4020b734a5
All checks were successful
CD Pipeline / deploy (push) Successful in 2m34s
E-1: K8s configmap 對齊三主機級聯（清舊配置遺留） - k8s/02-configmap.yaml: - 移除寫死的 OLLAMA_HOST=192.168.0.188:11434（早已過時） - 改注入 OLLAMA_HOST_PRIMARY/SECONDARY/FALLBACK 三個變數 - llama3:70b → qwen3:8b（與目前 prod 一致） - k8s/gcp/02-configmap.yaml: - 移除「OLLAMA_HOST=空 + AI_PROVIDER=gemini」自相矛盾 - 改三主機級聯 + AI_PROVIDER=ollama（與 docker-compose.yml 一致） E-2: ppt_audit 加 RAG「相似失敗修法」 - 對最近 3 筆 audit_status='failed'/'error' 跑 rag_service.query - 從 ai_insights 召回相似案例（cosine ≥ 0.6, top 2） - template 加「RAG 自動修法建議」card E-3: Telegram obs_health/obs_budget 加 inline action button (L2) - cmd:obs_health：任一主機標記異常時，inline 顯示「🩹 修 GCP-A/GCP-B/111」按鈕 - cmd:obs_heal:{label}：新 handler，在 Telegram 內直接觸發 AutoHeal （白名單 GCP-A/GCP-B/111，非 unhealthy 拒絕） - cmd:obs_budget：warn 時 inline 顯示「⚡ 立即重算節流狀態」按鈕 - cmd:obs_force_throttle：新 handler，Telegram 內直接觸發 evaluate_throttle_status 最終升級對應： - RAG 整合 4/6 → 5/6（補 ppt_audit） - AI 自動化：Web L2×4 + Telegram L2×2 = 6 個一鍵自動化入口 - K8s 配置一致性：三套 (docker-compose / k8s/* / k8s/gcp/*) 統一 指向 GCP-A → GCP-B → 111 三主機鏈 Phase 38+39+40+41 累計 7 commits 觀測台戰役完整收官。 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
66 lines
2.0 KiB
YAML
66 lines
2.0 KiB
YAML
# =============================================================================
|
||
# WOOO TECH - Momo Pro System
|
||
# Kubernetes ConfigMap
|
||
# =============================================================================
|
||
apiVersion: v1
|
||
kind: ConfigMap
|
||
metadata:
|
||
name: momo-config
|
||
namespace: momo
|
||
data:
|
||
# Telegram
|
||
TELEGRAM_CHAT_IDS: '["5619078117","7664770082","8184109358"]'
|
||
|
||
# LINE
|
||
LINE_GROUP_ID: "Cd150747b69821dfd209f6e5aa99c84ff"
|
||
|
||
# Email
|
||
EMAIL_HOST: "smtp.gmail.com"
|
||
EMAIL_PORT: "587"
|
||
EMAIL_HOST_USER: "yingpin_chen@pchome.tw"
|
||
EMAIL_SENDER: "yingpin_chen@pchome.tw"
|
||
EMAIL_RECEIVER: "yingpin831@gmail.com"
|
||
|
||
# PostgreSQL 連線設定 (config.py 使用的變數)
|
||
POSTGRES_HOST: "momo-postgres"
|
||
POSTGRES_PORT: "5432"
|
||
POSTGRES_USER: "momo"
|
||
POSTGRES_DB: "momo_analytics"
|
||
|
||
# 應用設定
|
||
USE_DATABASE_AUTH: "false"
|
||
USE_POSTGRESQL: "true"
|
||
DISABLE_AUTO_IMPORT: "false"
|
||
USE_HTTPS: "false"
|
||
PYTHONUNBUFFERED: "1"
|
||
FLASK_APP: "app.py"
|
||
DISABLE_LOGIN: "true"
|
||
|
||
# 密碼策略
|
||
PASSWORD_MIN_LENGTH: "4"
|
||
PASSWORD_REQUIRE_UPPERCASE: "false"
|
||
PASSWORD_REQUIRE_LOWERCASE: "false"
|
||
PASSWORD_REQUIRE_DIGIT: "false"
|
||
PASSWORD_REQUIRE_SPECIAL: "false"
|
||
|
||
# Ollama AI 服務（ADR-027 三主機級聯：GCP-A → GCP-B → 111）
|
||
# 不再寫死 OLLAMA_HOST，改由 services/ollama_service.py:resolve_ollama_host()
|
||
# 動態解析。三主機 env 由 OLLAMA_HOST_PRIMARY/SECONDARY/FALLBACK 控制。
|
||
OLLAMA_HOST_PRIMARY: "http://34.143.170.20:11434"
|
||
OLLAMA_HOST_SECONDARY: "http://34.21.145.224:11434"
|
||
OLLAMA_HOST_FALLBACK: "http://192.168.0.111:11434"
|
||
OLLAMA_MODEL: "qwen3:8b"
|
||
|
||
# Google Gemini AI 服務
|
||
# GEMINI_API_KEY: 請在 K8s Secret 中設定
|
||
GEMINI_MODEL: "gemini-1.5-flash"
|
||
AI_PROVIDER: "ollama"
|
||
|
||
# YouTube API (趨勢爬蟲)
|
||
# YOUTUBE_API_KEY: 請在 K8s Secret 中設定
|
||
|
||
# 外部服務連結（導航列顯示）
|
||
METABASE_URL: "https://monitor.wooo.work/metabase"
|
||
SUPERSET_URL: "https://monitor.wooo.work/superset"
|
||
GRIST_URL: "https://grist.wooo.work"
|