Chapter 3. Configuring certificates
3.1. Replacing the default ingress certificate
3.1.1. Understanding the default ingress certificate
By default, OpenShift Container Platform uses the Ingress Operator to create an internal CA and issue a wildcard certificate that is valid for applications under the .apps
sub-domain. Both the web console and CLI use this certificate as well.
The internal infrastructure CA certificates are self-signed. While this process might be perceived as bad practice by some security or PKI teams, any risk here is minimal. The only clients that implicitly trust these certificates are other components within the cluster. Replacing the default wildcard certificate with one that is issued by a public CA already included in the CA bundle as provided by the container userspace allows external clients to connect securely to applications running under the .apps
sub-domain.
3.1.2. Replacing the default ingress certificate
You can replace the default ingress certificate for all applications under the .apps
subdomain. After you replace the certificate, all applications, including the web console and CLI, will have encryption provided by specified certificate.
Prerequisites
-
You must have a wildcard certificate for the fully qualified
.apps
subdomain and its corresponding private key. Each should be in a separate PEM format file. - The private key must be unencrypted. If your key is encrypted, decrypt it before importing it into OpenShift Container Platform.
-
The certificate must include the
subjectAltName
extension showing*.apps.<clustername>.<domain>
. - The certificate file can contain one or more certificates in a chain. The wildcard certificate must be the first certificate in the file. It can then be followed with any intermediate certificates, and the file should end with the root CA certificate.
- Copy the root CA certificate into an additional PEM format file.
Procedure
Create a config map that includes only the root CA certificate used to sign the wildcard certificate:
$ oc create configmap custom-ca \ --from-file=ca-bundle.crt=</path/to/example-ca.crt> \1 -n openshift-config
- 1
</path/to/example-ca.crt>
is the path to the root CA certificate file on your local file system.
Update the cluster-wide proxy configuration with the newly created config map:
$ oc patch proxy/cluster \ --type=merge \ --patch='{"spec":{"trustedCA":{"name":"custom-ca"}}}'
Create a secret that contains the wildcard certificate chain and key:
$ oc create secret tls <secret> \1 --cert=</path/to/cert.crt> \2 --key=</path/to/cert.key> \3 -n openshift-ingress
Update the Ingress Controller configuration with the newly created secret:
$ oc patch ingresscontroller.operator default \ --type=merge -p \ '{"spec":{"defaultCertificate": {"name": "<secret>"}}}' \1 -n openshift-ingress-operator
- 1
- Replace
<secret>
with the name used for the secret in the previous step.
Additional resources
3.2. Adding API server certificates
The default API server certificate is issued by an internal OpenShift Container Platform cluster CA. Clients outside of the cluster will not be able to verify the API server’s certificate by default. This certificate can be replaced by one that is issued by a CA that clients trust.
3.2.1. Add an API server named certificate
The default API server certificate is issued by an internal OpenShift Container Platform cluster CA. You can add one or more alternative certificates that the API server will return based on the fully qualified domain name (FQDN) requested by the client, for example when a reverse proxy or load balancer is used.
Prerequisites
- You must have a certificate for the FQDN and its corresponding private key. Each should be in a separate PEM format file.
- The private key must be unencrypted. If your key is encrypted, decrypt it before importing it into OpenShift Container Platform.
-
The certificate must include the
subjectAltName
extension showing the FQDN. - The certificate file can contain one or more certificates in a chain. The certificate for the API server FQDN must be the first certificate in the file. It can then be followed with any intermediate certificates, and the file should end with the root CA certificate.
Do not provide a named certificate for the internal load balancer (host name api-int.<cluster_name>.<base_domain>
). Doing so will leave your cluster in a degraded state.
Procedure
Login to the new API as the
kubeadmin
user.$ oc login -u kubeadmin -p <password> https://FQDN:6443
Get the
kubeconfig
file.$ oc config view --flatten > kubeconfig-newapi
Create a secret that contains the certificate chain and private key in the
openshift-config
namespace.$ oc create secret tls <secret> \1 --cert=</path/to/cert.crt> \2 --key=</path/to/cert.key> \3 -n openshift-config
Update the API server to reference the created secret.
$ oc patch apiserver cluster \ --type=merge -p \ '{"spec":{"servingCerts": {"namedCertificates": [{"names": ["<FQDN>"], 1 "servingCertificate": {"name": "<secret>"}}]}}}' 2
Examine the
apiserver/cluster
object and confirm the secret is now referenced.$ oc get apiserver cluster -o yaml
Example output
... spec: servingCerts: namedCertificates: - names: - <FQDN> servingCertificate: name: <secret> ...
Check the
kube-apiserver
operator, and verify that a new revision of the Kubernetes API server rolls out. It may take a minute for the operator to detect the configuration change and trigger a new deployment. While the new revision is rolling out,PROGRESSING
will reportTrue
.$ oc get clusteroperators kube-apiserver
Do not continue to the next step until
PROGRESSING
is listed asFalse
, as shown in the following output:Example output
NAME VERSION AVAILABLE PROGRESSING DEGRADED SINCE kube-apiserver 4.10.0 True False False 145m
If
PROGRESSING
is showingTrue
, wait a few minutes and try again.NoteA new revision of the Kubernetes API server only rolls out if the API server named certificate is added for the first time. When the API server named certificate is renewed, a new revision of the Kubernetes API server does not roll out because the
kube-apiserver
pods dynamically reload the updated certificate.
3.3. Securing service traffic using service serving certificate secrets
3.3.1. Understanding service serving certificates
Service serving certificates are intended to support complex middleware applications that require encryption. These certificates are issued as TLS web server certificates.
The service-ca
controller uses the x509.SHA256WithRSA
signature algorithm to generate service certificates.
The generated certificate and key are in PEM format, stored in tls.crt
and tls.key
respectively, within a created secret. The certificate and key are automatically replaced when they get close to expiration.
The service CA certificate, which issues the service certificates, is valid for 26 months and is automatically rotated when there is less than 13 months validity left. After rotation, the previous service CA configuration is still trusted until its expiration. This allows a grace period for all affected services to refresh their key material before the expiration. If you do not upgrade your cluster during this grace period, which restarts services and refreshes their key material, you might need to manually restart services to avoid failures after the previous service CA expires.
You can use the following command to manually restart all pods in the cluster. Be aware that running this command causes a service interruption, because it deletes every running pod in every namespace. These pods will automatically restart after they are deleted.
$ for I in $(oc get ns -o jsonpath='{range .items[*]} {.metadata.name}{"\n"} {end}'); \ do oc delete pods --all -n $I; \ sleep 1; \ done
3.3.2. Add a service certificate
To secure communication to your service, generate a signed serving certificate and key pair into a secret in the same namespace as the service.
The generated certificate is only valid for the internal service DNS name <service.name>.<service.namespace>.svc
, and is only valid for internal communications. If your service is a headless service (no clusterIP
value set), the generated certificate also contains a wildcard subject in the format of *.<service.name>.<service.namespace>.svc
.
Because the generated certificates contain wildcard subjects for headless services, you must not use the service CA if your client must differentiate between individual pods. In this case:
- Generate individual TLS certificates by using a different CA.
- Do not accept the service CA as a trusted CA for connections that are directed to individual pods and must not be impersonated by other pods. These connections must be configured to trust the CA that was used to generate the individual TLS certificates.
Prerequisites:
- You must have a service defined.
Procedure
Annotate the service with
service.beta.openshift.io/serving-cert-secret-name
:$ oc annotate service <service_name> \1 service.beta.openshift.io/serving-cert-secret-name=<secret_name> 2
For example, use the following command to annotate the service
test1
:$ oc annotate service test1 service.beta.openshift.io/serving-cert-secret-name=test1
Examine the service to confirm that the annotations are present:
$ oc describe service <service_name>
Example output
... Annotations: service.beta.openshift.io/serving-cert-secret-name: <service_name> service.beta.openshift.io/serving-cert-signed-by: openshift-service-serving-signer@1556850837 ...
-
After the cluster generates a secret for your service, your
Pod
spec can mount it, and the pod will run after it becomes available.
Additional resources
- You can use a service certificate to configure a secure route using reencrypt TLS termination. For more information, see Creating a re-encrypt route with a custom certificate.
3.3.3. Add the service CA bundle to a config map
A pod can access the service CA certificate by mounting a ConfigMap
object that is annotated with service.beta.openshift.io/inject-cabundle=true
. Once annotated, the cluster automatically injects the service CA certificate into the service-ca.crt
key on the config map. Access to this CA certificate allows TLS clients to verify connections to services using service serving certificates.
After adding this annotation to a config map all existing data in it is deleted. It is recommended to use a separate config map to contain the service-ca.crt
, instead of using the same config map that stores your pod configuration.
Procedure
Annotate the config map with
service.beta.openshift.io/inject-cabundle=true
:$ oc annotate configmap <config_map_name> \1 service.beta.openshift.io/inject-cabundle=true
- 1
- Replace
<config_map_name>
with the name of the config map to annotate.
NoteExplicitly referencing the
service-ca.crt
key in a volume mount will prevent a pod from starting until the config map has been injected with the CA bundle. This behavior can be overridden by setting theoptional
field totrue
for the volume’s serving certificate configuration.For example, use the following command to annotate the config map
test1
:$ oc annotate configmap test1 service.beta.openshift.io/inject-cabundle=true
View the config map to ensure that the service CA bundle has been injected:
$ oc get configmap <config_map_name> -o yaml
The CA bundle is displayed as the value of the
service-ca.crt
key in the YAML output:apiVersion: v1 data: service-ca.crt: | -----BEGIN CERTIFICATE----- ...
3.3.4. Add the service CA bundle to an API service
You can annotate an APIService
object with service.beta.openshift.io/inject-cabundle=true
to have its spec.caBundle
field populated with the service CA bundle. This allows the Kubernetes API server to validate the service CA certificate used to secure the targeted endpoint.
Procedure
Annotate the API service with
service.beta.openshift.io/inject-cabundle=true
:$ oc annotate apiservice <api_service_name> \1 service.beta.openshift.io/inject-cabundle=true
- 1
- Replace
<api_service_name>
with the name of the API service to annotate.
For example, use the following command to annotate the API service
test1
:$ oc annotate apiservice test1 service.beta.openshift.io/inject-cabundle=true
View the API service to ensure that the service CA bundle has been injected:
$ oc get apiservice <api_service_name> -o yaml
The CA bundle is displayed in the
spec.caBundle
field in the YAML output:apiVersion: apiregistration.k8s.io/v1 kind: APIService metadata: annotations: service.beta.openshift.io/inject-cabundle: "true" ... spec: caBundle: <CA_BUNDLE> ...
3.3.5. Add the service CA bundle to a custom resource definition
You can annotate a CustomResourceDefinition
(CRD) object with service.beta.openshift.io/inject-cabundle=true
to have its spec.conversion.webhook.clientConfig.caBundle
field populated with the service CA bundle. This allows the Kubernetes API server to validate the service CA certificate used to secure the targeted endpoint.
The service CA bundle will only be injected into the CRD if the CRD is configured to use a webhook for conversion. It is only useful to inject the service CA bundle if a CRD’s webhook is secured with a service CA certificate.
Procedure
Annotate the CRD with
service.beta.openshift.io/inject-cabundle=true
:$ oc annotate crd <crd_name> \1 service.beta.openshift.io/inject-cabundle=true
- 1
- Replace
<crd_name>
with the name of the CRD to annotate.
For example, use the following command to annotate the CRD
test1
:$ oc annotate crd test1 service.beta.openshift.io/inject-cabundle=true
View the CRD to ensure that the service CA bundle has been injected:
$ oc get crd <crd_name> -o yaml
The CA bundle is displayed in the
spec.conversion.webhook.clientConfig.caBundle
field in the YAML output:apiVersion: apiextensions.k8s.io/v1 kind: CustomResourceDefinition metadata: annotations: service.beta.openshift.io/inject-cabundle: "true" ... spec: conversion: strategy: Webhook webhook: clientConfig: caBundle: <CA_BUNDLE> ...
3.3.6. Add the service CA bundle to a mutating webhook configuration
You can annotate a MutatingWebhookConfiguration
object with service.beta.openshift.io/inject-cabundle=true
to have the clientConfig.caBundle
field of each webhook populated with the service CA bundle. This allows the Kubernetes API server to validate the service CA certificate used to secure the targeted endpoint.
Do not set this annotation for admission webhook configurations that need to specify different CA bundles for different webhooks. If you do, then the service CA bundle will be injected for all webhooks.
Procedure
Annotate the mutating webhook configuration with
service.beta.openshift.io/inject-cabundle=true
:$ oc annotate mutatingwebhookconfigurations <mutating_webhook_name> \1 service.beta.openshift.io/inject-cabundle=true
- 1
- Replace
<mutating_webhook_name>
with the name of the mutating webhook configuration to annotate.
For example, use the following command to annotate the mutating webhook configuration
test1
:$ oc annotate mutatingwebhookconfigurations test1 service.beta.openshift.io/inject-cabundle=true
View the mutating webhook configuration to ensure that the service CA bundle has been injected:
$ oc get mutatingwebhookconfigurations <mutating_webhook_name> -o yaml
The CA bundle is displayed in the
clientConfig.caBundle
field of all webhooks in the YAML output:apiVersion: admissionregistration.k8s.io/v1 kind: MutatingWebhookConfiguration metadata: annotations: service.beta.openshift.io/inject-cabundle: "true" ... webhooks: - myWebhook: - v1beta1 clientConfig: caBundle: <CA_BUNDLE> ...
3.3.7. Add the service CA bundle to a validating webhook configuration
You can annotate a ValidatingWebhookConfiguration
object with service.beta.openshift.io/inject-cabundle=true
to have the clientConfig.caBundle
field of each webhook populated with the service CA bundle. This allows the Kubernetes API server to validate the service CA certificate used to secure the targeted endpoint.
Do not set this annotation for admission webhook configurations that need to specify different CA bundles for different webhooks. If you do, then the service CA bundle will be injected for all webhooks.
Procedure
Annotate the validating webhook configuration with
service.beta.openshift.io/inject-cabundle=true
:$ oc annotate validatingwebhookconfigurations <validating_webhook_name> \1 service.beta.openshift.io/inject-cabundle=true
- 1
- Replace
<validating_webhook_name>
with the name of the validating webhook configuration to annotate.
For example, use the following command to annotate the validating webhook configuration
test1
:$ oc annotate validatingwebhookconfigurations test1 service.beta.openshift.io/inject-cabundle=true
View the validating webhook configuration to ensure that the service CA bundle has been injected:
$ oc get validatingwebhookconfigurations <validating_webhook_name> -o yaml
The CA bundle is displayed in the
clientConfig.caBundle
field of all webhooks in the YAML output:apiVersion: admissionregistration.k8s.io/v1 kind: ValidatingWebhookConfiguration metadata: annotations: service.beta.openshift.io/inject-cabundle: "true" ... webhooks: - myWebhook: - v1beta1 clientConfig: caBundle: <CA_BUNDLE> ...
3.3.8. Manually rotate the generated service certificate
You can rotate the service certificate by deleting the associated secret. Deleting the secret results in a new one being automatically created, resulting in a new certificate.
Prerequisites
- A secret containing the certificate and key pair must have been generated for the service.
Procedure
Examine the service to determine the secret containing the certificate. This is found in the
serving-cert-secret-name
annotation, as seen below.$ oc describe service <service_name>
Example output
... service.beta.openshift.io/serving-cert-secret-name: <secret> ...
Delete the generated secret for the service. This process will automatically recreate the secret.
$ oc delete secret <secret> 1
- 1
- Replace
<secret>
with the name of the secret from the previous step.
Confirm that the certificate has been recreated by obtaining the new secret and examining the
AGE
.$ oc get secret <service_name>
Example output
NAME TYPE DATA AGE <service.name> kubernetes.io/tls 2 1s
3.3.9. Manually rotate the service CA certificate
The service CA is valid for 26 months and is automatically refreshed when there is less than 13 months validity left.
If necessary, you can manually refresh the service CA by using the following procedure.
A manually-rotated service CA does not maintain trust with the previous service CA. You might experience a temporary service disruption until the pods in the cluster are restarted, which ensures that pods are using service serving certificates issued by the new service CA.
Prerequisites
- You must be logged in as a cluster admin.
Procedure
View the expiration date of the current service CA certificate by using the following command.
$ oc get secrets/signing-key -n openshift-service-ca \ -o template='{{index .data "tls.crt"}}' \ | base64 --decode \ | openssl x509 -noout -enddate
Manually rotate the service CA. This process generates a new service CA which will be used to sign the new service certificates.
$ oc delete secret/signing-key -n openshift-service-ca
To apply the new certificates to all services, restart all the pods in your cluster. This command ensures that all services use the updated certificates.
$ for I in $(oc get ns -o jsonpath='{range .items[*]} {.metadata.name}{"\n"} {end}'); \ do oc delete pods --all -n $I; \ sleep 1; \ done
WarningThis command will cause a service interruption, as it goes through and deletes every running pod in every namespace. These pods will automatically restart after they are deleted.
3.4. Updating the CA bundle
3.4.1. Understanding the CA Bundle certificate
Proxy certificates allow users to specify one or more custom certificate authority (CA) used by platform components when making egress connections.
The trustedCA
field of the Proxy object is a reference to a config map that contains a user-provided trusted certificate authority (CA) bundle. This bundle is merged with the Red Hat Enterprise Linux CoreOS (RHCOS) trust bundle and injected into the trust store of platform components that make egress HTTPS calls. For example, image-registry-operator
calls an external image registry to download images. If trustedCA
is not specified, only the RHCOS trust bundle is used for proxied HTTPS connections. Provide custom CA certificates to the RHCOS trust bundle if you want to use your own certificate infrastructure.
The trustedCA
field should only be consumed by a proxy validator. The validator is responsible for reading the certificate bundle from required key ca-bundle.crt
and copying it to a config map named trusted-ca-bundle
in the openshift-config-managed
namespace. The namespace for the config map referenced by trustedCA
is openshift-config
:
apiVersion: v1 kind: ConfigMap metadata: name: user-ca-bundle namespace: openshift-config data: ca-bundle.crt: | -----BEGIN CERTIFICATE----- Custom CA certificate bundle. -----END CERTIFICATE-----
3.4.2. Replacing the CA Bundle certificate
Procedure
Create a config map that includes the root CA certificate used to sign the wildcard certificate:
$ oc create configmap custom-ca \ --from-file=ca-bundle.crt=</path/to/example-ca.crt> \1 -n openshift-config
- 1
</path/to/example-ca.crt>
is the path to the CA certificate bundle on your local file system.
Update the cluster-wide proxy configuration with the newly created config map:
$ oc patch proxy/cluster \ --type=merge \ --patch='{"spec":{"trustedCA":{"name":"custom-ca"}}}'