Files
harmony/iot/iot-operator-v0/chart/values.yaml
Jean-Gabriel Gill-Couture 99e661ce4d
All checks were successful
Run Check Script / check (pull_request) Successful in 2m44s
feat(iot-operator): helm chart + gen-chart-crd subcommand
Chapter 3 scaffolding. Chart layout mirrors the CloudNativePG
convention after reviewing the CRD-in-chart vs CRD-as-hook
tradeoff: CRDs live inside templates/ (so helm upgrade re-applies
schema changes) with helm.sh/resource-policy: keep so
helm uninstall never deletes them. Chart publication target is
hub.nationtech.io.

CRD yaml is generated at chart-release time by a new
`iot-operator-v0 gen-chart-crd` subcommand reading
Deployment::crd() — the runtime install path remains the typed
Score; only the chart deliverable uses generated yaml. Wrapped
with the helm conditional + annotations by templates/crds.yaml
via .Files.Get so the generated yaml stays pure.

Install / upgrade / uninstall-preserves-CRD validated against a
scratch k3d cluster; the operator pod naturally stays pending
because the hub.nationtech.io image hasn't been published yet.
2026-04-21 23:33:06 -04:00

80 lines
2.0 KiB
YAML

## iot-operator-v0 helm chart values.
## Cnpg-aligned conventions: image.* shape, crds.create toggle,
## non-root podSecurityContext, resources{} stub.
image:
repository: hub.nationtech.io/iot-operator-v0
# Defaults to .Chart.AppVersion when unset.
tag: ""
pullPolicy: IfNotPresent
imagePullSecrets: []
## Controller is singleton today — kube-rs Controller handles
## reconcile-level concurrency within the process. Multi-replica
## HA would need leader election wired up first (see ROADMAP
## for future work).
replicaCount: 1
## NATS connection. The operator connects to this URL on startup
## and stays connected; a flaky NATS link surfaces as pod restart.
nats:
url: nats://nats.iot-system.svc.cluster.local:4222
## Override the KV bucket name the operator writes desired-state
## into. Defaults to harmony-reconciler-contracts' BUCKET_DESIRED_STATE
## constant, so this is only useful for multi-tenant testbeds.
kvBucket: ""
## RUST_LOG directive passed through to the operator pod.
logLevel: info
## CRDs. Cnpg pattern: install by default, gated by this toggle so
## GitOps/cluster-admin teams managing CRDs out-of-band can flip
## it off. Resource-policy: keep annotation on the CRDs themselves
## means `helm uninstall` never deletes them (and therefore never
## deletes customer CRs).
crds:
create: true
serviceAccount:
create: true
# Leave empty to default to the release name.
name: ""
annotations: {}
rbac:
create: true
## Standard k8s resource requests/limits. Left unset by default;
## uncomment and tune before running in production.
resources: {}
# limits:
# cpu: 200m
# memory: 256Mi
# requests:
# cpu: 50m
# memory: 128Mi
podSecurityContext:
runAsNonRoot: true
runAsUser: 10001
runAsGroup: 10001
fsGroup: 10001
seccompProfile:
type: RuntimeDefault
containerSecurityContext:
allowPrivilegeEscalation: false
readOnlyRootFilesystem: true
capabilities:
drop:
- ALL
nodeSelector: {}
tolerations: []
affinity: {}
podAnnotations: {}
podLabels: {}