Exchanging messages over an ssl ingress with cert-manager's generated certs

Prerequisite

Before you start, you need to have access to a running Kubernetes cluster environment. A Minikube instance running on your laptop will do fine.

Start minikube with a parametrized dns domain name

$ minikube start --dns-domain='demo.artemiscloud.io'
😄  minikube v1.32.0 on Fedora 39
🎉  minikube 1.33.1 is available! Download it: https://github.com/kubernetes/minikube/releases/tag/v1.33.1
💡  To disable this notice, run: 'minikube config set WantUpdateNotification false'

✨  Automatically selected the kvm2 driver. Other choices: qemu2, ssh
👍  Starting control plane node minikube in cluster minikube
🔥  Creating kvm2 VM (CPUs=2, Memory=6000MB, Disk=20000MB) ...
🐳  Preparing Kubernetes v1.28.3 on Docker 24.0.7 ...
    ▪ Generating certificates and keys ...
    ▪ Booting up control plane ...
    ▪ Configuring RBAC rules ...
🔗  Configuring bridge CNI (Container Networking Interface) ...
    ▪ Using image gcr.io/k8s-minikube/storage-provisioner:v5
🔎  Verifying Kubernetes components...
🌟  Enabled addons: storage-provisioner, default-storageclass
🏄  Done! kubectl is now configured to use "minikube" cluster and "default" namespace by default

Enable nginx and ssl passthrough for minikube

$ minikube addons enable ingress
$ minikube kubectl -- patch deployment -n ingress-nginx ingress-nginx-controller --type='json' -p='[{"op": "add", "path": "/spec/template/spec/containers/0/args/-", "value":"--enable-ssl-passthrough"}]'

Make sure the domain of your cluster is resolvable

If you are running your OpenShift cluster locally, you might not be able to resolve the urls to IPs out of the blue. Follow this guide to configure your setup.

Deploy the operator

create the namespace

$ kubectl create namespace send-receive-project
$ kubectl config set-context --current --namespace=send-receive-project

Go to the root of the operator repo and install it:

$ ./deploy/install_opr.sh

Wait for the Operator to start (status: running).

$ kubectl get pod --namespace send-receive-project --watch

Create the chain of trust with Cert manager

Install cert manager

Follow the official documentation.

$ kubectl get pod --namespace cert-manager --watch
NAME                                       READY   STATUS              RESTARTS   AGE
cert-manager-7ddd8cdb9f-fx9mw              1/1     Running             0          13s
cert-manager-cainjector-57cd76c845-zg45d   1/1     Running             0          13s
cert-manager-webhook-cf8f9f895-kkj5r       0/1     ContainerCreating   0          13s
cert-manager-webhook-cf8f9f895-kkj5r       0/1     Running             0          17s
cert-manager-webhook-cf8f9f895-kkj5r       1/1     Running             0          21s

Create the root issuer

$ kubectl apply -f - <<EOF
apiVersion: cert-manager.io/v1
kind: Issuer
metadata:
  name: send-receive-root-issuer
  namespace: send-receive-project
spec:
  selfSigned: {}
EOF
$ kubectl get issuers  -n send-receive-project -o wide send-receive-root-issuer --watch
NAME                       READY   STATUS   AGE
send-receive-root-issuer   True             45s

Create the issuer certificate

$ kubectl apply -f - <<EOF
apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
  name: send-receive-issuer-cert
  namespace: send-receive-project
spec:
  isCA: true
  commonName: ArtemiseCloud send-receive issuer
  dnsNames:
    - issuer.demo.artemiscloud.io
  secretName: send-receive-issuer-cert-secret
  privateKey:
    algorithm: ECDSA
    size: 256
  issuerRef:
    name: send-receive-root-issuer
    kind: Issuer
EOF
$ kubectl get certificates  -n send-receive-project --watch
NAME                       READY   SECRET                            AGE
send-receive-issuer-cert   True    send-receive-issuer-cert-secret   24s

Create the issuer

$ kubectl apply -f - <<EOF
apiVersion: cert-manager.io/v1
kind: Issuer
metadata:
  name: send-receive-issuer
  namespace: send-receive-project
spec:
  ca:
    secretName: send-receive-issuer-cert-secret
EOF
$ kubectl get issuer  -n send-receive-project --watch
NAME                       READY   AGE
send-receive-issuer        True    10s
send-receive-root-issuer   True    50s

Download the issuer’s CA

Note: to execute the following commands you’ll need jq.

We need to download a certificate so that our artermis client is be able to open a secure connection to the broker.

In this tutorial we configure the deployment CR so that upon starting a new broker, cert-manager is using the send-receive-issuer to generate the broker’s certificate. The issuer is signing the generated certs with its CA: send-receive-issuer-cert. In turns this means that a client using the send-receive-issuer-cert in its trust store will trust any of the issuer’s generated certs.

The issuer CA can be found in the send-receive-issuer-cert-secret secret:

$ kubectl get secrets send-receive-issuer-cert-secret -o json | jq -r '.data."tls.crt"' | base64 -d
-----BEGIN CERTIFICATE-----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-----END CERTIFICATE-----

Store the output in the /tmp/IssuerCA.pem file:

$ kubectl get secrets send-receive-issuer-cert-secret -o json | jq -r '.data."tls.crt"' | base64 -d > /tmp/IssuerCA.pem

Deploy the ActiveMQ Artemis Broker

Ingress and ssl configuration

The acceptor is configured as follows:

  • port: 62626
  • sslEnabled: true activates ssl
  • expose: true and exposeMode: ingress are triggering an ingress creation
  • sslSecret: send-receive-sslacceptor-0-svc-ing-ptls provides the name of the sercret store that’ll be used to secure the connection. The -ptls suffix indicates that it will be provide by cert-manager so the artemis operator will not validate its present when the Artemis CR is deployed.
  • ingressHost: ing.$(ITEM_NAME).$(CR_NAME)-$(BROKER_ORDINAL).$(CR_NAMESPACE).$(INGRESS_DOMAIN) customizes the url to access the acceptor from outside the cluster.

An annotation is used to ask cert-manager to produce the sslSecret needed to secure the connection. It is configured as follows:

  • the selector attaches the annotation to the acceptor’s ingress by its name
  • cert-manager.io/issuer: send-receive-issuer links to the previously created issuer, cert-manager will use it to generate the secrets.
  • spec.hosts list the domains for which the certificate applies, it has to match the url of the ingress set in the acceptor.
  • spec.secretName matches the sslSecret value of the acceptor and must be set to the name of the ingress + -ptls. Here: send-receive-sslacceptor-0-svc-ing-ptls.

broker configuration

Two queues are added to exchange messages on. These are configured via the broker properties. Two queues are setup, one called APP.JOBS that is of type ANYCAST and one called APP.COMMANDS that is of type MULTICAST.

Apply the configuration and start the broker

$ kubectl apply -f - <<'EOF'
apiVersion: broker.amq.io/v1beta1
kind: ActiveMQArtemis
metadata:
  name: send-receive
  namespace: send-receive-project
spec:
  ingressDomain: demo.artemiscloud.io
  acceptors:
    - name: sslacceptor
      port: 62626
      sslEnabled: true
      expose: true
      exposeMode: ingress
      sslSecret: send-receive-sslacceptor-0-svc-ing-ptls
      ingressHost: ing.$(ITEM_NAME).$(CR_NAME)-$(BROKER_ORDINAL).$(CR_NAMESPACE).$(INGRESS_DOMAIN)
  resourceTemplates:
    - selector:
        kind: Ingress
        name: send-receive-sslacceptor-0-svc-ing
      annotations:
        cert-manager.io/issuer: send-receive-issuer
      patch:
        kind: Ingress
        spec:
          tls:
            - hosts:
                - ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io
              secretName: send-receive-sslacceptor-0-svc-ing-ptls
  brokerProperties:
    - addressConfigurations."APP.JOBS".routingTypes=ANYCAST
    - addressConfigurations."APP.JOBS".queueConfigs."APP.JOBS".routingType=ANYCAST
    - addressConfigurations."APP.COMMANDS".routingTypes=MULTICAST
EOF

Wait for the Broker to be ready:

$ kubectl get ActivemqArtemis  --watch
NAME           READY   AGE
send-receive   True    51s

Check that the ingress is available and has an IP address:

$ kubectl get ingress --show-labels
NAME                                 CLASS   HOSTS                                                                     ADDRESS         PORTS     AGE   LABELS
send-receive-sslacceptor-0-svc-ing   nginx   ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io  192.168.39.68   80, 443   18s   ActiveMQArtemis=send-receive,application=send-receive-app,statefulset.kubernetes.io/pod-name=send-receive-ss-0

Exchange messages between a producer and a consumer

Download Artemis

Download the latest release of ActiveMQ Artemis, decompress the tarball and locate the artemis executable.

The client needs a bit more information than just the raw URL of the ingress to open a working connection:

  • sslEnabled is set to true
  • trustStorePath is set to the issuer CA downloaded at the previous Steps
  • trustStoreType is set to PEM
  • useTopologyForLoadBalancing is set to false Leaving the value to true would lead to some errors where the client will try to open a direct line of communication with the broker without going through the ingress. The ingress will provide a round robin load balancing when multiple brokers can be reached.
Test the connection
$ ./artemis check queue --name TEST --produce 10 --browse 10 --consume 10 --url 'tcp://ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io:443?sslEnabled=true&trustStorePath=/tmp/IssuerCA.pem&trustStoreType=PEM&useTopologyForLoadBalancing=false' --verbose
Executing org.apache.activemq.artemis.cli.commands.check.QueueCheck check queue --name TEST --produce 10 --browse 10 --consume 10 --url tcp://send-receive-ss-0.send-receive-hdls-svc.send-receive-project.svc.demo.artemiscloud.io:443?sslEnabled=true&trustStorePath=/tmp/IssuerCA.pem&trustStoreType=PEM --verbose 
Home::/home/tlavocat/dev/activemq-artemis/artemis-distribution/target/apache-artemis-2.34.0-SNAPSHOT-bin/apache-artemis-2.34.0-SNAPSHOT, Instance::null
Connection brokerURL = tcp://ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io:443?sslEnabled=true&trustStorePath=/tmp/IssuerCA.pem&trustStoreType=PEM
Running QueueCheck
Checking that a producer can send 10 messages to the queue TEST ... success
Checking that a consumer can browse 10 messages from the queue TEST ... success
Checking that a consumer can consume 10 messages from the queue TEST ... success
Checks run: 3, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 6.402 sec - QueueCheck

ANYCAST

For this use case, run first the producer, then the consumer.

$ ./artemis producer --destination queue://APP.JOBS --url "tcp://ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io:443?sslEnabled=true&trustStorePath=/tmp/IssuerCA.pem&trustStoreType=PEM&useTopologyForLoadBalancing=false"
Producer ActiveMQQueue[APP.JOBS], thread=0 Started to calculate elapsed time ...

Producer ActiveMQQueue[APP.JOBS], thread=0 Produced: 1000 messages
Producer ActiveMQQueue[APP.JOBS], thread=0 Elapsed time in second : 9 s
Producer ActiveMQQueue[APP.JOBS], thread=0 Elapsed time in milli second : 9104 milli seconds
$ ./artemis consumer --destination queue://APP.JOBS --url "tcp://ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io:443?sslEnabled=true&trustStorePath=/tmp/IssuerCA.pem&trustStoreType=PEM&useTopologyForLoadBalancing=false"
Connection brokerURL = tcp://ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io:443?sslEnabled=true&trustStorePath=/home/tlavocat/IssuerCA.pem&trustStoreType=PEM
Consumer:: filter = null
Consumer ActiveMQQueue[APP.JOBS], thread=0 wait until 1000 messages are consumed
Received 1000
Consumer ActiveMQQueue[APP.JOBS], thread=0 Consumed: 1000 messages
Consumer ActiveMQQueue[APP.JOBS], thread=0 Elapsed time in second : 0 s
Consumer ActiveMQQueue[APP.JOBS], thread=0 Elapsed time in milli second : 116 milli seconds
Consumer ActiveMQQueue[APP.JOBS], thread=0 Consumed: 1000 messages
Consumer ActiveMQQueue[APP.JOBS], thread=0 Consumer thread finished

MULTICAST

For this use case, run first the consumer(s), then the producer. More details there.

  1. in n other terminal(s) connect n consumer(s):
$ ./artemis consumer --destination topic://APP.COMMANDS --url "tcp://ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io:443?sslEnabled=true&trustStorePath=/tmp/IssuerCA.pem&trustStoreType=PEM&useTopologyForLoadBalancing=false"
Consumer:: filter = null
Consumer ActiveMQTopic[APP.COMMANDS], thread=0 wait until 1000 messages are consumed
Received 1000
Consumer ActiveMQTopic[APP.COMMANDS], thread=0 Consumed: 1000 messages
Consumer ActiveMQTopic[APP.COMMANDS], thread=0 Elapsed time in second : 8 s
Consumer ActiveMQTopic[APP.COMMANDS], thread=0 Elapsed time in milli second : 8434 milli seconds
Consumer ActiveMQTopic[APP.COMMANDS], thread=0 Consumed: 1000 messages
Consumer ActiveMQTopic[APP.COMMANDS], thread=0 Consumer thread finished
  1. connect the producer to start broadcasting messages.
$ ./artemis producer --destination topic://APP.COMMANDS --url "tcp://ing.sslacceptor.send-receive-0.send-receive-project.demo.artemiscloud.io:443?sslEnabled=true&trustStorePath=/tmp/IssuerCA.pem&trustStoreType=PEM&useTopologyForLoadBalancing=false"
Producer ActiveMQTopic[APP.COMMANDS], thread=0 Started to calculate elapsed time ...

Producer ActiveMQTopic[APP.COMMANDS], thread=0 Produced: 1000 messages
Producer ActiveMQTopic[APP.COMMANDS], thread=0 Elapsed time in second : 0 s
Producer ActiveMQTopic[APP.COMMANDS], thread=0 Elapsed time in milli second : 653 milli seconds