Overview¶
The prometheus-stack chart in ovesorg/deployment-charts is a vendored copy
of the upstream kube-prometheus-stack Helm chart. Installing it gives you a
complete, opinionated Kubernetes monitoring solution in one shot.
What problem does it solve?¶
Running Prometheus on Kubernetes "by hand" means stitching together at least the following:
- Prometheus server (with the right scrape configuration for the kubelet, API server, etcd, scheduler, controller-manager, kube-proxy, CoreDNS)
- Alertmanager (with routing, receivers, silences)
- Grafana (datasource wiring, dashboards, sidecars)
kube-state-metricsandnode-exporter(and a way to scrape them)- Alerting rules and recording rules
- TLS, RBAC, ServiceAccounts, NetworkPolicies, PodDisruptionBudgets
The chart wraps all of this. You get a single helm install that deploys the
Prometheus Operator plus a set of Custom Resources (Prometheus,
Alertmanager, ServiceMonitor, PodMonitor, PrometheusRule, ...). The
Operator then materializes those CRs into actual StatefulSets and
Deployments and keeps them reconciled.
What is not in the chart¶
The chart focuses on cluster-level monitoring with Prometheus. It is not:
- A logging stack (use Loki / Elastic / Splunk separately)
- A tracing stack (use Tempo / Jaeger / OpenTelemetry separately)
- A long-term metric store (the chart can integrate with Thanos / Cortex / Mimir, but does not deploy them)
- A SIEM or APM product
High-level mental model¶
flowchart LR
subgraph User
H[helm install] --> CRDs
H --> CRs[Custom Resources<br/>Prometheus / Alertmanager / ...]
end
subgraph Operator
OP[Prometheus Operator<br/>Deployment]
end
subgraph Workloads
P[Prometheus<br/>StatefulSet]
A[Alertmanager<br/>StatefulSet]
G[Grafana<br/>Deployment]
KSM[kube-state-metrics<br/>Deployment]
NE[node-exporter<br/>DaemonSet]
end
CRDs --> OP
CRs --> OP
OP -->|reconciles| P
OP -->|reconciles| A
P -->|scrapes| KSM
P -->|scrapes| NE
P -->|scrapes| KubeAPI[Kubernetes API server]
P -->|scrapes| Kubelet[Kubelet / cAdvisor]
P -->|sends alerts| A
G -->|queries| P
In practice you only ever interact with:
- The chart's
values.yaml(during install / upgrade). - Kubernetes Custom Resources of kinds
ServiceMonitor,PodMonitor,PrometheusRule, and occasionallyAlertmanagerConfig. - The Grafana, Prometheus and Alertmanager UIs.
The Operator handles everything else.
What this guide covers¶
- Architecture — what gets deployed and how the pieces talk to each other.
- Components — what each subchart and template group is responsible for.
- Installation — prerequisites and the recommended install command.
- Configuration — the most-edited parts of
values.yaml. - Accessing services — reaching the Prometheus, Alertmanager, and Grafana UIs.
- Customization — adding your own alerts, dashboards, and scrape targets.
- Operations — upgrades, backups, and troubleshooting.
- Reference — terminology, external references, and the values-reference map.