Metrics

Navios OTel supports OpenTelemetry metrics for monitoring your application's performance.

Enabling Metrics

app.usePlugin(defineOtelPlugin({
  serviceName: 'my-api',
  exporter: 'otlp',
  exporterOptions: {
    endpoint: 'http://localhost:4318/v1/traces',
  },
  autoInstrument: {
    http: true,
  },
  metrics: {
    enabled: true,
    requestDuration: true,  // Histogram of request durations
    errorCount: true,       // Counter of errors
  },
}))

Installation for Metrics Export

yarn add @opentelemetry/exporter-metrics-otlp-http

Default Metrics

When metrics are enabled, the following are automatically collected:

Request Duration Histogram

• Name: http.server.request.duration
• Type: Histogram
• Unit: Milliseconds
• Attributes:
  • http.method: HTTP method (GET, POST, etc.)
  • http.route: Route pattern
  • http.status_code: Response status code

Error Count

• Name: http.server.error.count
• Type: Counter
• Attributes:
  • http.method: HTTP method
  • http.route: Route pattern
  • error.type: Error type (e.g., ValidationError, InternalError)

Custom Metrics

Use the MeterToken to create custom metrics:

import { inject, Injectable } from '@navios/di'
import { MeterToken } from '@navios/otel'

@Injectable()
export class OrderService {
  private readonly meter = inject(MeterToken)

  // Counter
  private readonly ordersCreated = this.meter.createCounter('orders.created', {
    description: 'Number of orders created',
  })

  // Histogram
  private readonly orderProcessingTime = this.meter.createHistogram('orders.processing_time', {
    description: 'Time to process an order',
    unit: 'ms',
  })

  async createOrder(data: OrderData) {
    const start = Date.now()

    try {
      const order = await this.repository.create(data)

      this.ordersCreated.add(1, {
        'order.type': data.type,
        'customer.tier': data.customerTier,
      })

      return order
    } finally {
      this.orderProcessingTime.record(Date.now() - start, {
        'order.type': data.type,
      })
    }
  }
}

Metrics Configuration Options

Option	Type	Default	Description
metrics.enabled	boolean	false	Enable metrics collection
metrics.requestDuration	boolean	true	Track request duration histogram
metrics.errorCount	boolean	true	Track error count

Viewing Metrics

Grafana

Query examples for Prometheus data source:

# Average request duration
rate(http_server_request_duration_sum[5m]) / rate(http_server_request_duration_count[5m])

# Request rate
rate(http_server_request_duration_count[5m])

# Error rate
rate(http_server_error_count[5m])

Jaeger

Jaeger supports metrics through its OTLP endpoint when configured with Prometheus backend.