Redis becomes mandatory infrastructure (PostgreSQL-parity): the
REDIS_ENABLED flag, NullCache backend, and every conditional caching
code path are removed. /health/ready reports not_ready (503) when
Redis is unreachable and cached endpoints return 503 instead of
silently degrading; invalidation helpers never raise after a
committed write.
The bundled redis container joins the core profile only via
docker-compose.dev.yml (production points REDIS_* at an external
instance, same no-shadowing pattern as postgres), the backend suite
runs against real Redis (test-redis on 127.0.0.1:55433, CI redis:8
service container, per-xdist-worker key prefixes), and operator docs
describe the required-dependency setup (upgrading.md gains the
v0.20.0 section).
- /health/ready returns 503 (not 200) when not ready; body no longer leaks
raw backend error text (api + web tiers)
- SYSTEM_MAINTENANCE now gates the web /api/* proxy (503 MAINTENANCE)
instead of only hiding the SPA UI
- channel mutations invalidate messages/packets/packet_groups/dashboard
caches so visibility/key changes apply immediately (was stale up to 1h)
- async DB engine: pool_size/max_overflow/pre_ping matching the sync
engine (survives Postgres restarts); adispose() awaited from the API
lifespan (async connections were leaked by the sync dispose)
- webhooks: permanent 4xx (except 408/429) fail fast instead of retrying;
dispatch queue bounded at 1000 with drop-oldest + warning; shutdown
logs dropped events; matches_event evaluated once
- SPA: apiGet/apiPostForm redirect to OIDC login on 401; chart date
labels and route timestamps rendered in UTC/configured timezone
- e2e compose collector healthcheck uses 'meshcore-hub health collector'
(image has no pgrep)
- docs/upgrading.md: fold unreleased notes into v0.18.0 and document the
above
Replaces the httpx-based smoke tests in tests/e2e/ with a browser E2E suite
under e2e/, running against a self-contained throwaway stack that never
touches the local development database.
Stack & data isolation (e2e/docker-compose.test.yml):
- Own ephemeral Postgres 17 (schema via the `migrate` service / Alembic),
distinct project name + named volumes, no host DB port, no \${VAR}
interpolation. `make e2e-down` destroys everything.
- OIDC enabled with a known session secret; WEB_AUTO_REFRESH_SECONDS=2 so
polling is assertable; CONTENT_HOME mounts a test markdown page.
- Deterministic seeder (e2e/seed_data.py) run via global setup inside the
collector container: nodes/observers with `area` tags, adverts, messages
on the public (17) + a custom channel, raw packets + path hops keyed to
node prefixes, a route + health/history, profiles + adoptions, and
event_observers rows so the observer filter/badges resolve.
Auth & tests:
- Forged signed `meshcore-session` cookies (e2e/mint_session.py,
itsdangerous) for admin/member identities -> storageState; no real IdP
needed. 31 specs across 12 files cover global nav/theme, profile
menu+edit, home hero, dashboard widgets, list filters/auto-refresh/row
actions, observer toggles, the path-node overlay, map filters +
show-labels, members, markdown pages, and routes add/edit/delete
(persistence, validation, confirm dialog). workers:1 / fullyParallel:false
against the single shared backend; targeted data-testids added to React
components for stable selectors.
Fixes surfaced while running the suite on fresh Postgres:
- Migration 5e3b712ccf10 aborted on Postgres: its route-health backfill
queries the live Route model (which now has max_path_length) before that
column exists. The swallowed error left the transaction aborted, killing
the subsequent alembic_version stamp. Wrapped the backfill in a SAVEPOINT
so a failure rolls back cleanly without blocking the migration.
- Restored the full ABUSE_* env set required by the MQTT broker, and set
the web service's API_KEY to the admin key (admin writes go through the
proxy as a Bearer token).
31/31 passing; tsc (frontend+e2e), pytest (1460), and pre-commit green.
Louis King