observability-stack/crds/prometheus-operator/README.md

Prometheus Operator CRDs

This directory contains the CustomResourceDefinitions (CRDs) for the Prometheus Operator.

Why Separate CRDs?

These CRDs are managed separately from the main kube-prometheus-stack Helm chart to avoid:

1. Large Annotation Error: Helm stores the full manifest in annotations, which can exceed Kubernetes' 262144-byte limit for CRDs
2. CRD Installation Race Conditions: Ensures CRDs are installed before any resources that depend on them

Deployment Method

These CRDs are manually deployed using kubectl server-side apply:

kubectl apply --server-side=true --force-conflicts -f manifests/prometheus-operator-crds/

Note: Server-side apply is required to bypass the annotation size limit that affects large CRDs.

CRDs Included

• crd-alertmanagerconfigs.yaml - AlertmanagerConfig resources
• crd-alertmanagers.yaml - Alertmanager resources
• crd-podmonitors.yaml - PodMonitor resources
• crd-probes.yaml - Probe resources
• crd-prometheusagents.yaml - PrometheusAgent resources
• crd-prometheuses.yaml - Prometheus resources
• crd-prometheusrules.yaml - PrometheusRule resources
• crd-scrapeconfigs.yaml - ScrapeConfig resources
• crd-servicemonitors.yaml - ServiceMonitor resources
• crd-thanosrulers.yaml - ThanosRuler resources

Updating CRDs

When updating the kube-prometheus-stack chart version:

1. Copy the new CRDs from charts/kube-prometheus-stack/charts/crds/crds/ to this directory
2. Reapply them: kubectl apply --server-side=true --force-conflicts -f manifests/prometheus-operator-crds/
3. Then update the main stack via ArgoCD

Related Configuration

• Main stack values: manifests/kube-prometheus-stack/values.yaml has crds.enabled: false
• Main ArgoCD app: argocd-apps/kube-prometheus-stack.yaml