New to KubeDB? Please start here.

Monitoring Milvus with KubeDB

KubeDB has native support for monitoring via Prometheus. This guide will give you an overview of how monitoring works for a Milvus database.

Before You Begin

• You should be familiar with the following KubeDB concepts:
  • Milvus

How Monitoring Works

Milvus components expose Prometheus metrics on port 9091. KubeDB wires those metrics up for you through spec.monitor:

spec:
  monitor:
    agent: prometheus.io/operator
    prometheus:
      serviceMonitor:
        labels:
          release: prometheus
        interval: 10s

• spec.monitor.agent: prometheus.io/operator tells KubeDB to integrate with the Prometheus Operator.
• spec.monitor.prometheus.serviceMonitor.labels are applied to the generated ServiceMonitor. They must match the serviceMonitorSelector of your Prometheus object so the operator picks it up (here, release: prometheus).
• spec.monitor.prometheus.serviceMonitor.interval sets the scrape interval.

When monitoring is enabled, KubeDB creates and maintains:

1. The primary Milvus service, which exposes gRPC (19530), metrics (9091), and REST (8080).
2. A dedicated stats Service named <db>-stats that exposes the metrics port (9091) and carries the kubedb.com/role: stats label.
3. A ServiceMonitor named <db>-stats that selects the stats service and scrapes its metrics port at /metrics.

The Prometheus Operator then reconciles the ServiceMonitor into the running Prometheus configuration and begins scraping Milvus metrics.

In the next doc, we will see a step-by-step guide on monitoring a Milvus database using the Prometheus Operator.