O-2.1: OTEL Collector DaemonSet (filelog receiver) - 收集所有 K3s 節點 Pod stdout/stderr → SigNoz ClickHouse - CRI log parser (Go time layout for +08:00 timezone) - filter processor 排除 kube-system debug noise - observability namespace PSA privileged (log 目錄需 root) - 資源限制: 50m-200m CPU / 64-128Mi Memory O-2.2: kubernetes-event-exporter - K8s Event → 結構化 JSON Log → SigNoz - Warning/Error 全量保留, Normal 過濾高頻事件 - 解決: Event 預設僅保留 ~1hr 的致命盲區 O-3: Prometheus remote_write 配置模板 - 白名單: ~50 關鍵 metric series (node/container/kube/api/db) - 目標: 90 天長期儲存於 SigNoz ClickHouse 已部署驗證: 3 Pod Running, 0 error, filelog 正常監控所有 namespace Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
4.6 KiB
4.6 KiB