Files
oikos/scripts/cutover-checklist.md
dtoro ec119566fd
Some checks failed
Desktop App / Build Linux (amd64) (push) Waiting to run
Desktop App / Attach to Release (push) Blocked by required conditions
ci / build-test (push) Has been cancelled
ci / docker-build (push) Has been cancelled
ci / web (push) Has been cancelled
scratch matrix approval notifier, add chat-native MCP approval tools
Removes the entire Matrix-based notifier (internal/notifier/) that polled
for pending approvals, sent Matrix alerts, and checked for reaction-based
approve/deny. Approval decisions now work on any chat platform (Hermes
desktop, Telegram, Discord, WhatsApp, CLI) via two new MCP tools:

- list_approvals — query pending/recent approvals by status or entity
- decide_approval — approve/deny via same API endpoint as UI + nomos

Config fields removed: MatrixHomeserver, MatrixUserID, MatrixToken,
MatrixRoomID, ApprovalHMACSecret. Docker notifier: service removed.
Approval HMAC token generation removed (unused by code).

The existing chat-assent path in nomos (cmd/nomos/assent.go) and the
control-room Approve button keep working unchanged — both call the
shared POST /api/v1/approvals/{id}/decision endpoint.
2026-08-15 20:56:28 +02:00

2.7 KiB

Cutover checklist — Phase 6: apps/105 → Docker stack on mac-mini

Status: [x] = done, [ ] = pending

Pre-cutover

  • Backup: pg_dump oikos > backups/pre-cutover-20260707.sql (145K)
  • CI green: pushed to main, .gitea/workflows/ci.yml exists
  • Deploy test: Docker stack running with api + scheduler + nomos
  • Caddy config: compose/caddy/Caddyfile.oikos pushed to dtoro/caddy-conf (ed20908). Auto-deploys to caddy (121).
  • DNS: oikos.hubris.network already resolves to 192.168.8.175 (mac-mini mesh)
  • Secrets: Infisical bootstrapped + migration complete 2026-07-07. All 11 SOPS secrets migrated to Infisical (oikos project, dev env). Machine identity oikos-api has RW access verified via Go SDK. ENCRYPTION_KEY must be 32-char raw string (docs incorrect). SOPS fallback preserved for DR. secrets-issuance decommissioned — stopped/disabled on apps/105; superseded by Infisical.
  • Watchdog: crontab entry added (every 2 min → scripts/watchdog.sh). Path fixed 2026-07-07 (was stale worktree path).

Cutover

  • Stop apps/105 services: homelab-mcp-deploy, secrets-issuance, secrets-issuance-deploy, oikos-console, oikos-console-deploy
  • Disable apps/105 services: all 5 units disabled
  • Deploy to mac-mini: Docker stack running (docker compose --profile full up -d)
  • Caddy reload: pushed to dtoro/caddy-conf — auto-deploy triggers on LXC 121.
  • DNS verify: oikos.hubris.network → 192.168.8.175

Post-cutover verification

  • ./scripts/verify-phase6.sh — all 14 checks pass
  • Nomos query: curl http://localhost:8092/query -d '{"query":"fleet health"}' → HTTP 200
  • Agent activity: curl http://localhost:8090/api/v1/agent-activity → returns data
  • Scheduler ticking: 30s ticks logged
  • Notifier polling: running
  • Watchdog tested: 2026-07-07 — full cycle verified. API down → failure counting → Matrix alert sent (!alerts:hubris.network) → API recovery → counter reset. Crontab path fixed (was stale worktree path, now /Users/dtoro/Projects/oikos). MATRIX_TOKEN wired in crontab env.

Rollback drill

  • ./scripts/rollback.sh — rehearsed 2026-07-07. Backup (674KB) → stop stack → restore DB → checkout 7ac2521 → rebuild → health OK (20 tools working) → re-deploy latest (21 tools). Full cycle verified.

Cleanup

  • Remove Gitea webhooks for apps/105 (ids 10, 11, 14) from dtoro/oikos — deleted 2026-07-08 via Gitea API
  • Archive apps/105 LXC (keep for 30 days, then destroy) → ssh hubris pct stop 105 && pct snapshot 105 archive-$(date +%Y%m%d)
  • Update auto-deploy docs — apps/105 entries marked deprecated