Deploy Extension Package#

You can deploy the extension packages manually or automatically, based on your requirements:

Deploy Extension Package Manually#

You can manually deploy pre-curated deployment packages in the catalog, like any other deployment package in the Edge Orchestrator.

The extension packages will be visible in a separate tab called Extensions in Deployment Packages page. For more information on deploying a package, refer to Set up a Deployment.

The extension applications will also be visible in a separate tab called Extensions in Applications page.

If you need to deploy a dependency package and its associated applications before deploying the user application, ensure the dependent deployment packages are successfully installed on all edge clusters.

If you want to deploy dependent packages manually, you do not have to include a deploymentRequirements section in the user application profile.

The Edge Orchestrator will check if the extension or dependency deployment package is already deployed. If it is, the Edge Orchestrator will reuse the existing dependency deployment instead of creating a new one.

Currently, you cannot create or delete extension deployment packages or extension applications; extension applications are pre-curated for use in extension deployment packages.

You can select from different pre-configured profiles when editing an extension deployment package, but you cannot edit an extension application.

Extension deployment packages provide options to override certain values as needed. If you need to change the pre-curated extension deployment packages, contact Intel customer support.

Follow these steps to deploy a dependency deployment package manually:

  1. Go the Deployments tab and click Deployment Packages from the left pane.

  2. Click the Extensions tab in the Deployment Packages page.

  3. Click the three-dot button that corresponds to the package you want to deploy, and select Deploy. In this example, you are deploying the Virtualization dependency deployment package.

  4. Select the profile and click Next to override profile values.

  5. Review the overriding values and click Next to select the deployment profile.

  6. Select Automatic and click Next to provide the deployment details.

  7. Specify the deployment details for the Virtualization Extension and click Next.

  8. Review the deployment details and click Deploy.

  9. Click Deployments from the left pane and note the Virtualization Package being deployed and run.

Deploy Package Automatically through Dependency#

You can establish a dependency between a user application and a deployment package. In this scenario, you need to deploy the primary deployment package. The dependent deployment package will be automatically deployed to all clusters where the user application will be deployed.

If you use automatic deployment for dependent deployment packages, you do not have to include the deploymentRequirements section in the user application profile.

Each user application can have a list of dependent deployment packages. When deploying a user application, all dependent deployment packages will be deployed in parallel with the user application.

When two or more user applications depend on a specific extension or dependency deployment package, Edge Orchestrator will check if the extension or dependency deployment package is already deployed. If it is, Edge Orchestrator will reuse the existing dependency deployment instead of creating a new one.

The forbidsMultipleDeployments flag allows the Edge Orchestrator to determine whether multiple instances of the dependency deployment package can run simultaneously. For pre-curated extension deployment packages, the forbidsMultipleDeployments flag is set to true, by default. If the flag is set to true,` AppOrch` does not allow creating multiple Deployments for this package. Any clusters that require this package will be automatically added to the target cluster list of the existing deployment.

You can include pre-curated extension deployment packages as part of your target application or create your own.

When upgrading deployments, you can upgrade dependency packages by updating the version of the dependency deployment package in the application profile.

Deleting user deployments must come first before deleting dependency extension deployments.

For example, to illustrate the various steps of automatic deployment of extension packages, consider a scenario where you want to deploy a deployment package for a virtual machine application, which must be deployed after Virtualization and CDI.

In this case, you can create the deployment package or use the pre-curated Virtualization Package. You can then add the deployment package information to the list of dependent deployment packages in the virtual machine application. After you have deployed the deployment package for the virtual machine application, Virtualization and CDI will be deployed, followed by the virtual machine application.

To define a dependency, you need to write the application and deployment package yaml files and import them into the user interface:

  1. Create the dependent deployment packages (dependent-deployment-package.yaml) and their applications (dependent-application.yaml). For details, refer to Create Deployment Packages and Add Application.

    “dependent-application.yaml” file:

    specSchema: "Application"
schemaVersion: "0.1"
$schema: "https://schema.intel.com/catalog.orchestrator/0.1/schema"

name: <application_name>
version: <application_version>
description: <application_description>

# User can use pre-created helm registry name which is a part of
# Edge Orchestrator and
# load the helm chart of the application to this registry.
# For more details, see :doc:`/user_guide/package_software/push_registry`.
# User can optionally create new registry in UI and use that name in this
# section. For more details, see :doc:`../add_registry`.
helmRegistry: <helm_registry_name>
chartName: <chart_name>
chartVersion: <chart_version>

# If user application docker image is part of a private repo, then user will
# need to add the imageRegistry details per application in this file.
# User can use pre-created docker image registry "harbor-docker-oci". To create
# registry see :ref:`add_registry`.
# For more details on how to push images to the registry, see :doc:`/user_guide/package_software/push_registry`.
imageRegistry: "harbor-docker-oci"

# Application Profiles are collections of settings that are used when launching an application.
# These settings are application-specific and are typically known to the developer.
# Profiles allow the developer to take this bundle of settings and
# give it a convenient name which can be used to deploy the application.
profiles:
   - name: <application_profile_name>
      displayName: <application_profile_display_name>
      valuesFileName: /path/to/values.yaml
      # If the user wants to use dependent extension deployment_packages for applications, then
      # the below section should be added. For more details on extension packages see :doc:`/user_guide/package_software/extensions/deploy_extension_package`
      # application. The dependent deployment package should be uploaded
      # to the catalog before uploading the user deployment package.
      deploymentRequirements:
      - name: <dependent_deployment_package_name>
         version: <dependent_deployment_package_version>
         deploymentProfileName: <dependent_deployment_package_profile>
      - name: <dependent_deployment_package_name>
         version: <dependent_deployment_package_version>
         deploymentProfileName: <dependent_deployment_package_profile>

#example

specSchema: "Application"
schemaVersion: "0.1"
$schema: "https://schema.intel.com/catalog.orchestrator/0.1/schema"

name: nginx
version: 1.1.3
description: "Nginx for Kubernetes"

helmRegistry: "rs-helm"
chartName: "edge/edge-node/nginx"
chartVersion: "1.1.3"

profiles:
   - name: default
      displayName: "Default"
      valuesFileName: "values-nginx-default.yaml"

    Note

    /path/to/values.yaml contains the overriding values for the Helm* chart.

    “dependent-deployment-package.yaml” file:

    specSchema: "DeploymentPackage"
schemaVersion: "0.1"
$schema: "https://schema.intel.com/catalog.orchestrator/0.1/schema"

name: <deployment_package_name>
displayName: <deployment_package_display_name>
description: <deployment_package_description>
version: <deployment_package_version>

# User can add multiple applications in a deployment package
applications:
      - name: <application_name>
      version: <application_version>

# Deployment profile consists of application profiles for each
# application which is part of the deployment package. Application
# profiles contain the helm chart values which the user wants to update.

# User can create multiple deployment profiles. These will be
# available as options to choose from when user deploys this package through UI
deploymentProfiles:
      - name: <deployment_profile_name>
      displayName: <deployment_profile_display_name>
      applicationProfiles:
         - application: <application_name>
         profile: <application_profile_name>

# This flag decides if multiple deployments of the deployment
# package is allowed or not
forbidsMultipleDeployments: <true/false>

# This field lets user choose the namespace in which the various
# applications will be deployed. If this field is not set, then orchestrator
# generates unique namespace for each deployment and all applications of the
# deployment package are deployed in this namespace. Some system
# namespaces are forbidden to be used for user application E.g: kube-system.
defaultNamespaces:
   <application_1_name>: namespace_1
   <application_2_name>: <namespace_1/namespace_2>
forbidsMultipleDeployments: <flag_forbids_multiple_deployments>

# example
specSchema: "DeploymentPackage"
schemaVersion: "0.1"
$schema: "https://schema.intel.com/catalog.orchestrator/0.1/schema"

name: "nginx-pkg"
description: "nginx support for k8s cluster"
version: 0.0.12
forbidsMultipleDeployments: true

applications:
   - name: nginx
      version: 1.1.3

defaultNamespaces:
   nginx: nginx

deploymentProfiles:
   - name: "default-profile"
      displayName: "Default Configuration"
      applicationProfiles:
      - application: "nginx"
         profile: "default"

  .. note::
To ensure that the deployment of the dependent package is not
duplicated, set `forbidsMultipleDeployments: true`
(default value is false, which means duplicated deployment is allowed).

    “profile.yaml” file:

    #example

# User can add any or all of the fields of the helm chart values file in the profile file for the application.
# Each application should have a separate profile.yaml file. Each application can have multiple profile.yaml files with unique names.
image:
   containerDisk:
      registry: <registry-name>
      repository: <repo-name>
      tag: <image-tag>

    Note

    You can create the deployment-package.yaml, application.yaml, and profile.yaml files and import these together to create a deployment-package in the catalog.

  2. Create the target deployment package and its application.

    “target-application.yaml” file:

    specSchema: "Application"
schemaVersion: "0.1"
$schema: "https://schema.intel.com/catalog.orchestrator/0.1/schema"

name: <application_name>
version: <application_version>
description: <application_description>

helmRegistry: <helm_registry_name>
chartName: <chart_name>
chartVersion: <chart_version>

profiles:
- name: <application_profile_name>
   displayName: <application_profile_display_name>
   valuesFileName: /path/to/values.yaml
   deploymentRequirements:
      - name: <dependent_deployment_package_name>
        version: <dependent_deployment_package_version>
        deploymentProfileName: <dependent_deployment_package_profile>
      - name: <dependent_deployment_package_name>
         version: <dependent_deployment_package_version>
         deploymentProfileName: <dependent_deployment_package_profile>

# example
specSchema: "Application"
schemaVersion: "0.1"
$schema: "https://schema.intel.com/catalog.orchestrator/0.1/schema"

name: my-example-app
version: 1.0.0
description: "My application"

helmRegistry: "harbor-helm-oci"
chartName: "my-example-app"
chartVersion: "1.0.0"

imageRegistry: "harbor-docker-oci"

profiles:
   - name: "default"
   displayName: "Default"
   valuesFileName: "values.yaml"
   deploymentRequirements:
      - name: "nginx-pkg"
         version: "0.0.12"
         deploymentProfile: "default-profile"

    Note

    Application Profile has deploymentRequirements that is the list of dependent deployment packages. Each element of this list has name, version, and profile of the dependent deployment package.

    “target-deployment-package.yaml” file:

    specSchema: "DeploymentPackage"
schemaVersion: "0.1"
$schema: "https://schema.intel.com/catalog.orchestrator/0.1/schema"

name: <deployment_package_name>
displayName: <deployment_package_display_name>
description: <deployment_package_description>
version: <deployment_package_version>

applications:
- name: <application_name>
   version: <application_version>

deploymentProfiles:
- name: <deployment_profile_name>
   displayName: <deployment_profile_display_name>
   applicationProfiles:
      - application: <application_name>
        profile: <application_profile_name>


# example
specSchema: "DeploymentPackage"
schemaVersion: "0.1"
$schema: "https://schema.intel.com/catalog.orchestrator/0.1/schema"

name: "my-example-package"
displayName: "My example package"
description: "Package with my application"
version: "1.0.0"

applications:
   - name: my-example-app
   version: 1.0.0

deploymentProfiles:
   - name: "default"
   displayName: "Default"
   applicationProfiles:
      - application: "my-example-app"
         profile: "default"

  3. Import the yaml files to the Edge Orchestrator. For more information on importing yaml files, refer to Import Deployment Package.

The nginx-pkg package will be deployed automatically when “my-example-package” is deployed.

When you try to delete the extension deployment package (nginx-pkg) when the primary deployment package is running, you will see an error message. Follow these steps to delete the primary deployment package (my-example-package) without deleting the extension package:

  1. Go the Deployments tab and click Deployments from the left pane.

  2. Click the three-dot button corresponding to my-example-package, and select Delete.

  3. In the confirmation prompt, click Delete.

Now, follow these steps to delete the extension deployment package (nginx-pkg):

  1. Go to the Deployments tab and click Deployments from the left pane.

  2. Click the three-dot button corresponding to nginx-pkg, and select Delete.

  3. In the confirmation prompt, click Delete.