Comprehensive Guide to Deploying and Configuring Prometheus Monitoring on Kubernetes

Monitoring is essential for reliable system operation, especially for large platforms like Kubernetes.

Prometheus, originally developed by SoundCloud and now a CNCF graduated project, replaces the older heapster‑InfluxDB‑Grafana stack with a flexible, multi‑dimensional data model, powerful query language, and simple HTTP‑based metric collection.

Overview

Key features include a metric name with key/value pairs, a flexible query language, local storage, HTTP pull model, push‑gateway support, service discovery, and rich visualization options.

Components

Prometheus Server : scrapes metrics and stores time‑series data.

exporter : exposes metrics for Prometheus to collect.

pushgateway : allows push‑based metric ingestion.

alertmanager : handles alert routing and deduplication.

Installation

Download the binary from https://prometheus.io/download and start it with a YAML configuration file, e.g.

$ ./prometheus --config.file=prometheus.yml

A minimal prometheus.yml defines global , rule_files , and scrape_configs sections.

Kubernetes Deployment

Configuration is managed with a ConfigMap, a Deployment, and a Service. RBAC objects ( ServiceAccount , ClusterRole , ClusterRoleBinding ) grant Prometheus read access to cluster resources.

Persistent storage is provided via a PersistentVolume / PersistentVolumeClaim and the pod’s securityContext ensures proper permissions.

Application Monitoring

Any service exposing a /metrics endpoint can be scraped. CoreDNS, Redis with redis-exporter , and custom exporters are demonstrated.

After updating the ConfigMap, reload Prometheus with curl -X POST "http:// :9090/-/reload" to apply new scrape jobs.