Files
mg4/server/README.md
T
nilsandClaude Opus 5.5 a5db5604d3 Code review fixes: app 2.9.0, server 2.2.0
Telemetry no longer loses queued batches. The server drops a bad value,
or a sample without a plausible time, instead of rejecting the batch,
and answers 200 with what it dropped. The app sends only values inside
the server's ranges (speed as a magnitude, so reversing is fine) and no
samples before the head unit's clock is set.

Server:
- Pairing hands out and accepts a token only after the device is saved;
  a failed save leaves the request pending.
- A paired token serves only the installation id it was paired with
  (403 otherwise); the shared INGEST_TOKEN still serves any.
- Wrong-password lockouts and open pairing requests are limited per
  client address (last X-Forwarded-For entry), so a stranger cannot
  lock the owner out; a car starting again replaces its own request.
- ABRP forwarding skips samples stamped ahead of the server's clock,
  which used to stop it until a restart.

App:
- Settings written to plaintext during a keystore failure move back
  into encrypted storage; the stored installation id wins.
- The installation id is derived from ANDROID_ID on first start, so a
  reinstall finds its history; a stored or imported id still wins.
- The scanned config file is deleted after import (it holds the token).
- GPS stops while live telemetry is off or the server unconfigured, and
  pauses while parked and idle, keeping the fix the car parked with.
- Range estimates use the monotonic clock.
- One HTTP helper (ServerHttp) for telemetry, sync and pairing.
- Last Turkish strings translated; mise logs shows the NilsMG4 tags.

Grafana reads InfluxDB with its own read-only token (new grafana-influx
Secret, created by hand as described in k8s/README.md).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QLJfZQZeAKAwSqViTmLzki
2026-10-18 00:36:19 +02:00

109 lines
5.0 KiB
Markdown

# Nils' MG4 — car server
A small Go service at car.k8s.stinnesbeck.com: the API the car writes to, and the ABRP
forwarder. Data lives in InfluxDB 2; Grafana reads it from there for the dashboards
(`../k8s/grafana/`).
```
car ──live telemetry───▶ server ──▶ ABRP api.iternio.com/1/tlm/send
──daily chunks─────▶ │
──charging sessions▶ ▼
InfluxDB 2 ◀── Grafana
```
## Configuration
Environment only; the server refuses to start when a required value is missing.
| Name | Notes |
|---|---|
| `INFLUX_URL`, `INFLUX_TOKEN`, `INFLUX_ORG` | required; the token needs read/write on the bucket |
| `INFLUX_BUCKET` | default `mg4` |
| `INGEST_TOKEN` | required, at least 24 characters: `openssl rand -hex 32` |
| `ABRP_API_KEY`, `ABRP_USER_TOKEN` | optional; both needed to forward |
| `ABRP_MIN_INTERVAL_SECONDS` | default 5 |
| `APPROVAL_PASSWORD` | optional, at least 12 characters; switches pairing on (see below) |
| `PUBLIC_URL` | the address a phone uses, for the pairing QR code; derived from the request if empty |
| `LISTEN_ADDR` | default `:8080` |
The ABRP credentials never reach the car.
## Endpoints
The ingest endpoints accept `Authorization: Bearer <token>` with either the shared `INGEST_TOKEN`
or a paired car's own token. A paired token only works for the installation id it was paired
with (403 otherwise); the shared token works for any.
- `POST /v1/telemetry` — `{ installationId, samples: [...] }`, up to 500 samples in ABRP's `tlm`
field names and units (`utc` in epoch seconds; flags as 0/1 or booleans). Unknown fields are
dropped, known ones bounded: a value out of range is dropped on its own, and a sample without
a plausible time (before 2001, or more than an hour ahead) is dropped whole. The rest is stored
and the answer is `200 {stored, droppedCount, dropped}`, even when nothing was usable, since
the car would only send it again. The newest sample is forwarded to ABRP if it is under
5 minutes old and not ahead of the server's clock.
- `PUT /v1/consumption/daily` — one delta chunk of a day's consumption.
- `PUT /v1/charging-sessions/{sessionId}` — a finished DC session with its curve.
- `GET /v1/consumption/history/{installationId}` — the car reads its own totals back.
- `GET /v1/health` — `{"status":"ok"}`.
A malformed request is a 400 naming the field; a database failure is a 500 without detail.
## Pairing a car by QR code
So nobody types a token in the car. Modelled on the OAuth device authorization grant (RFC 8628);
on only when `APPROVAL_PASSWORD` is set.
1. The car calls `POST /v1/pair/start` `{installationId}` and shows the returned
`verificationUriComplete` as a QR code, plus the short `userCode` (e.g. `K7F-3QX`).
2. The phone opens `/pair?code=…`, the owner checks the code and approves with the approval
password. Five wrong passwords from one address lock that address out for a minute; other
addresses are unaffected (the address is the last `X-Forwarded-For` entry, which Traefik sets).
An address may hold three open requests; a car starting again replaces its own.
3. The car polls `POST /v1/pair/poll` `{deviceCode}`: 202 pending, 200 with its own token
(handed out once), 403 denied, 410 expired (after 10 minutes) or unknown.
Each car gets its own 43-character token; only its SHA-256 is stored (measurement `device` in
the same bucket), and the token is only handed out, and only accepted, once the device is
persisted, so a restart never loses a paired car. Manage cars with the approval password:
```sh
curl -H "Authorization: Bearer $APPROVAL_PASSWORD" https://car.k8s.stinnesbeck.com/v1/devices
curl -X POST -H "Authorization: Bearer $APPROVAL_PASSWORD" https://car.k8s.stinnesbeck.com/v1/devices/<id>/revoke
```
## Storage
One bucket, `installationId` tag on every point:
| Measurement | Timestamp | Tags | Fields |
|---|---|---|---|
| `telemetry` | sample `utc` | — | the ABRP fields that were sent |
| `consumption_chunk` | `recordedAt` | `date` | km, kwh, hours, socDrop, sohPercent |
| `charging_session` | `startedAt` | `sessionId` | endedAtMs, duration, energy, cost, start/end SOC |
| `charging_curve` | point time | `sessionId` | powerKw, socPercent |
| `device` | paired at | `deviceId` | tokenHash, installationId, revoked |
A point with the same measurement, tags and timestamp replaces the earlier one, so the car's
retries are idempotent: a cached chunk is re-sent with its original `recordedAt`. Days and the
lifetime total are summed from the chunks when read. Numbers are always written as floats,
because InfluxDB fixes a field's type on first write.
## Development
```sh
nix develop # from the repository root: Go, govulncheck, InfluxDB 2
cd server
go test ./...
govulncheck ./...
```
## Container image
```sh
podman build -t git.stinnesbeck.com/nils/mg4-server:<version> .
podman push git.stinnesbeck.com/nils/mg4-server:<version>
```
The build runs `go vet` and the tests. The image is a static binary on distroless (about
11 MB), running as UID 65532 on port 8080.