Grafana bridge fix

.gitignore

@@ -223,3 +223,6 @@ generated-days/
 # Node dependencies
 node_modules/
 .claude/settings.local.json
+
+# Superpowers (AI agent specs/plans - not for version control)
+docs/superpowers/

docs/containers/file-uploader.md

@@ -1,6 +1,6 @@
 # File uploader
 
-The file uploader is a Flask application that streams CAN CSV logs into InfluxDB 3. It exposes a simple web UI for selecting the destination bucket and monitoring progress.
+The file uploader is a Flask application that streams CAN CSV logs into InfluxDB 3. It exposes a simple web UI for selecting the destination **season** (InfluxDB table within the configured database) and monitoring progress.
 
 ## Ports
 
@@ -10,8 +10,8 @@ The file uploader is a Flask application that streams CAN CSV logs into InfluxDB
 
 | Variable | Description | Default |
 | --- | --- | --- |
-| `INFLUXDB_URL` | API endpoint for bucket discovery and writes. | `http://influxdb3:8181` |
-| `INFLUXDB_TOKEN` | Token with write access to the target bucket. | `dev-influxdb-admin-token` |
+| `INFLUXDB_URL` | API endpoint for table discovery and writes. | `http://influxdb3:8181` |
+| `INFLUXDB_TOKEN` | Token with write access to the target database. | `dev-influxdb-admin-token` |
 | `FILE_UPLOADER_WEBHOOK_URL` | Optional webhook invoked when uploads finish. | empty |
 | `SLACK_WEBHOOK_URL` | Fallback webhook if the dedicated uploader value is unset. | empty |
 
@@ -25,6 +25,6 @@ The file uploader is a Flask application that streams CAN CSV logs into InfluxDB
 ## Usage
 
 1. Visit http://localhost:8084.
-2. Choose a target bucket from the drop-down (populated from the InfluxDB API).
+2. Choose a target season (table) from the drop-down (populated from the InfluxDB API).
 3. Upload one or more CSV files exported from the vehicle logger.
 4. Monitor progress via the live event stream; notifications are sent upon completion if a webhook is configured.

installer/.env.example

@@ -3,6 +3,20 @@
 # ------------------------------------------------------------
 DBC_FILE_PATH=example.dbc
 
+# ------------------------------------------------------------
+# File uploader — team DBCs from GitHub (optional)
+# ------------------------------------------------------------
+# Fine-grained PAT or classic PAT with contents:read on Western-Formula-Racing/DBC
+GITHUB_DBC_TOKEN=
+# GITHUB_DBC_REPO=Western-Formula-Racing/DBC
+# GITHUB_DBC_BRANCH=main
+
+# Optional limits for .zip uploads (file-uploader); defaults are generous for team use
+# UPLOAD_ZIP_MAX_ARCHIVE_BYTES=2147483648
+# UPLOAD_ZIP_MAX_MEMBER_BYTES=4294967296
+# UPLOAD_ZIP_MAX_TOTAL_UNCOMPRESSED_BYTES=25769803776
+# UPLOAD_ZIP_MAX_CSV_IN_ZIP=5000
+
 # ------------------------------------------------------------
 # InfluxDB credentials
 # ------------------------------------------------------------

installer/VPS_RECOVERY.md

@@ -0,0 +1,197 @@
+# VPS Recovery Guide — OVH vps-1969c8c2
+
+## What happened
+
+The server ran out of memory (OOM). The Linux kernel started killing processes to survive, including:
+- `containerd` (Docker runtime) — taking down all containers
+- `cloudflared` binary — deleted from disk
+- `tailscaled` binary — deleted from disk
+- `containerd-shim-runc-v2` binary — deleted from disk
+
+OVH detected the unresponsive server and rebooted it into **rescue mode**.
+
+---
+
+## Step 1 — Exit rescue mode
+
+OVH boots into rescue mode automatically when the server crashes hard. You need to manually switch it back.
+
+1. Go to [OVH control panel](https://www.ovh.com/manager/) → Bare Metal Cloud → VPS → `vps-1969c8c2.vps.ovh.ca`
+2. Find the **Boot** field (shows `RESCUE`) — click the pencil/edit icon
+3. Change to **Hard disk** (normal mode)
+4. Click **Reboot**
+
+> The server will come up clean — no Docker containers will auto-start (all are `restart=no` or `restart=unless-stopped` but Docker itself won't be running until the daemon starts).
+
+---
+
+## Step 2 — Fix SSH known_hosts
+
+The rescue OS has a different host key, so SSH will warn you. After rebooting to normal mode, clear the old key:
+
+```bash
+ssh-keygen -R 148.113.191.22
+```
+
+Then connect:
+```bash
+ssh [email protected]
+# or via Tailscale:
+ssh ubuntu@ovh-daq-server
+```
+
+---
+
+## Step 3 — Restore missing binaries
+
+The OOM killer can delete binaries from disk. Check and fix each one:
+
+### cloudflared
+```bash
+sudo systemctl status cloudflared
+# If "status=203/EXEC" — binary is missing
+
+curl -L https://github.com/cloudflare/cloudflared/releases/latest/download/cloudflared-linux-amd64 \
+  -o /tmp/cloudflared
+sudo mv /tmp/cloudflared /usr/bin/cloudflared
+sudo chmod +x /usr/bin/cloudflared
+sudo systemctl restart cloudflared
+sudo systemctl status cloudflared
+```
+
+### tailscale
+```bash
+sudo systemctl status tailscaled
+# If "status=203/EXEC" — binary is missing
+
+sudo apt-get install --reinstall tailscale -y
+sudo systemctl restart tailscaled
+tailscale status
+```
+
+### containerd (Docker runtime)
+```bash
+# If Docker containers fail to start with:
+# "containerd-shim-runc-v2: file does not exist"
+
+sudo apt-get install --reinstall containerd.io -y
+sudo systemctl daemon-reload
+sudo systemctl start docker
+docker info | grep "Server Version"
+```
+
+---
+
+## Step 4 — Start the Docker stack
+
+```bash
+cd /home/ubuntu/projects/daq-server-components/installer
+docker compose up -d
+```
+
+Wait ~30 seconds for InfluxDB to become healthy, then verify:
+
+```bash
+docker ps --format 'table {{.Names}}\t{{.Status}}'
+```
+
+Expected running containers:
+| Container | Notes |
+|---|---|
+| influxdb3 | Should show `(healthy)` |
+| influxdb3-explorer | InfluxDB UI |
+| grafana | Dashboard |
+| grafana-bridge | Grafana API bridge |
+| file-uploader | |
+| data-downloader-api | Waits for influxdb3 healthy |
+| data-downloader-scanner | |
+| data-downloader-frontend | |
+| health-monitor | |
+| sandbox | |
+| code-generator | |
+| slackbot | Exits cleanly if `ENABLE_SLACK=false` |
+| startup-data-loader | Runs once then exits — normal |
+
+> `lap-detector` is intentionally disabled. To run it: `docker compose --profile disabled up lap-detector -d`
+
+---
+
+## Step 5 — Verify Cloudflare tunnel
+
+```bash
+sudo systemctl status cloudflared
+# Should show "Registered tunnel connection" in logs
+```
+
+Check that https://grafana.westernformularacing.org loads.
+
+---
+
+## Investigating an OOM crash
+
+If the server crashed again and you're in rescue mode:
+
+```bash
+# Mount original disk
+mkdir -p /mnt/vps
+mount /dev/sdb1 /mnt/vps
+
+# Check what got OOM-killed and when
+journalctl --directory=/mnt/vps/var/log/journal \
+  --since="2 hours ago" --no-pager \
+  | grep -iE "oom|killed|memory" | head -50
+
+# Check disk usage
+df -h /mnt/vps
+du -sh /mnt/vps/var/lib/docker /mnt/vps/var/lib/containerd /mnt/vps/var/log/journal
+
+# Vacuum logs if journal is large (>500MB)
+journalctl --vacuum-size=200M
+```
+
+---
+
+## Preventing OOM crashes
+
+Memory limits are now set in `docker-compose.yml`. Key limits:
+
+| Service | Limit |
+|---|---|
+| influxdb3 | 4096M |
+| file-uploader | 1536M |
+| data-downloader-api | 1024M |
+| sandbox | 1024M |
+| grafana | 512M |
+| others | 128–512M |
+
+**Total ceiling: ~9GB** across all services (server has 8GB RAM + 8GB swap).
+
+If OOM happens again, check which container hit its limit:
+```bash
+docker stats --no-stream
+# or check logs:
+sudo journalctl -u docker --since="1 hour ago" | grep -i "oom\|killed"
+```
+
+Daily restart is scheduled at 4 AM to clear any memory accumulation:
+```bash
+crontab -l  # shows: 0 4 * * * docker compose restart
+# Logs at: /var/log/docker-restart.log
+```
+
+---
+
+## Quick reference
+
+| Service | Port |
+|---|---|
+| InfluxDB | 9000 |
+| InfluxDB Explorer | 8888 |
+| Grafana | 8087 (also via Cloudflare tunnel) |
+| Grafana Bridge | 3001 |
+| File Uploader | 8084 |
+| Data Downloader API | 8000 |
+| Data Downloader Frontend | 3000 |
+| Lap Detector (disabled) | 8050 |
+
+Tailscale IP: `100.72.11.60` (hostname: `ovh-daq-server`)

installer/data-downloader/backend/config.py

@@ -24,7 +24,7 @@ def _parse_seasons(raw: str | None) -> List[SeasonConfig]:
     """Parse SEASONS env var: "WFR25:2025:222 76 153,WFR26:2026:..."."""
     if not raw:
         # Default fallback if not set
-        return [SeasonConfig(name="WFR25", year=2025, database="WFR25", table="WFR25", color="#DE4C99")]
+        return [SeasonConfig(name="WFR25", year=2025, database=os.getenv("INFLUX_DATABASE", "WFR"), table="WFR25", color="#DE4C99")]
 
     seasons = []
     for part in raw.split(","):
@@ -46,13 +46,14 @@ def _parse_seasons(raw: str | None) -> List[SeasonConfig]:
 
             color = parts[2] if len(parts) > 2 else None
 
-            # DB and table name both match season name by convention (WFR25→WFR25, WFR26→WFR26)
-            seasons.append(SeasonConfig(name=name, year=year, database=name, table=name, color=color))
+            # All seasons share one database; table name matches season name (WFR25, WFR26, etc.)
+            shared_db = os.getenv("INFLUX_DATABASE", "WFR")
+            seasons.append(SeasonConfig(name=name, year=year, database=shared_db, table=name, color=color))
         except ValueError:
             continue
 
     if not seasons:
-         return [SeasonConfig(name="WFR25", year=2025, database="WFR25", table="WFR25")]
+         return [SeasonConfig(name="WFR25", year=2025, database=os.getenv("INFLUX_DATABASE", "WFR"), table="WFR25")]
 
     # Sort by year descending (newest first)
     seasons.sort(key=lambda s: s.year, reverse=True)

installer/docker-compose.local.yml

@@ -77,6 +77,9 @@ services:
       - INFLUXDB_TOKEN=${INFLUXDB_ADMIN_TOKEN:-apiv3_dev-influxdb-admin-token}
       - INFLUXDB_URL=${INFLUXDB_URL:-http://influxdb3:8181}
       - DBC_FILE_PATH=/installer/example.dbc
+      - GITHUB_DBC_TOKEN=${GITHUB_DBC_TOKEN:-}
+      - GITHUB_DBC_REPO=${GITHUB_DBC_REPO:-Western-Formula-Racing/DBC}
+      - GITHUB_DBC_BRANCH=${GITHUB_DBC_BRANCH:-main}
     deploy:
       resources:
         limits: