Fix MEDIUM findings from 2026-09-24 maintenance review

- Backend retries DB connection at startup (up to 180s) so host reboots
  no longer crash-loop it; add backend and frontend healthchecks
- Docker log rotation (json-file 10m x 3) on all services
- ntfy alerts use X-Real-IP (set by nginx after real_ip resolution)
  instead of the client-controlled first X-Forwarded-For entry
- Frontend build on Node 24 LTS with package-lock.json + npm ci;
  axios 1.20.0, vite 5.4.21
- README: backup/restore/rollback runbook, real-IP proxy trust notes
- Release-Notes/v1.1.md; version 1.1.0

Co-Authored-By: Claude Opus 5.5 <[email protected]>

README.md

@@ -49,7 +49,7 @@ A self-hosted web app for managing homeschool schedules, tracking daily learning
 | Frontend server | nginx (Docker) |
 | Backend API | FastAPI (Python 3.12) |
 | Real-time | WebSockets via FastAPI |
-| Database | MySQL 8 |
+| Database | MySQL 8.4 LTS |
 | ORM | SQLAlchemy 2.0 (async) |
 | Auth | JWT — PyJWT + passlib/bcrypt |
 | Orchestration | Docker Compose |
@@ -374,6 +374,8 @@ Nginx enforces the following rate limits (per IP):
 
 Requests over the limit receive a `429 Too Many Requests` response.
 
+Limits are keyed on the real client IP. `frontend/nginx.conf` trusts the reverse proxy and Cloudflare (`set_real_ip_from`) and walks `X-Forwarded-For` past them. **If the NPM host IP changes, or you front the app with a different proxy/CDN, update those `set_real_ip_from` lines.** Otherwise every request appears to come from the proxy and all clients share one limit. The backend reads the resolved IP from `X-Real-IP` for ntfy alerts.
+
 ### Login Lockout
 
 After 5 consecutive failed login attempts, a parent account is locked for 15 minutes. The lock clears automatically after the cooldown, or immediately when a super admin resets the user's password. The lockout threshold and duration are configured in `backend/app/routers/auth.py` (`_LOGIN_MAX_ATTEMPTS`, `_LOGIN_LOCKOUT_MINUTES`).
@@ -411,3 +413,71 @@ docker compose down -v
 # Restart without rebuilding
 docker compose up
 ```
+
+---
+
+## Backup, Restore & Rollback
+
+All persistent state is in the MySQL volume (`homeschool_mysql_data`). The live `.env` is not in git, so keep a copy somewhere safe.
+
+### Back up the database
+
+```bash
+mkdir -p ~/backups/homeschool
+docker exec homeschool_db sh -c 'exec mysqldump -uroot -p"$MYSQL_ROOT_PASSWORD" \
+  --single-transaction --routines --triggers --events --databases "$MYSQL_DATABASE"' \
+  | gzip > ~/backups/homeschool/homeschool-$(date +%Y%m%d-%H%M).sql.gz
+```
+
+Copy the dump **off the Docker host**. A backup on the same disk doesn't survive a host failure.
+
+### Restore the database from a dump
+
+```bash
+docker compose stop backend frontend
+zcat ~/backups/homeschool/<file>.sql.gz | \
+  docker exec -i homeschool_db sh -c 'exec mysql -uroot -p"$MYSQL_ROOT_PASSWORD"'
+docker compose start backend frontend
+```
+
+The dump includes `CREATE DATABASE`/`DROP TABLE` statements, so it replaces the current tables.
+
+### Before an update: record a rollback point
+
+```bash
+# App images are built locally and tagged :latest, so a rebuild overwrites them
+docker tag homeschool-backend  homeschool-backend:rollback-$(date +%Y%m%d)
+docker tag homeschool-frontend homeschool-frontend:rollback-$(date +%Y%m%d)
+# Take a database backup (above). For a MySQL version upgrade, also copy the volume,
+# because MySQL can't be downgraded in place:
+docker compose stop
+docker volume create homeschool_mysql_data_pre_<date>
+docker run --rm -v homeschool_mysql_data:/from:ro -v homeschool_mysql_data_pre_<date>:/to \
+  alpine sh -c 'cp -a /from/. /to/'
+```
+
+### Roll back an app update
+
+```bash
+docker tag homeschool-backend:rollback-<date>  homeschool-backend:latest
+docker tag homeschool-frontend:rollback-<date> homeschool-frontend:latest
+docker compose up -d --no-build
+```
+
+If the update changed the schema, restore the pre-update dump too (above).
+
+### Roll back a MySQL version upgrade
+
+1. `git checkout` the previous `image: mysql:…` line in `docker-compose.yml`.
+2. `docker compose stop`.
+3. Copy the pre-upgrade volume back over `homeschool_mysql_data`:
+   ```bash
+   docker run --rm -v homeschool_mysql_data_pre_<date>:/from:ro -v homeschool_mysql_data:/to \
+     alpine sh -c 'rm -rf /to/* && cp -a /from/. /to/'
+   ```
+   Or recreate the volume and restore the dump.
+4. `docker compose up -d`.
+
+### Releases
+
+Deployed versions are tagged `vX.Y.Z` in git, with notes in `Release-Notes/vX.Y.md`. To see what's running, compare `docker image inspect homeschool-backend --format '{{.Created}}'` to the tag date.

Release-Notes/v1.1.md

@@ -0,0 +1,36 @@
+# v1.1
+
+## v1.1.0 — 2026-09-24
+
+Maintenance release from the 2026-09-24 review (`reports/maintenance-2026-09-24.md`).
+
+### Security
+- Rate limits now apply to each client. nginx resolves the real client IP through Cloudflare → NPM (`set_real_ip_from` + `real_ip_recursive`). Before this, every request looked like it came from NPM, so all users shared one login limit.
+- ntfy alerts show the real client IP (`X-Real-IP` from nginx). They no longer use the first `X-Forwarded-For` entry, which the client controls.
+- Python dependency updates:
+  - PyJWT 2.15.0 (fixes an auth bypass)
+  - anyio 4.15.1
+  - fastapi 0.141.1 / starlette 1.7.0
+  - cryptography 50.0.1
+  - python-multipart 0.0.32
+  - alembic 1.20.0 / Mako 1.4.3
+- Base images:
+  - python 3.12.14-slim, with `apt-get upgrade` and without the unused gcc/mysqlclient headers
+  - nginx 1.30.5-alpine (stable), with `apk upgrade`
+  - Node 24 LTS for the build stage
+- axios 1.20.0 and vite 5.4.21. Frontend dependencies are now locked in `package-lock.json` and installed with `npm ci`.
+
+### Platform
+- MySQL 8.0.40 (EOL) → **8.4.11 LTS**. The data dictionary is upgraded in place, and **you can't downgrade in place**. Restore the pre-upgrade volume copy or dump instead; see README "Backup, Restore & Rollback".
+
+### Reliability
+- The backend retries the DB connection for up to 180s at startup. Before this it crash-looped after host reboots, because the restart policy ignores `depends_on`.
+- Healthchecks on backend (`/api/health`) and frontend. The db healthcheck has `start_period: 180s`.
+- Docker log rotation on all services: json-file, 10 MB × 3.
+
+### Fixes
+- The meeting alert catch-up window fix (`8e92ae6`) is now deployed.
+
+### Known / accepted
+- npm audit: vite ≤6.4.2 and esbuild advisories affect only the **dev server**, which production never runs (it serves the static build). Fixing them fully needs Vite 6.4+/7.
+- mysql:8.4.11 has upstream findings in Oracle's image: curl, bundled Python tools, and gosu's Go stdlib.

backend/app/main.py

@@ -1,5 +1,7 @@
+import asyncio
 import logging
 import random
+import time
 from contextlib import asynccontextmanager
 
 from fastapi import FastAPI, WebSocket, WebSocketDisconnect
@@ -67,8 +69,25 @@ async def _add_index_if_missing(conn, index_name: str, table: str, column: str):
             raise
 
 
+async def _wait_for_db(timeout: float = 180, interval: float = 2) -> None:
+    """Retry until MySQL accepts connections. On host reboot Docker's restart
+    policy ignores depends_on, so the backend can start before the DB is up."""
+    deadline = time.monotonic() + timeout
+    while True:
+        try:
+            async with engine.connect() as conn:
+                await conn.execute(text("SELECT 1"))
+            return
+        except OperationalError as e:
+            if time.monotonic() >= deadline:
+                raise
+            logger.info("Database not ready (%s), retrying in %ss", e.orig, interval)
+            await asyncio.sleep(interval)
+
+
 @asynccontextmanager
 async def lifespan(app: FastAPI):
+    await _wait_for_db()
     # Create tables on startup (Alembic handles migrations in prod, this is a safety net)
     async with engine.begin() as conn:
         await conn.run_sync(Base.metadata.create_all)

backend/app/routers/admin.py

@@ -9,6 +9,7 @@ from app.auth.jwt import create_admin_token, create_access_token, hash_password
 from app.config import get_settings
 from app.dependencies import get_db, get_admin_user
 from app.models.user import User
+from app.utils.client_ip import client_ip
 from app.utils.ntfy import notify
 
 router = APIRouter(prefix="/api/admin", tags=["admin"])
@@ -23,7 +24,7 @@ async def admin_login(body: dict, request: Request):
         logger.warning("Failed super-admin login attempt for username=%s", username)
         raise HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail="Invalid admin credentials")
     token = create_admin_token({"sub": "admin"})
-    ip = request.headers.get("X-Forwarded-For", request.client.host if request.client else "unknown").split(",")[0].strip()
+    ip = client_ip(request)
     ua = request.headers.get("User-Agent", "unknown")
     await notify(
         title="Homeschool Dashboard Super Admin Login",

backend/app/routers/auth.py

@@ -18,6 +18,7 @@ from app.models.user import User
 from app.models.subject import Subject
 from app.schemas.auth import LoginRequest, RegisterRequest, TokenResponse
 from app.schemas.user import UserOut
+from app.utils.client_ip import client_ip
 from app.utils.ntfy import notify
 
 router = APIRouter(prefix="/api/auth", tags=["auth"])
@@ -59,7 +60,7 @@ async def register(body: RegisterRequest, response: Response, request: Request,
     refresh = create_refresh_token({"sub": str(user.id)})
     response.set_cookie(REFRESH_COOKIE, refresh, **COOKIE_OPTS)
 
-    ip = request.headers.get("X-Forwarded-For", request.client.host if request.client else "unknown").split(",")[0].strip()
+    ip = client_ip(request)
     ua = request.headers.get("User-Agent", "unknown")
     await notify(
         title="Homeschool Dashboard New User Registered",
@@ -83,7 +84,7 @@ async def login(body: LoginRequest, response: Response, request: Request, db: As
         raise HTTPException(status_code=401, detail="Invalid credentials")
 
     now = datetime.now(timezone.utc).replace(tzinfo=None)
-    ip = request.headers.get("X-Forwarded-For", request.client.host if request.client else "unknown").split(",")[0].strip()
+    ip = client_ip(request)
     ua = request.headers.get("User-Agent", "unknown")
 
     if user.locked_until and user.locked_until > now:

backend/app/utils/client_ip.py

@@ -0,0 +1,10 @@
+from fastapi import Request
+
+
+def client_ip(request: Request) -> str:
+    """Real client IP as resolved by the frontend nginx (real_ip module).
+
+    X-Forwarded-For is client-controlled and must not be trusted here; nginx
+    sets X-Real-IP from $remote_addr after walking past trusted proxies.
+    """
+    return request.headers.get("X-Real-IP") or (request.client.host if request.client else "unknown")

docker-compose.yml

@@ -1,3 +1,9 @@
+x-logging: &default-logging
+  driver: json-file
+  options:
+    max-size: "10m"
+    max-file: "3"
+
 services:
   db:
     image: mysql:8.4.11
@@ -20,6 +26,7 @@ services:
       start_period: 180s
     mem_limit: 512m
     cpus: 1.0
+    logging: *default-logging
 
   backend:
     build: ./backend
@@ -41,8 +48,15 @@ services:
         condition: service_healthy
     networks:
       - homeschool_net
+    healthcheck:
+      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/api/health', timeout=3)"]
+      interval: 30s
+      timeout: 5s
+      retries: 3
+      start_period: 180s
     mem_limit: 512m
     cpus: 1.0
+    logging: *default-logging
 
   frontend:
     build: ./frontend
@@ -54,8 +68,14 @@ services:
       - backend
     networks:
       - homeschool_net
+    healthcheck:
+      test: ["CMD", "wget", "-q", "-O", "/dev/null", "http://127.0.0.1/"]
+      interval: 30s
+      timeout: 5s
+      retries: 3
     mem_limit: 128m
     cpus: 0.5
+    logging: *default-logging
 
 networks:
   homeschool_net:

frontend/Dockerfile

@@ -1,10 +1,10 @@
 # Stage 1: Build Vue.js app
-FROM node:20.20.1-alpine AS builder
+FROM node:24.21.0-alpine AS builder
 
 WORKDIR /app
 
-COPY package.json package-lock.json* ./
-RUN npm install
+COPY package.json package-lock.json ./
+RUN npm ci
 
 COPY . .
 RUN npm run build

frontend/package.json

@@ -1,6 +1,6 @@
 {
   "name": "homeschool-frontend",
-  "version": "1.0.0",
+  "version": "1.1.0",
   "private": true,
   "scripts": {
     "dev": "vite",
@@ -8,13 +8,13 @@
     "preview": "vite preview"
   },
   "dependencies": {
-    "axios": "1.7.7",
+    "axios": "1.20.0",
     "pinia": "2.2.4",
     "vue": "3.5.12",
     "vue-router": "4.4.5"
   },
   "devDependencies": {
     "@vitejs/plugin-vue": "5.1.4",
-    "vite": "5.4.10"
+    "vite": "5.4.21"
   }
 }

reports/maintenance-2026-09-24.md

@@ -134,7 +134,7 @@ This section was first blocked on backups. The user then asked for the HIGH find
 | Real-IP rate limiting tested | PASS | Test nginx: client A was limited after its burst, client B was unaffected, and a forged XFF prefix resolved to the true client. |
 | Deploy | PASS* | The MySQL data upgrade (80040 → 80411) took about 90s, longer than the healthcheck window, so compose aborted the dependent containers. The user started them manually. Fixed by adding `start_period: 180s` to the db healthcheck. |
 | Sections 2/3 re-run | PASS (partial) | All containers up, backend logs clean, public `/api/health` 200. Live bundle now contains `8e92ae6`. nginx logs real client IPs. The user confirmed the site is back up. |
-| Version tag / release notes | DEFERRED | The repo has no tagging or `Release-Notes/` convention yet. |
+| Version tag / release notes | PASS | `v1.1.0`, `Release-Notes/v1.1.md` |
 | Old images cleanup | DEFERRED | Keep the rollback tags and volume copy until the update has been stable for about a week. Then remove them: `docker rmi homeschool-{backend,frontend}:rollback-20260924 mysql:8.0.40` and `docker volume rm homeschool_mysql_data_pre84_20260924`. |
 
 ### Changes applied
@@ -184,13 +184,13 @@ This section was first blocked on backups. The user then asked for the HIGH find
 3. ~~Fixable HIGH/CRITICAL CVEs in all images and Python deps.~~ Fixed by the rebuild and dependency bumps.
 4. ~~MySQL 8.0 is EOL.~~ Upgraded to 8.4.11 LTS.
 
-### MEDIUM
+### MEDIUM (all resolved 2026-09-24, v1.1.0)
 5. ~~The frontend is not deployed at HEAD.~~ Resolved by the rebuild.
-6. No Docker log rotation.
-7. The backend crash-loops on every host reboot (DB startup race). The db `start_period` does **not** fix this: on reboot, Docker's restart policy ignores `depends_on`. It needs connect-retry in the backend startup. backend and frontend have no healthchecks.
-8. Node 20 (build stage) is EOL, and there is no `package-lock.json`.
-9. There is no rollback/restore runbook, and there are no release tags.
-10. Client IP in ntfy alerts can be spoofed through `X-Forwarded-For`.
+6. ~~No Docker log rotation.~~ Added `x-logging` anchor (json-file, 10m × 3) to all services.
+7. ~~The backend crash-loops on reboot, and there are no backend/frontend healthchecks.~~ `_wait_for_db()` in `main.py` retries for up to 180s. Tested: the backend started before its DB, logged retries, and came up with 0 restarts. Healthchecks were added to both services.
+8. ~~Node 20 EOL and no lockfile.~~ Now node:24.21.0-alpine with `package-lock.json` and `npm ci`. axios → 1.20.0, vite → 5.4.21. Remaining npm audit items (vite ≤6.4.2 / esbuild) are dev-server-only and accepted; the fix needs Vite 6.4+.
+9. ~~No rollback/restore runbook or release tags.~~ README now has "Backup, Restore & Rollback" and documents real-IP proxy trust. Added `Release-Notes/v1.1.md` and tagged `v1.1.0`.
+10. ~~Client IP in ntfy alerts can be spoofed.~~ New `app/utils/client_ip.py` reads `X-Real-IP`, which nginx sets after real_ip resolution. Tested: a forged XFF is ignored.
 
 ### LOW
 11. db memory is at 82% of its 512 MiB limit.
@@ -208,7 +208,7 @@ Portainer/NPM drift, functional smoke tests, ntfy test, off-host/VM backups, HTT
 - Failed (critical/high): 7. That is backups (counted once, CRITICAL), rate limiting, image CVEs, dependency CVEs, runtime EOL, log rotation and frontend drift.
 - Failed (non-critical) / WARN: 18
 - Pending user confirmation: 14
-- Deferred: 2 (release tag, old image cleanup)
-- Updates applied: 4 (real-IP rate limiting, Python deps, base images, MySQL 8.4)
+- Deferred: 1 (old image cleanup)
+- Updates applied: 9 (HIGH: real-IP rate limiting, Python deps, base images, MySQL 8.4. MEDIUM: log rotation, DB-wait + healthchecks, Node 24 + lockfile, runbook + release tag, trusted client IP)
 
-**Overall: NEEDS ATTENTION.** All HIGH findings are resolved. The CRITICAL no-backups finding is still open: the pre-update dump and volume copy are on-host only, with no schedule.
+**Overall: NEEDS ATTENTION.** All HIGH and MEDIUM findings are resolved. The CRITICAL no-backups finding is still open: the pre-update dump and volume copy are on-host only, with no schedule.