing live on a dedicated VPS; core storage and identity live behind private gateways.
+• Telemetry must be completely decoupled from application runtimes.
+• If an alert is based on an arbitrary percentage ratio instead of absolute time-to-exhaustion, delete it.
+
+When you operate both the codebase and the host it runs on, there is no finger-pointing between the developer and the infrastructure provider. You own the transport layer as completely as the application layer.
+
+For those running dense container footprints: what is your primary watchdog test to catch a service that is technically "running" but practically dead?
+
+Write-up: https://blog.hoelee.com/posts/running-production-infrastructure-solo/
+```
+
+---
+
+## 三、 外部顾问审阅评估反馈单
+
+- [ ] **审阅结论**：[ ] 通过 / [ ] 需微调 / [ ] 需重构 / [ ] 建议暂缓
+- **1. 职场形象评估（是否符合高级技术雇佣人设，无老板/卖货感）**：
+- **2. 技术真实性与逻辑推导（因果链条与数据是否严谨）**：
+- **3. 语言地道度与人设去油（是否纯正地道、零业余玩具感）**：
+- **4. 优化修改建议（如有优化建议或修改文案，请在此批注）**：