infra/stacks/platform/modules/kyverno/main.tf
Viktor Barzin 31f3fc0773
[ci skip] fix OOMKill: prometheus (4Gi), kyverno-reports (512Mi), grampsweb (512Mi)
- Prometheus server: explicit 1Gi req / 4Gi limit (was inheriting 512Mi LimitRange default)
- Kyverno reports controller: 128Mi req / 512Mi limit (was 128Mi Helm default)
- Grampsweb: 256Mi req / 512Mi limit for both containers (was 256Mi LimitRange default)
2026-03-02 21:39:14 +00:00

181 lines
4.4 KiB
HCL

resource "kubernetes_namespace" "kyverno" {
metadata {
name = "kyverno"
labels = {
"istio-injection" : "disabled"
}
}
}
resource "helm_release" "kyverno" {
namespace = kubernetes_namespace.kyverno.metadata[0].name
create_namespace = false
name = "kyverno"
atomic = true
repository = "https://kyverno.github.io/kyverno/"
chart = "kyverno"
version = "3.6.1"
values = [yamlencode({
# When Kyverno is unavailable, allow pod creation to proceed without
# mutation/validation rather than blocking all admissions cluster-wide.
features = {
forceFailurePolicyIgnore = {
enabled = true
}
}
reportsController = {
resources = {
limits = {
memory = "512Mi"
}
requests = {
cpu = "100m"
memory = "128Mi"
}
}
}
admissionController = {
replicas = 2
container = {
resources = {
limits = {
memory = "768Mi"
}
requests = {
cpu = "100m"
memory = "128Mi"
}
}
}
# More tolerant liveness probe — API server slowness shouldn't kill the pod
livenessProbe = {
httpGet = {
path = "/health/liveness"
port = 9443
scheme = "HTTPS"
}
initialDelaySeconds = 15
periodSeconds = 30
timeoutSeconds = 5
failureThreshold = 4
successThreshold = 1
}
# Spread replicas across nodes for HA
topologySpreadConstraints = [
{
maxSkew = 1
topologyKey = "kubernetes.io/hostname"
whenUnsatisfiable = "DoNotSchedule"
labelSelector = {
matchLabels = {
"app.kubernetes.io/component" = "admission-controller"
"app.kubernetes.io/instance" = "kyverno"
}
}
}
]
}
})]
}
# To unlabel all:
# kubectl label deployment,statefulset,daemonset --all-namespaces -l tier tier-
#
# Uses namespaceSelector to match tiers — no API call needed.
# One rule per tier so Kyverno resolves the tier value from its informer cache.
resource "kubernetes_manifest" "mutate_tier_from_namespace" {
manifest = {
apiVersion = "kyverno.io/v1"
kind = "ClusterPolicy"
metadata = {
name = "sync-tier-label-from-namespace"
}
spec = {
rules = [for tier in local.governance_tiers : {
name = "sync-tier-${tier}"
match = {
any = [
{
resources = {
kinds = ["Deployment", "StatefulSet", "DaemonSet"]
namespaceSelector = {
matchLabels = {
tier = tier
}
}
}
}
]
}
exclude = {
any = [
{
resources = {
namespaces = ["kube-system", "metallb-system", "n8n"]
}
}
]
}
mutate = {
patchStrategicMerge = {
metadata = {
labels = {
"+(tier)" = tier
}
}
}
}
}]
}
}
}
# resource "kubernetes_manifest" "enforce_pod_tier_label" {
# manifest = {
# apiVersion = "kyverno.io/v1"
# kind = "ClusterPolicy"
# metadata = {
# name = "enforce-pod-tier-label"
# annotations = {
# "policies.kyverno.io/description" = "Rejects any pod that does not have a tier label."
# }
# }
# spec = {
# # 'Enforce' blocks the creation. 'Audit' just reports it.
# validationFailureAction = "Enforce"
# background = true
# rules = [
# {
# name = "check-for-tier-label"
# match = {
# any = [
# {
# resources = {
# kinds = ["Pod"]
# }
# }
# ]
# }
# validate = {
# message = "The label 'tier' is required for all pods in this cluster."
# pattern = {
# metadata = {
# labels = {
# "tier" = "?*" # The "?*" syntax means the value must not be empty
# }
# }
# }
# }
# }
# ]
# }
# }
# }