Introduction
I recently had the pleasure of delivering a presentation at Kscope25 by ODTUG in Dallas, demonstrating an Oracle Database 23ai running on Kubernetes using Oracle Kubernetes Engine (OKE) in Oracle Cloud Infrastructure (OCI) all driven from my MacBook.
After my session I was asked ‘Is it was possible to run Oracle Database 23ai on Kubernetes on a MacBook, if you don’t have access to an on-premises or Cloud Kubernetes environment ?‘
Well the answer is yes, and in this blog I will show how we can create an Oracle Database 23ai (23.8.0.25) using a Kubernetes StatefulSet, how to check the Alert log, connect to the running database, and finally how to delete it.
Getting Started
If you don’t already have podman and minikube installed and configure on your MacBook, you may want to visit my blog post on installing podman and minikube on a MacBook.
Kubectl
To confirm we are using a minikube Kubernetes cluster start a terminal window and type kubectl config get-contexts.
% kubectl config get-contexts
From the below, we can see my current context is minikube.
Use kubectl get nodes to see details of the single node cluster, for example:
% kubectl get nodes -o wide
Podman Desktop
The Podman Desktop provides a Kubernetes Dashboard, from here we can get a per namespace overview of our local Kubernetes environment.
The Dashboard reports Nodes, Deployments, Pods, Services, Ingress & Routes, Persistent Volume Claims, ConfigMaps & Secrets, Jobs and CronJobs.
We will return to the Podman Kubernets Dashboard once we have created our Oracle 23ai database.
Minikube
Minikube provides the ability to push an image into the minikube vm using the minikube image load command.
However, for this blog post I show how we can also ssh into the minikube vm using minikube ssh and pull the Oracle database image from there. I am using this approach as I need to use the –tls-verify = false option on my MacBook.
Before we shell into the minikube vm, let’s confirm what images are already downloaded using the minikube image list command.
Now, SSH into the minikube vm using minikube ssh, then sudo to root using sudo su – and use podman pull to pull the container image using the below.
# podman pull container-registry.oracle.com/database/free:latest --tls-verify=false
Once complete, exit out of the minikube vm and repeat the minikube image list command, we should now see Oracle Database 23ai Free image within the minikube machine.
% minikube image list
Oracle Database 23ai
We are now ready to create a containerised Oracle 23ai Database in our running Kubernetes cluster using the Oracle image already downloaded.
Database Creation
To automate the creation of an Oracle Database 23ai I have created a self-contained Kubernetes StatefulSet which you can find here.
The yaml file creates the following:
- Namespace – dedicated Kubernetes called oracle-namespace
- Secret to store the Oracle Database Password
- ConfigMap for Oracle environmental variables
- Service definition for the Oracle database listener
- VolumeClaimTemplates for persistent storage
- Statefulset to manage the pod.
To create the Oracle Database 23ai use kubectl apply -f <filename.yaml>
% kubectl apply -f 23aiFree_ss.yaml
Database Alert log
Use kubectl logs pod/<pod name> -n <namespace> to inspect the Oracle database Alert log file, for example.
% kubectl logs pod/oracle23aifree-0 -n oracle-namespace
Use the –follow option to continual monitor the log output.
Periodical check the log output, and wait to see
DATABASE IS READY TO USE!
Database Connection – kubectl
To connect to our newly created Oracle 23ai Database we can use execute sqlplus within the database container, for example
% kubectl exec -it pod/oracle23aifree-0 -n oracle-namespace -- sqlplus system/Mysecretpassword@FREE
Or the pluggable database with
% kubectl exec -it pod/oracle23aifree-0 -n oracle-namespace -- sqlplus system/Mysecretpassword@FREEPDB1
SQL> select INSTANCE_NAME, HOST_NAME, VERSION_FULL, EDITION from v$instance;
Minikube Tunnel
An alternative method of connecting to our Oracle 23ai database is by using the minikube tunnel to provide localhost access.
In the below you can see my Kubernetes LoadBalancer service EXTERNAL-IP is showing <pending>. Let’s fix that.
% kubectl get all -n oracle-namespace
Start a minikube tunnel in a separate terminal window and return to the original terminal window.
If we now perform kubectl get service -n <namespace> we should see the EXTERNAL-IP has now be allocated the localhost IP address (127.0.0.1), which we can now use to connect to the database.
% kubectl get service -n oracle-namespace
Database Connection – minikube tunnel
With the minikube tunnel in place, connect using SQLcl with localhost, as below.
% sql system/Mysecretpassword@//localhost:1521/FREE
Podman Desktop
Return to the Podman Desktop, navigate to Kubernetes -> Dashboard and select Namespace -> oracle-namespace from the pull-down list on the righthand side.
From here we can explore the Kubernetes objects we have created, for example, click on Pods, select oracle23aifree-0.
From here we can see:
- Summary – summary of the configuration of the Pod
- Logs – View of the Oracle Alert log file
- Inspect – View of the running Pods JSON
- Kube – Editable YAML of the running Pod
- Terminal – Shell access into the Pod
Database Deletion
To delete the Oracle 23ai (23.8.0.25) Database, use kubectl delete -f <filename.yaml> for example.
% kubectl delete -f 23aiFree_ss.yaml
Summary
In this blog post I have shared how to create an educational environment on a MacBook which can use to learn Kubernetes and Oracle 23ai new features.
In a future post I share how we can use the Oracle Database Kubernetes Operator (aka OraOperator) to simplify database operations.