Files

T

Developer d4f95b64d4 chore(observability): externalize stack from root compose files

Manage now connects to existing Grafana/Prometheus/Alertmanager instances
and never deploys its own stack.

- docker-compose.yml / docker-compose.dev.yml: removed prometheus, loki,
  alloy, grafana, alertmanager, node-exporter services, the monitoring
  network, and observability named volumes; they now ship only backend +
  frontend. Dev frontend now joins the web network so the Vite dev proxy
  can reach the backend.
- backend: alertmanager_url default is now empty; /api/monitoring/alerts
  and /alertmanager-status return graceful "not configured" responses
  when ALERTMANAGER_URL is unset. Added not-configured tests.
- docker-compose.observability.yml: kept as the optional standalone
  example; header clarifies Manage does not deploy it.
- Removed orphaned combined monitoring/prometheus/prometheus.yml
  (standalone stack uses prometheus.standalone.yml).
- Docs (README, REQUIREMENTS decision log, monitoring-logging-design,
  observability-runbooks, context.md, MIGRATION_PLAN, frontend/README,
  CHANGELOG) updated to the connect-to-existing model.

VITE_GRAFANA_URL / VITE_PROMETHEUS_URL remain as optional frontend
deep-link overrides. .env.example still needs a manual update (safety
policy blocks assistant edits): set ALERTMANAGER_URL empty/optional and
move standalone-only vars out of the root file.

2026-06-23 21:20:07 +00:00

4.5 KiB

Raw Blame History

Changelog

All notable changes to Manage. Breaking changes are marked with BREAKING.

[Unreleased]

Changed — Observability is now external only

Removed all observability services from docker-compose.yml and docker-compose.dev.yml. They now deploy only the backend and frontend. The monitoring network and the prometheus/loki/alloy/grafana/ alertmanager/node-exporter services and their named volumes were deleted, and the GRAFANA_APP_HOST Traefik rule was removed.
Manage now connects to existing Grafana/Prometheus/Alertmanager instances and never ships its own stack. The previous in-compose stack is preserved as an optional, deploy-it-yourself example in docker-compose.observability.yml (config under monitoring/, documented in docs/observability-runbooks.md).
The backend alertmanager_url default is now empty. The /api/monitoring/alerts and /alertmanager-status endpoints return a graceful "not configured" response when ALERTMANAGER_URL is unset.
Removed the now-orphaned combined monitoring/prometheus/prometheus.yml; the standalone stack uses monitoring/prometheus/prometheus.standalone.yml.
VITE_GRAFANA_URL/VITE_PROMETHEUS_URL remain as optional frontend deep-link overrides. ALERTMANAGER_URL/ALERTMANAGER_WEBHOOK_URL are optional.

Added — Service registry

Runtime service registry persisted in the backend SQLite database. External services (Grafana, Prometheus, Jellyfin, Nextcloud, SSH task runner) are now configured in the app instead of via environment variables.
Services page (/services) to create, list, and delete service instances.
Service detail pages (/services/:serviceType/:serviceId) to edit name/enabled state, rotate secrets, and view the widgets a service provides.
Service definitions live as Pydantic modules in backend/.../integrations/, each declaring its config schema, secret fields, and widget kinds.
Multi-instance support: multiple Grafana/Jellyfin/etc. instances per type.
SSH task runner service records run history in a new service_task_runs table, shown on the runner's service page.

Changed

Dashboard widgets are now service-bound (reference a service instance + widget kind) or built-in (backups, static text). The "Add widget" flow is pick-service → pick-widget-kind → configure.
Deleting a service cascade-deletes widgets that reference it.

Security

Service secrets (API keys, tokens, passphrases) are encrypted at rest with Fernet.

BREAKING

Saved Actions (server tasks) now target ssh_tasks service instances instead of monitoring machines. The default_machine_id field on saved tasks was replaced with default_service_id; the legacy saved_task_runs table was dropped and run history now lives in service_task_runs. Re-create SSH task runner services on the Services page and re-link saved actions after upgrading.

MANAGE_ENCRYPTION_KEY is now required to start the backend. Generate one with:

python -c "from cryptography.fernet import Fernet; print(Fernet.generate_key().decode())"

The GRAFANA_URL and PROMETHEUS_URL backend environment variables were removed; Grafana/Prometheus URLs now live on service records configured in the UI. Re-create them on the Services page after upgrading.
The legacy widget/addon-pages model (/addons/:addonId, /api/widgets/types, /api/widgets/sources) was removed in favor of the service registry.
Default dashboard widget seeding was removed; a fresh install starts with an empty dashboard. Add widgets from the dashboard's edit dialog after configuring services.

Notes / follow-ups

Machine-level Jellyfin/Jellyseerr app config still powers the Media/Users/Files pages. Migrating those onto the service registry is a separate follow-up change (see openspec/changes/service-registry/design.md §12.5).

Follow-up #1 — remove dead machine Jellyfin/Jellyseerr fields

With Jellyfin/Jellyseerr now resolved from the service registry, the machine-level Jellyfin/Jellyseerr fields are dead config. Removed from dependencies.py (dead _jellyseerr_client_for; _resolve_machine simplified to SSH-only), services/settings_store.py, routers/settings.py (MachineInput), frontend types, the Settings.tsx form, and frontend test fixtures. Existing DB rows may still carry these keys in config_json; they are inert and get dropped on the next machine save. No data migration required.

4.5 KiB Raw Blame History