alerting-stacklisted

# Walter-VM alerting stack — the simple version PagerDuty and Opsgenie are overkill for a one-operator self-hosted setup. You want: "ping me on Telegram if something breaks". Four tiers cover every realistic failure mode. ## The bot (one bot, all tiers) You already have: - Telegram bot via @BotFather (skill: `telegram-bot-cli`) - Bot token in `WALTER_TELEGRAM_BOT_TOKEN` - Your chat ID in `WALTER_TELEGRAM_CHAT_ID` Every tier below pushes to the same bot. One pinned chat, one mute toggle, one "shut up" button when you're sleeping. ## Tier 1 — Service uptime (Uptime-Kuma → Telegram) **Covers**: any HTTP/TCP/DNS service down, including Plane / Forgejo / LiteLLM / Infisical / n8n / public sites you care about. **Setup** (in https://status.${WALTER_DOMAIN}): 1. Settings → **Notifications** → **Setup Notification** 2. Type: **Telegram** 3. Friendly name: `walter-bot` 4. Bot Token: paste `WALTER_TELEGRAM_BOT_TOKEN` 5. Chat ID: paste `WALTER_TELEGRAM_CHAT_ID` 6. Click **Test** — should arrive instantly. 7. **Default Enabled** + **Apply on all existing monitors** — checked. 8. Save. Then add monitors for: | Monitor | Type | Target | Interval | |---|---|---|---| | Plane | HTTP(s) keyword | https://plane.${WALTER_DOMAIN}, expect 200/302/CF Access redirect | 60s | | Forgejo | HTTP(s) | https://git.${WALTER_DOMAIN} | 60s | | LiteLLM | HTTP(s) | https://llm.${WALTER_DOMAIN}/health | 60s | | Infisical | HTTP(s) | https://secrets.${WALTER_DOMAIN} | 60s | | n8n | HTTP(s) | https://n8