Setup

Requirements

• Have a look at Getting Started with Amazon EKS

• Install kubectl 1.21, eksctl and AWS CLI.

• Create an AWS IAM user with at least minimum iam policies for EKS

• Configure your AWS CLI to use your EKS enabled user.

Create EKS cluster

Go to k8s folder:

cd k8s

Create an EKS cluster with some nodes:

eksctl create cluster --name afuentes-webinar-k8s-ha --nodes 3 --version 1.21 --node-type t2.medium --region eu-west-1

Install IKO (InterSystems Kubernetes Operator)

• Create a secret called intersystems-pull-secret to pull images from intersystems container repository:

kubectl create secret docker-registry intersystems-pull-secret \
--docker-server=https://containers.intersystems.com \
--docker-username='<username>' \
--docker-password='<password>' \
--docker-email='<email>'

• Download & unzip iris_operator-3.3.0.120 from WRC SW Distribution.

• Edit iris_operator-3.3.0.120/chart/iris-operator/values.yaml file, and update imagePullSecrets section as:

imagePullSecrets:
  - name: intersystems-pull-secret

• Install IKO chart using helm in your AWS EKS cluster:

helm install intersystems iris_operator-3.3.0.120/chart/iris-operator

• After that, show running pods, you should see InterSystems IRIS Operator:

kubectl get pods

Deploy IRIS in Kubernetes

Create secrets

Define a secret for the IRIS license:

kubectl create secret generic iris-key-secret --from-file=iris.key

Create iris-cpf config map

Create a config-map for configuration files:

kubectl create cm iris-cpf --from-file common.cpf --from-file data.cpf --from-file compute.cpf --from-file CSP.ini

Storage

Define storage class:

kubectl apply -f storage.yaml

Deploy

• Have a look at the iris.yaml configuration.

• Deploy the InterSystems IRIS configuration:

kubectl apply -f iris.yaml

• Show running pods

kubectl get pods
NAME                                          READY   STATUS    RESTARTS   AGE
intersystems-iris-operator-84568955dc-mm6gg   1/1     Running   0          17m
webinar-arbiter-0                             1/1     Running   0          13m
webinar-compute-0                             1/1     Running   0          5m53s
webinar-compute-1                             1/1     Running   0          3m12s
webinar-data-0-0                              1/1     Running   0          12m
webinar-data-0-1                              1/1     Running   0          10m
webinar-webgateway-0                          1/1     Running   0          101s
webinar-webgateway-1                          0/1     Running   0          50s

• Show services, pay attention to the services that references LoadBalancers to webgateways, compute and data pods.

kubectl get svc
NAME                         TYPE           CLUSTER-IP       EXTERNAL-IP                                                               PORT(S)                          AGE
ecp-service                  LoadBalancer   10.100.221.197   aaa5ff964a875403f8d412fb7452adb6-471642292.eu-west-1.elb.amazonaws.com    1972:32665/TCP,52773:32276/TCP   13m
intersystems-iris-operator   ClusterIP      10.100.13.53     <none>                                                                    443/TCP                          17m
iris-svc                     ClusterIP      None             <none>                                                                    <none>                           13m
kubernetes                   ClusterIP      10.100.0.1       <none>                                                                    443/TCP                          41m
webgateway-service           LoadBalancer   10.100.158.12    aaab9045e05d8410aa8cb9a66ed348b0-1326787197.eu-west-1.elb.amazonaws.com   80:31840/TCP                     13m
webinar                      LoadBalancer   10.100.153.110   aaad50e7786e440d19c59f9fb82e0ae2-1725045822.eu-west-1.elb.amazonaws.com   1972:30645/TCP,52773:32655/TCP   13m
webinar-webgateway           LoadBalancer   10.100.191.167   aaa936e8c427d42c996aa24cf9c7f136-1002690704.eu-west-1.elb.amazonaws.com   80:31857/TCP                     6s

• After initialization is completed, you should be able to access to:

  • Data pods - http://webinar-lb:52773/csp/sys/UtilHome.csp
  • Compute pods - http://compute-service-lb:52773/csp/sys/UtilHome.csp
  • Webgateway pods - http://webgateway-service-lb/

• You can even connect to a specific pod in a interactive shell:

kubectl exec --stdin --tty webinar-data-0-0 -- /bin/bash

Run some tests

Keep in mind that in a real environment you could just pull your own Docker image with your application.

In this case, for the sake of simplicity, we will load some code and create web applications.

Load code in the IRIS instance

• We will load & compile some test code in IRIS
• Using Visual Code, set up a connection to your data pod and load and compile the code in /src.

Create web applications

• Connect to webinar-data-0-0, webinar-data-0-1, webinar-compute-0 and webinar-compute-1 and create a simple test web application using:

write ##class(API.Util).CreateWebApp("IRISCLUSTER", "/external/person", "API.RestHandler")

Python client (NativeAPI)

Try running a simple Python (NativeAPI) example that will connect to your compute-lb.

Install nativeAPI (Python)

pip3 install nativeAPI_wheel/irisnative-1.0.0-cp34-abi3-linux_x86_64.whl 

Create an client\ingest.ini file and set up python client configuration like:

[iris]
ip = <compute-service-lb>
port = 1972
namespace = IRISCLUSTER
username = superuser
password = <password>

Run the client:

cd client
python ingest.py

Connected
Inserting data...
id = 139
id = 140
id = 141
id = 142
id = 143
...

Now, while inserting data try to delete you Primary Data pod:

kubectl delete pod webinar-data-0-0

Have a look at the Mirror Monitor in the Data Management Portal and check what happened.

Check how Kubernetes has already started again the pod for you.

REST API

• Have a look at the simple REST API in src/API/RestHandler.cls.
• Open Postman, load client/webinar-k8s-ha.postman_collection.json and check the request.
• The request will use webgateway-lb as target.

Delete Kubernetes deployment

After you have finished testing your deployment, you can delete it.

Go to k8s folder:

cd k8s

Delete the InterSystems IRIS deployment:

kubectl delete -f iris.yaml

Delete persistent volumes:

kubectl delete pvc $(kubectl get pvc -o name | cut -d/ -f2)

Delete iris service:

kubectl delete service iris-svc

Optionally, you can even uninstall IKO:

helm uninstall intersystems

And finally, delete the AWS EKS cluster:

eksctl delete cluster --name afuentes-webinar-k8s-ha --region eu-west-1