Syncing Secrets Across Namespaces

It may be required for multiple components across namespaces to consume the same Secret that has been created by a single Certificate. The recommended way to do this is to use extensions such as:

• reflector with support for auto secret reflection
• kubernetes-replicator secret replication

Serving a wildcard to ingress resources in different namespaces (default SSL certificate)

Most ingress controllers, including ingress-nginx, Traefik, and Kong support specifying a single certificate to be used for ingress resources which request TLS but do not specify tls.[].secretName. This is often referred to as a "default SSL certificate". As long as this is correctly configured, ingress resources in any namespace will be able to use a single wildcard certificate. Wildcard certificates are not supported with HTTP01 validation and require DNS01.

Sample ingress snippet:

apiVersion: networking.k8s.io/v1
kind: Ingress
#[...]
spec:
  rules:
  - host: service.example.com
  #[...]
  tls:
  - hosts:
    - service.example.com
    #secretName omitted to use default wildcard certificate

Syncing arbitrary secrets across namespaces using extensions

In order for the target Secret to be synced, you can use the secretTemplate field for annotating the generated secret with the extension specific annotation (See CertificateSecretTemplate).

Using reflector

The example below shows syncing a certificate's secret from the cert-manager namespace to multiple namespaces (i.e. dev, staging, prod). Reflector will ensure that any namespace (existing or new) matching the allowed condition (with regex support) will get a copy of the certificate's secret and will keep it up to date. You can also sync other secrets (different name) using reflector (consult the extension's README).

---
apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
  name: source
  namespace: cert-manager
spec:
  secretName: source-tls
  commonName: source
  issuerRef:
    name: source-ca
    kind: Issuer
    group: cert-manager.io
  secretTemplate:
    annotations:
      reflector.v1.k8s.emberstack.com/reflection-allowed: "true"
      reflector.v1.k8s.emberstack.com/reflection-allowed-namespaces: "dev,staging,prod"  # Control destination namespaces
      reflector.v1.k8s.emberstack.com/reflection-auto-enabled: "true" # Auto create reflection for matching namespaces
      reflector.v1.k8s.emberstack.com/reflection-auto-namespaces: "dev,staging,prod" # Control auto-reflection namespaces

Using kubernetes-replicator

Replicator supports both push- and pull-based replication. Push-based replication will "push out" the TLS secret into namespaces when new ones are created, or when the secret changes. Pull-based replication makes it possible to create an empty TLS secret in the destination namespace and select a "source" resource from which the data is replicated from. The following example shows the pull-based approach:

apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
  name: source
  namespace: cert-manager
spec:
  secretName: source-tls
  commonName: source
  issuerRef:
    name: source-ca
    kind: Issuer
  secretTemplate:
    annotations:
      replicator.v1.mittwald.de/replication-allowed: "true"  # permit replication
      replicator.v1.mittwald.de/replication-allowed-namespaces: "dev,test,prod-[0-9]*"  # comma separated list of namespaces or regular expressions
---
apiVersion: v1
kind: Secret
metadata:
  name: tls-secret-replica
  namespace: prod-1
  annotations:
    replicator.v1.mittwald.de/replicate-from: cert-manager/source-tls
type: kubernetes.io/tls
# Normally, we'd create an empty destination secret, but secrets of type
# 'kubernetes.io/tls' are treated in a special way and need to have properties
# data["tls.crt"] and data["tls.key"] to begin with, though they may be empty.
data:
  tls.key: ""
  tls.crt: ""