You are viewing documentation for Kubeflow 1.4

This is a static snapshot from the time of the Kubeflow 1.4 release.
For up-to-date information, see the latest version.

Configure Azure MySQL database to store metadata

How to configure an external Azure Database for MySQL with kustomize to store metadata

This section shows how to use Kubeflow kustomize to configure an external Azure MySQL database to store metadata.

Microsoft Azure Database for MySQL is a relational database service based on MySQL. It provides built-in high availability, data protection using automatic backups and point-in-time-restore for up to 35 days, and automated maintenance for underlying hardware, operating system and database engine to keep the service secure and up to date. Learn more about Azure Database for MySQL.

Table of contents:

  1. Create an Azure database for MySQL
  2. Deploy Kubeflow to use the Azure metadata overlay
  3. Update Kubeflow resources

Create an Azure database for MySQL

First, you need to create an Azure database for MySQL on Azure through either Azure Portal or using Azure CLI:

Remember your Server Name, Admin username, and Password - you’ll be using them later in this guide.

Deploy Kubeflow to use the Azure metadata overlay

You have created the MySQL server database. Next, you need to deploy Kubeflow to use the Azure metadata overlay.

  1. Follow the Install Kubeflow on AKS (Azure Kubernetes Service) guide until the step where you have to build and apply the CONFIG_URI.

  2. If you followed the previous step, you should have downloaded the configuration file. Next, you need to customize the configuration before deploying Kubeflow. Run the following command:

wget -O kfctl_azure.yaml ${CONFIG_URI}

where the ${CONFIG_URL} is the URL pointing to Kubeflow manifest for Azure (for example, .../kubeflow/manifests/v1.1-branch/kfdef/kfctl_azure.v1.1.0.yaml) that you specified in step 1.

  1. Before deploying Kubeflow, use kfctl build to create configuration files:
kfctl build -V -f kfctl_azure.yaml
  1. Under /stacks/azure and resources replace - ../../metadata/v3 with metadata to enable using Azure MySQL.

The updated kustomization.yaml in stacks/azure should look similar to this:

  # Metadata
  # - ../../metadata/v3
  # Uncomment the line below if you want to use Azure MySQL
  - ./metadata
  1. Edit params.env to provide parameters to configuration map as follows:
MYSQL_HOST={YOUR_DB_SERVER_NAME}.mysql.database.azure.com
MYSQL_DATABASE=mlmetadata
MYSQL_PORT=3306
MYSQL_ALLOW_EMPTY_PASSWORD=true

where {YOUR_DB_SERVER_NAME} is your server name.

  1. Edit secrets.env to create a secret with the admin username and password based on your database configuration. Make sure the user name follows the pattern with an “@”, like the one showed below:
MYSQL_USERNAME={ADMIN_USERNAME}@{YOUR_DB_SERVER_NAME}
MYSQL_PASSWORD={ADMIN_PASSWORD}

Deploy Kubeflow

Having completed the previous steps to set up the MySQL server, you can deploy Kubeflow:

kfctl apply -V -f kfctl_azure.yaml

Your metadata database should now be using the Azure Database for MySQL.

To configure the pipeline database using an external Azure Database for MySQL, go to KFP customizations for Azure.


Last modified 20.08.2021: Fix typos in docs (#2884) (e1639edb)