Kubernetes infrastructure is dynamic by design — pods restart, deployments scale, nodes come and go. Traditional monitoring tools built around static hostnames struggle in that environment, which is exactly the gap kube-prometheus-stack (sometimes shortened to just "the kubernetes stack") was built to close.
This category is the big-picture entry point: what full-stack Kubernetes observability actually requires, how metrics, dashboards, and alerting fit together, and why this particular combination of tools became the default choice for teams monitoring production Kubernetes clusters.
Concretely, "full-stack" here means four things working together: metric collection at every layer (nodes, Kubernetes objects, and application-level custom metrics), pre-built visualization that requires no PromQL to get value from on day one, alert routing that reaches the right on-call channel, and a declarative, CRD-driven configuration model that fits naturally into GitOps workflows. Platform engineers and SREs use this category as the conceptual map before diving into any single component in depth.
Discover what kube-prometheus-stack is, how it works, and why DevOps teams use it for production-ready Kubernetes observability. Prometheus, Grafana, Alertmanager explained.
Read ArticleNo — it's the most widely adopted open-source option, but managed alternatives (Datadog, New Relic) and other open-source stacks exist; see our Comparisons category for the trade-offs.
Node-level resource metrics, Kubernetes object state (pods, deployments, jobs), and application-level custom metrics — kube-prometheus-stack covers all three out of the box via Node Exporter, kube-state-metrics, and ServiceMonitors.
