Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Hazelcast for OpenShift

Hazelcast can be used as a Caching Layer for applications deployed on OpenShift.

This sample is a complete guideline on how to set up the local OpenShift environment, start Hazelcast cluster, configure Management Center, and finally run a sample client application.

Table of Contents

Step-by-step instruction

Step 1: Install OpenShift environment

There are multiple options to run the OpenShift environment:

Note: This guide can be followed on any environment, however some of the steps differ, for example, accessing Persistent Storage. What's more, some of the steps are not possible to execute on certain environments, for example, connecting with external client to the Hazelcast deployed on OpenShift Online (since it does not provide public IPs). In such cases, the Minishift solution is presented, because Minishift is free and simple to start from.

Minishift toolkit (version 3.3.0) is used to help with running OpenShift locally. Use the following steps to set it up:

  1. Install OpenShift Container Development Kit (CDK) as described here

  2. Configure CDK and run a first Hello World OpenShift application as described here

    The Red Hat guide is complete in general, however here are a few hints that can save your time:

    • Complete the "Setup" section before you run minishift setup-cdk
    • In case of any issues with Minishift, you can enable fine level logging with the option --show-libmachine-logs -v5
    • In case of using Windows and Hyper-V:
      • In one of the points you need to create the virtual switch, you can use this guide to do it; then the name of the created switch is the one you need to use instead of "External (Wireless)"
      • Make sure to add user to Hyper-V Administrator group
      • Hyper-V does not well support NAT networks, so your router must accept the Virtual Machine connecting directly to the same network interface (in case of problems, you may see a meaningless error Too many retries waiting for SSH to be available); the solution is to use VirutalBox instead of Hyper-V or play with the experimental Minishift features
  3. Make sure your minishift and oc tools are installed and ready to use

$ minishift version
minishift v1.11.0+d7f374a
CDK v3.3.0-1

$ oc version
oc v3.9.0-alpha.3+78ddc10
kubernetes v1.9.1+a0ce1bc657
features: Basic-Auth
  1. Make also sure that you're able to access the OpenShift Web Console.
$ minishift console

Log in as Username: "developer", Password: "developer".


Step 2: Start Hazelcast cluster

Note that, in case of Hazelcast Enterprise OpenShift (both Centos and RHEL, you will need a valid Hazelcast license key. If you don't have one, you can either use Hazelcast OpenShift Origin instead or get a trial key from this link.

1) Create Project

Log into OpenShift Platform. The login command is available when you click in the Web Console: "Copy Login Command".

Copy Login Command

$ oc login --token=OnqI91DRdpni6ChmpMDwsRkGd5cjkmXKp7xKsI7u8QE

Then, you can create a new project.

$ oc new-project hazelcast

Note that, in case of Hazelcast Enterprise OpenShift RHEL, you will need to authenticate to Red Hat Container Catalog as described here.

2) Start Hazelcast cluster

Change the directory to Hazelcast Enterprise ($ cd hazelcast-cluster/hazelcast-enterprise-openshift-centos), Hazelcast Enterprise RHEL ($ cd hazelcast-cluster/hazelcast-enterprise-openshift-rhel) or Hazelcast ($ cd hazelcast-cluster/hazelcast-openshift-origin).

Then, start the cluster with the following command.

$ oc new-app -f hazelcast.yaml \
  -p NAMESPACE=$(oc project -q) \
  -p ENTERPRISE_LICENSE_KEY=<hazelcast_enterprise_license>

Used parameters:

  • NAMESPACE: must be the same as the OpenShift project's name
  • ENTERPRISE_LICENSE_KEY: Hazelcast Enterprise License (not needed for Hazelcast OpenShift Origin)

3) Check that Hazelcast is running

To check all created OpenShift resources, use the following command.

$ oc get all
deploy/management-center   1         1         1            1           7m

NAME                              DESIRED   CURRENT   READY     AGE
rs/management-center-3761922137   1         1         1         7m

deploy/management-center   1         1         1            1           7m

NAME                     DESIRED   CURRENT   AGE
statefulsets/hazelcast   3         3         7m

NAME                                    READY     STATUS    RESTARTS   AGE
po/hazelcast-0                          1/1       Running   0          7m
po/hazelcast-1                          1/1       Running   0          6m
po/hazelcast-2                          1/1       Running   0          5m
po/management-center-3761922137-kc6nx   1/1       Running   0          7m

NAME                            TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)    AGE
svc/hazelcast-service           ClusterIP   None         <none>        5701/TCP   7m
svc/management-center-service   ClusterIP   None         <none>        8080/TCP   7m

Please check that the STATUS is Running for all PODs. Then, to check the logs for each replica, use the following command:

$ oc logs po/hazelcast-2

Kubernetes Namespace: hazelcast
Kubernetes Service DNS: hazelcast-service.hazelcast.svc
# CLASSPATH=/data/hazelcast/*:/opt/hazelcast/*:/opt/hazelcast/external/*:
Members [3] {
        Member []:5701 - b047e291-ebd6-4edc-8b9c-d06fcb3b9965
        Member []:5701 - f5e6cf50-d83a-42c5-b152-2571a929fd12
        Member []:5701 - a50f8468-0852-45d2-966a-74301e04d45e this

Note that you can also perform all the operations and observe the results in the OpenShift Web Console.

Hazelcast PODs

4) Delete Hazelcast cluster

If you want to delete all resources (Replication Controller, Service, PODs, Storage, Config Map), you could use the following commands:

$ oc delete all --all
$ oc delete pvc --all
$ oc delete configmap --all

Step 3: Access Management Center

Management Center application (Hazelcast Enterprise only) is already started together with Hazelcast members when using hazelcast.yaml. Nevertheless, in order to make it usable, you need to expose its service.

$ oc expose svc/management-center-service

Then, it's accessible via the exposed route, which you can check by:

$ oc get route
NAME                        HOST/PORT                                                  PATH      SERVICES                    PORT      TERMINATION   WILDCARD
management-center-service   management-center-service-hazelcast.             management-center-service   8080                    None

Then, you can access Management Center by opening management-center-service-hazelcast. (/mancenter for versions prior to 3.10) in your browser.

Management Center

Step 4: Run a sample Hazelcast client application

If you're interested not only in setting up the Hazelcast cluster, but also in using it in the client application, you can follow the following guidelines.

Note that OpenShift sample uses the fabric8 maven plugin to build Docker image. Fabric8 requires 3.3.x or higher maven version, therefore make sure that you have proper maven version installed on your machine.

Note also that the commands below are presented for the Minishift, because they use access to the Docker Registry provided with OpenShift. In case of OpenShift Online (in which you don't have access to the OpenShift's Docker Registry), you may use an external registry (for example, Docker Hub).

1) Build Maven dependencies

Make sure you're in the main openshift directory and install the snapshot JAR files from the root directory:

$ mvn -f ../../pom.xml clean install

2) Build "ocp-demo-frontend" Docker image

Note that in order to build the Docker image, you need to have your OpenShift Docker Engine configured. In case of Minishift, you can do it using the guidelines from the minishift docker-env command, so in case of Unix-based systems:

$ eval $(minishift docker-env)

Run the following command to build the Docker image:

$ mvn -f client-apps/ocp-demo-frontend/pom.xml fabric8:build

3) Push Docker image to the local OpenShift registry

Check if your image is already in the OpenShift registry.

$ oc get is
NAME                DOCKER REPO                                   TAGS      UPDATED
ocp-demo-frontend   latest    39 seconds ago

In case you see the message No resources found, you need to manually push the image with the following commands:

$ docker login -u developer -p $(oc whoami -t) $(minishift openshift registry)
$ docker tag client-apps/ocp-demo-frontend $(minishift openshift registry)/$(oc project -q)/ocp-demo-frontend
$ docker push $(minishift openshift registry)/$(oc project -q)/ocp-demo-frontend

Then, you should see oc-demo-frontend in the output for $ oc get is.

4) Start "ocp-demo-frontend" application

To start the application, use the following command:

$ oc new-app --image-stream=ocp-demo-frontend --name=hazelcast-client-app

You can check that the application is running correctly:

$ oc get all -l app=hazelcast-client-app
NAME                                     REVISION   DESIRED   CURRENT   TRIGGERED BY
deploymentconfigs/hazelcast-client-app   1          1         1         config,image(ocp-demo-frontend:latest)

NAME                              READY     STATUS    RESTARTS   AGE
po/hazelcast-client-app-1-rcrfx   1/1       Running   0          2m

NAME                        DESIRED   CURRENT   READY     AGE
rc/hazelcast-client-app-1   1         1         1         2m

NAME                       TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)                      AGE
svc/hazelcast-client-app   ClusterIP   <none>        8080/TCP,8778/TCP,9779/TCP   2m

5) Expose the application

In order to make the application accessible from outside the OpenShift environment, you need to expose it using the following command:

$ oc expose svc/hazelcast-client-app

Then, you should be able to access the application via the exposed route. You can check the route using the following command:

$ oc get routes
NAME                   HOST/PORT                                             PATH      SERVICES               PORT       TERMINATION   WILDCARD
hazelcast-client-app   hazelcast-client-app-hazelcast.             hazelcast-client-app   8080-tcp                 None

Now, if you open in the browser hazelcast-client-app-hazelcast., you should see the following home screen.


You can check that the application really works together with the Hazelcast cluster by doing some operation in the application, for example, entering "12" in the "Data Operations->Count" and clicking "Auto Pilot". Then, in the Management Center application, you should see that the entries are added.

Management Center

6) Enable Entry Processor

If you want to play with "Entry Processor" from the code sample, you need to place the Entry Processor JAR into the /data directory of Hazelcast. Currently, you use ConfigMap with hazelcast.xml as the volume mounted to /data. However, if we want to copy the custom JAR, you need to change it and use OpenShift Persistent Volume.

Let's first change the directory to Hazelcast Enterprise ($ cd hazelcast-cluster/hazelcast-enterprise-openshift-centos), Hazelcast Enterprise RHEL ($ cd hazelcast-cluster/hazelcast-enterprise-openshift-rhel) or Hazelcast ($ cd hazelcast-cluster/hazelcast-openshift-origin).

Then, replace ConfigMap with the Persistent Storage.

oc volume rc/hz-rc --add --overwrite --claim-size 1Gi --mount-path /data/hazelcast

Now, copy hazelcast.xml and ocp-entry-processor-0.1-SNAPSHOT.jar into the volume. In case of Minishift, you need to first check the volume name.

$ oc get pvc
pvc-6qrqc   Bound     pv0090    100Gi      RWO,ROX,RWX                   1h

The volume name is pv0090 and the following commands copy the necessary files.

$ scp -i $HOME/.minishift/machines/minishift/id_rsa hazelcast-configuration/hazelcast.xml docker@$(minishift ip):/mnt/sda1/var/lib/minishift/openshift.local.pv/pv0090/
$ scp -i $HOME/.minishift/machines/minishift/id_rsa ../../client-apps/ocp-entry-processor/target/ocp-entry-processor-0.1-SNAPSHOT.jar docker@$(minishift ip):/mnt/sda1/var/lib/minishift/openshift.local.pv/pv0090/

Short explanation of the commands above:

  • $HOME/.minishift/machines/minishift/id_rsa - ssh key to Minishift VM is stored in the Minishift's home directory
  • hazelcast.xml - custom configuration of Hazelcast
  • minishift ip - command to return the IP address of the Minishift VM
  • /mnt/sda1/var/lib/minishift/openshift.local.pv/pv0090/ - location of the Persistent Volume pv0090 in Minishift VM

Then, after restarting all PODs (delete them, they will automatically start again), you can play with "Entry Processor".

oc delete po --all

Entry Processor

The results should be visible in Management Center.

Management Center

Authenticate to Red Hat Container Catalog

NOTE: Executing the following commands requires oc in version 3.9 or higher!

The hazelcast-cluster/hazelcast-enterprise-openshift-rhel/hazelcast.yaml uses images from Red Hat Container Catalog, which requires setting Red Hat credentials. In order to do it, you need to execute the following command after creating the OpenShift project.

$ oc create secret docker-registry rhcc \
   --docker-server=registry.connect.redhat.com \
   --docker-username=<red_hat_username> \
   --docker-password=<red_hat_password> \
$ oc secrets link default rhcc --for=pull

External Hazelcast Client

Client application presented in this tutorial works only if deployed inside the OpenShift environment, because Kubernetes Discovery SPI does not support external clients. If you need to connect to the Hazelcast cluster deployed on OpenShift, then you can expose a service with externalIP and connect to the cluster with Smart Routing disabled.

Note: Your OpenShift environment needs to provide public IP addresses, so for example, the solution won't work on the OpenShift Online environment.

You can create a service with automatically assigned external IP and port.

$ oc expose rc hz-rc --type=LoadBalancer --name=hazelcast-ingress

The following command checks the external port under which the service is published.

$ oc get service hazelcast-ingress -o custom-columns=EXTERNAL_PORT:.spec.ports[0].nodePort

Then, it's possible to access the Hazelcast cluster using server's public IP and that port. In case of Minishift, you can check its IP by:

$ minishift ip

Finally, you can connect to the cluster with the following Java client code:

public class Client {
    public static void main(String[] args) throws Exception {
        ClientConfig clientConfig = new ClientConfig();
        HazelcastInstance client = HazelcastClient.newHazelcastClient(clientConfig);
        IMap map = client.getMap("test-map");
        map.put("testKey", "testValue");


Debugging containerized applications in the OpenShift cluster can be difficult. In order to attach to the running POD, you can use the following command:

oc exec -ti <pod_name> -- bash