Archive / 全站文章庫
主線是 KAI K8s之路，其餘內容都該繞著它讀
這裡不是把所有文章平鋪展示，而是先把雙語系列 KAI K8s之路 / KAI's K8s Journey 放在前景。其他 Kubernetes、GitOps、 observability 與事故筆記，則作為 field notes 穿插更新。
Series spine
KAI K8s之路 / KAI's K8s Journey
Chapter 01 ZH + EN
KAI K8s之路 01｜先別急著背資源，先搞懂 Pod 到底在管什麼
這篇先從 Kubernetes 最核心的抽象開始：為什麼它不是直接管理單個 container，而是用 Pod 當作最小調度單位。
KAI K8s之路 01｜先別急著背資源，先搞懂 Pod 到底在管什麼
這篇先從 Kubernetes 最核心的抽象開始：為什麼它不是直接管理單個 container，而是用 Pod 當作最小調度單位。
EnglishKAI's K8s Journey 01: Before Deployments, understand what a Pod is managing
Start with the core Kubernetes abstraction first: why the platform schedules Pods instead of treating each container as the main unit of orchestration.
中文內容
中文主線與補充筆記
先把概念講清楚，再一路走到 rollout、GitOps 與真實環境裡的坑。
查看全部中文文章English track
English notes with production bias
Keep the series readable in English while leaving room for operational notes.
Browse English postsField notes
現場補充、小知識、事故趣事與建議
一次探針誤判，如何把服務打進重啟循環
有些事故不是程式壞掉，而是健康檢查把原本能活下來的服務判成該重啟。這篇記一個很典型的坑。
GitOps review checks I want before Argo CD touches production
A compact pre-sync checklist for GitOps changes so an innocent YAML edit does not become surprising production behavior.
Kubernetes rollout guardrails that catch bad deploys early
A compact baseline for probes, rollout settings, and quick triage before a Kubernetes release becomes an outage.
Agent runtime observability needs spans, tool events, and failure memory
Basic logging is not enough for agent systems; you need traces, tool-level outcomes, and a short path from failure back to prompt or policy.