# Rate Limit Monitoring Setup

This document explains how to set up automated rate limit monitoring with alerts.

## Overview

The rate limit monitoring system consists of:
1. **Metrics Collection** - Tracks all rate limit checks in-memory
2. **Alert Configuration** - Database table with configurable thresholds
3. **Monitor Function** - Edge function that checks metrics and triggers alerts
4. **Cron Job** - Scheduled job that runs the monitor function periodically

## Setup Instructions

### Step 1: Enable Required Extensions

Run this SQL in your Supabase SQL Editor:

```sql
-- Enable pg_cron for scheduling
CREATE EXTENSION IF NOT EXISTS pg_cron;

-- Enable pg_net for HTTP requests
CREATE EXTENSION IF NOT EXISTS pg_net;
```

### Step 2: Create the Cron Job

Run this SQL to schedule the monitor to run every 5 minutes:

```sql
SELECT cron.schedule(
  'monitor-rate-limits',
  '*/5 * * * *', -- Every 5 minutes
  $$
  SELECT
    net.http_post(
        url:='https://api.thrillwiki.com/functions/v1/monitor-rate-limits',
        headers:='{"Content-Type": "application/json", "Authorization": "Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZSIsInJlZiI6InlkdnRtbnJzenlicW5iY3FiZGN5Iiwicm9sZSI6ImFub24iLCJpYXQiOjE3NTgzMjYzNTYsImV4cCI6MjA3MzkwMjM1Nn0.DM3oyapd_omP5ZzIlrT0H9qBsiQBxBRgw2tYuqgXKX4"}'::jsonb,
        body:='{}'::jsonb
    ) as request_id;
  $$
);
```

### Step 3: Verify the Cron Job

Check that the cron job was created:

```sql
SELECT * FROM cron.job WHERE jobname = 'monitor-rate-limits';
```

### Step 4: Configure Alert Thresholds

Visit the admin dashboard at `/admin/rate-limit-metrics` and navigate to the "Configuration" tab to:

- Enable/disable specific alerts
- Adjust threshold values
- Modify time windows

Default configurations are automatically created:
- **Block Rate Alert**: Triggers when >50% of requests are blocked in 5 minutes
- **Total Requests Alert**: Triggers when >1000 requests/minute
- **Unique IPs Alert**: Triggers when >100 unique IPs in 5 minutes (disabled by default)

## How It Works

### 1. Metrics Collection

Every rate limit check (both allowed and blocked) is recorded with:
- Timestamp
- Function name
- Client IP
- User ID (if authenticated)
- Result (allowed/blocked)
- Remaining quota
- Rate limit tier

Metrics are stored in-memory for the last 10,000 checks.

### 2. Monitoring Process

Every 5 minutes, the monitor function:
1. Fetches enabled alert configurations from the database
2. Analyzes current metrics for each configuration's time window
3. Compares metrics against configured thresholds
4. For exceeded thresholds:
   - Records the alert in `rate_limit_alerts` table
   - Sends notification to moderators via Novu
   - Skips if a recent unresolved alert already exists (prevents spam)

### 3. Alert Deduplication

Alerts are deduplicated using a 15-minute window. If an alert for the same configuration was triggered in the last 15 minutes and hasn't been resolved, no new alert is sent.

### 4. Notifications

Alerts are sent to all moderators via the "moderators" topic in Novu, including:
- Email notifications
- In-app notifications (if configured)
- Custom notification channels (if configured)

## Monitoring the Monitor

### Check Cron Job Status

```sql
-- View recent cron job runs
SELECT * FROM cron.job_run_details 
WHERE jobid = (SELECT jobid FROM cron.job WHERE jobname = 'monitor-rate-limits')
ORDER BY start_time DESC 
LIMIT 10;
```

### View Function Logs

Check the edge function logs in Supabase Dashboard:
`https://supabase.com/dashboard/project/ydvtmnrszybqnbcqbdcy/functions/monitor-rate-limits/logs`

### Test Manually

You can test the monitor function manually by calling it via HTTP:

```bash
curl -X POST https://api.thrillwiki.com/functions/v1/monitor-rate-limits \
  -H "Content-Type: application/json"
```

## Adjusting the Schedule

To change how often the monitor runs, update the cron schedule:

```sql
-- Update to run every 10 minutes instead
SELECT cron.alter_job('monitor-rate-limits', schedule:='*/10 * * * *');

-- Update to run every hour
SELECT cron.alter_job('monitor-rate-limits', schedule:='0 * * * *');

-- Update to run every minute (not recommended - may generate too many alerts)
SELECT cron.alter_job('monitor-rate-limits', schedule:='* * * * *');
```

## Removing the Cron Job

If you need to disable monitoring:

```sql
SELECT cron.unschedule('monitor-rate-limits');
```

## Troubleshooting

### No Alerts Being Triggered

1. Check if any alert configurations are enabled:
```sql
SELECT * FROM rate_limit_alert_config WHERE enabled = true;
```

2. Check if metrics are being collected:
   - Visit `/admin/rate-limit-metrics` and check the "Recent Activity" tab
   - If no activity, the rate limiter might not be in use

3. Check monitor function logs for errors

### Too Many Alerts

- Increase threshold values in the configuration
- Increase time windows for less sensitive detection
- Disable specific alert types that are too noisy

### Monitor Not Running

1. Verify cron job exists and is active
2. Check `cron.job_run_details` for error messages
3. Verify edge function deployed successfully
4. Check network connectivity between cron scheduler and edge function

## Database Tables

### `rate_limit_alert_config`
Stores alert threshold configurations. Only admins can modify.

### `rate_limit_alerts`
Stores history of all triggered alerts. Moderators can view and resolve.

## Security

- Alert configurations can only be modified by admin/superuser roles
- Alert history is only accessible to moderators and above
- The monitor function runs without JWT verification (as a cron job)
- All database operations respect Row Level Security policies

## Performance Considerations

- In-memory metrics store max 10,000 entries (auto-trimmed)
- Metrics older than the longest configured time window are not useful
- Monitor function typically runs in <500ms
- No significant database load (simple queries on small tables)

## Future Enhancements

Possible improvements:
- Function-specific alert thresholds
- Alert aggregation (daily/weekly summaries)
- Custom notification channels per alert type
- Machine learning-based anomaly detection
- Integration with external monitoring tools (Datadog, New Relic, etc.)