Skip to main content

​
Overview

Keep-rs exposes Prometheus-compatible metrics and a real-time dashboard for monitoring bot performance. The metrics server runs on port 9898 by default.

​
Endpoints

The HTTP server provides several endpoints: 
curl http://localhost:9898/metrics

​
Metrics Endpoint

URL: http://localhost:9898/metrics Returns Prometheus-formatted metrics in text/plain format. This endpoint can be scraped by Prometheus for monitoring and alerting.

​
Dashboard Endpoint

URL: http://localhost:9898/dashboard or http://localhost:9898/ Provides a web-based dashboard showing:
  • High-risk users and liquidatable positions
  • Oracle price information
  • Real-time margin status
  • Position details

​
Dashboard API

URL: http://localhost:9898/api/dashboard Returns JSON data for the dashboard, including:
  • High-risk users with margin requirements
  • Oracle prices with staleness indicators
  • Current slot and timestamps

​
Health Check

URL: http://localhost:9898/health Returns HTTP 200 OK if the server is running. Use this for container health checks.

​
Available Metrics

All metrics use the rfb_ prefix (Rust Filler Bot).

​
Transaction Metrics

# Number of transactions sent
# Labels: intent
rfb_tx_sent_total{intent="fill"} 1234
rfb_tx_sent_total{intent="liquidate"} 56

​
Fill Metrics

# Number of expected fills
# Labels: intent
rfb_fill_expected_total{intent="maker"} 500
rfb_fill_expected_total{intent="taker"} 300

​
Trigger Order Metrics

# Number of expected triggered orders
rfb_trigger_expected_total 150

# Number of actual triggered orders
rfb_trigger_actual_total 148

​
Liquidation Metrics

# Number of liquidation attempts
# Labels: type
rfb_liquidation_attempts_total{type="perp"} 45
rfb_liquidation_attempts_total{type="spot"} 12

​
Swap Quote Metrics

# Swap quote request latency in milliseconds
rfb_swap_quote_latency_ms 250

# Number of Jupiter quote failures
rfb_jupiter_quote_failures_total 5

# Number of Titan quote failures
rfb_titan_quote_failures_total 2

​
Performance Metrics

# Histogram: Slots taken to confirm transactions
# Labels: intent
rfb_tx_confirmation_slots_bucket{intent="fill",le="10"} 850
rfb_tx_confirmation_slots_bucket{intent="fill",le="20"} 1150
rfb_tx_confirmation_slots_bucket{intent="fill",le="+Inf"} 1200

​
Configuration

​
Setting Metrics Port

The metrics server port can be configured via environment variable: 
export METRICS_PORT=9898
The default port is 9898. Make sure to update port mappings if you change this value.

​
Prometheus Integration

1

Configure Prometheus

Add Keep-rs to your prometheus.yml:
prometheus.yml
scrape_configs:
  - job_name: 'keeprs'
    static_configs:
      - targets: ['localhost:9898']
    scrape_interval: 15s
2

Start Prometheus

prometheus --config.file=prometheus.yml
3

Query Metrics

Access Prometheus UI at http://localhost:9090 and query metrics:
# Transaction success rate
rate(rfb_tx_confirmed_total[5m]) / rate(rfb_tx_sent_total[5m])

# Fill efficiency
rfb_fill_actual_total / rfb_fill_expected_total

# Average confirmation time
rate(rfb_tx_confirmation_slots_sum[5m]) / rate(rfb_tx_confirmation_slots_count[5m])

​
Grafana Dashboards

Create custom dashboards to visualize bot performance:

​
Key Metrics to Monitor

Transaction Performance
  • Transaction send rate
  • Confirmation rate and time
  • Failure rate by reason
  • Compute units consumed
Fill Bot Performance
  • Fill rate (expected vs actual)
  • Trigger order execution rate
  • Market-specific fill rates
Liquidator Performance
  • Liquidation attempt rate
  • Success rate by type (perp/spot)
  • Response time to liquidatable positions
System Health
  • Swap quote latency
  • Quote provider failures
  • Oracle price freshness

​
Example Grafana Queries

sum(rate(rfb_tx_confirmed_total{result="success"}[5m])) / 
sum(rate(rfb_tx_sent_total[5m])) * 100

​
Alerting

Set up alerts for critical conditions: 
alert: HighTransactionFailureRate
expr: |
  sum(rate(rfb_tx_failed_total[5m])) / 
  sum(rate(rfb_tx_sent_total[5m])) > 0.1
for: 5m
labels:
  severity: warning
annotations:
  summary: "High transaction failure rate (>10%)"

​
Dashboard State API

The /api/dashboard endpoint returns detailed information about the liquidator bot: 
{
  "high_risk_users": [
    {
      "pubkey": "user_account_pubkey",
      "authority": "authority_pubkey",
      "total_collateral": 1000000,
      "margin_requirement": 800000,
      "free_margin": 200000,
      "free_margin_ratio": 0.25,
      "status": "HighRisk",
      "last_updated_slot": 123456789,
      "last_updated_ms": 1699999999999,
      "positions": [
        {
          "market_type": "Perp",
          "market_index": 0,
          "base_asset_amount": 1000000000,
          "quote_asset_amount": -50000000000
        }
      ]
    }
  ],
  "oracle_prices": [
    {
      "market_type": "Perp",
      "market_index": 0,
      "price": 50000000000,
      "last_updated_slot": 123456789,
      "last_updated_ms": 1699999999999,
      "age_slots": 5,
      "age_ms": 2500,
      "is_stale": false
    }
  ],
  "current_slot": 123456794,
  "last_updated_ms": 1699999999999
}

​
Margin Status Values

  • Liquidatable: User can be liquidated (total collateral < margin requirement)
  • HighRisk: User is not liquidatable but has low free margin (<20% of requirement)
  • Safe: User has sufficient margin

​
Monitoring Best Practices

  1. Set up automated alerts for transaction failures and liquidation issues
  2. Monitor fill efficiency to ensure the bot is capturing opportunities
  3. Track confirmation times to optimize priority fees
  4. Watch swap quote latency for DEX integration issues
  5. Review dashboard regularly for high-risk users and oracle health
  6. Set up log aggregation for detailed error analysis
  7. Monitor compute unit usage to optimize transaction costs
The metrics endpoint is lightweight and can be scraped frequently (every 10-15 seconds) without impacting bot performance.

Build docs developers (and LLMs) love

Get started for freeTalk to us