Upgrading Edge Infrastructure Manager
=====================================
Edge Orchestrator is expected to support seamless upgrades of the Infrastructure Manager helm charts between versions 3.0 and 3.1.
Therefore, intermittent changes in the infra-charts repository must be verified against Edge Orchestrator version 3.0 to ensure they do not introduce breaking changes,
and that the Orchestrator remains fully functional.
Changes in the SB and NB APIs, as well as database schema updates, are validated during the CI phase through a set of automated lint and integration checks.
To test the upgradeability of infra charts with changes proposed in a given infra-charts pull request, we introduced the `upgrade-test `_ workflow.
This test can be manually triggered on any PR in the infra-charts repository, and is also executed periodically on the main branch to ensure long-term compatibility.
Additionally, users can manually test a newly released version of an Infra Helm chart in a deployed Edge Orchestrator using Argo CD's web UI.
Specifically, the Orchestrator operator can modify the Helm chart version in the following Argo CD Infra manager applications of the Edge Manageability Framework:
- infra-managers
- infra-core
- infra-external
- infra-onboarding
Through the platform's web UI, the user can verify the upgraded Orchestrator by onboarding an Edge Node (EN), monitoring its status, and performing an EN upgrade.
Upgradeability Requirements
---------------------------
The following requirements must be met to consider the system upgradeable:
1. The Orchestrator operator can perform an in-place upgrade of the Edge Infrastructure Manager applications.
2. Any new version of OS Profiles is automatically populated, and end users can upgrade their configured Edge Nodes in place.
3. Downtime is allowed for individual components or services during upgrade activities.
4. The Orchestrator upgrade must not impact the operational state of existing Edge Nodes.
5. If the upgrade fails, manual recovery is possible through rollback procedures.
Steps to Upgrade Edge Manageability Apps
----------------------------------------
The following section provides step-by-step instructions to perform an upgrade of one of the Edge Manageability applications (e.g., ``infra-managers``) and to verify the above requirements.
Prerequisites
^^^^^^^^^^^^^
1. Deploy an Edge Orchestrator v3.0
2. Obtain your JWT_TOKEN
3. Create User, Organization, and Project
Register and Onboard an EN
^^^^^^^^^^^^^^^^^^^^^^^^^^
Perform EN registration and onboarding by following the steps in the `EN registration guide `_.
On the ``Host`` page of the Web UI, observe the status of the EN until it is fully provisioned and onboarded.
Upgrade Infra Manager Version in Argo CD
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1. Log into the Argo CD UI for your deployment.
2. Disable autosync in the ``root-app`` application under ``Details``.
3. Search for the infra application you want to upgrade (e.g., ``infra-managers``).
4. Open its manifest and enable editing.
5. Update the Helm chart version by modifying the targetRevision field in the application's manifest, then save the changes.
6. Wait until the application reaches the ``Healthy`` state (this usually takes up to a minute). Meanwhile, monitor the EN status — it should remain unchanged.
Steps to Test Upgraded Edge Manager
-----------------------------------
The upgraded Edge Orchestrator should display all applications in a Healthy state.
The functionality of the Edge Manager can be further verified by confirming it is able to discover new OS profiles and perform EN OS updates.
Verify OS Profiles
^^^^^^^^^^^^^^^^^^
After the Edge Orchestrator upgrade, OS Profiles should still be automatically populated in the Web UI.
To verify if the OS profiles are automatically repopulated follow the steps:
1. Log into the web UI for the deployment.
2. Go to ``Settings`` and open the ``OS Profiles`` tab to see all OS profiles.
3. Identify the ``ResourceID`` of an unused profile:
.. code-block:: bash
CLUSTER_FQDN=edgeorchestration.example.com
PROJ_NAME=sample-project
curl -X GET \
-H "Authorization: Bearer ${JWT_TOKEN}" \
https://api.${CLUSTER_FQDN}/v1/projects/${PROJ_NAME}/compute/os
4. Use the OS profile’s ResourceID to delete an unused profile via an API call (e.g., RESOURCE_ID):
.. code-block:: bash
RESOURCE_ID=os-cd58ac2
curl -v -X DELETE "https://api.${CLUSTER_FQDN}/v1/projects/${PROJ_NAME}/compute/os/${RESOURCE_ID}" \
-H "accept: application/json" \
-H "Authorization: Bearer ${JWT_TOKEN}"
Soon, the deleted OS profile should be rediscovered by OS Resource Manager and displayed again in the Web UI. The OS profiles are updated by OS Resource Manager every 60 minutes.
Trigger EN Update
^^^^^^^^^^^^^^^^^
An update of the onboarded EN OS will test if the EN and the Infrastructure Manager are still fully functional after the Edge Orchestrator upgrade.
1. Manually add an OS Profile using the latest EMT image details, for example:
.. code-block:: bash
OS_RESOURCES='{
"architecture": "x86_64",
"imageId": "3.0.20250514.2200",
"imageUrl": "files-edge-orch/repository/microvisor/non_rt/edge-readonly-3.0.20250514.2200.raw.gz",
"installedPackages": "",
"kernelCommand": "",
"name": "Edge Microvisor Toolkit 3.0.20250514",
"osType": "OPERATING_SYSTEM_TYPE_IMMUTABLE",
"profileName": "microvisor-nonrt",
"repoUrl": "files-edge-orch/repository/microvisor/non_rt/edge-readonly-3.0.20250514.2200.raw.gz",
"securityFeature": "SECURITY_FEATURE_NONE",
"sha256": "f56ba0b338434813b70ecd8b81dec9a8f389a344ba868454825b4442217f6428",
"updateSources": [],
"osProvider": "OPERATING_SYSTEM_PROVIDER_INFRA"
}'
curl -X POST \
-H "Accept: application/json" \
-H "Authorization: Bearer ${JWT_TOKEN}" \
--data "$OS_RESOURCES" \
--header "Content-Type: application/json" \
https://api.${CLUSTER_FQDN}/v1/projects/${PROJ_NAME}/compute/os
2. Identify your EN instance ResourceID in the list of instances and set ``INSTANCE`` environment variable accordingly:
.. code-block:: bash
curl -X GET \
-H "Authorization: Bearer ${JWT_TOKEN}" \
https://api.${CLUSTER_FQDN}/v1/projects/${PROJ_NAME}/compute/instances
3. Identify the OS Profile ResourceID in the list of profiles and set ``OSPROFILE`` environment variable accordingly.
.. code-block:: bash
curl -X GET \
-H "Authorization: Bearer ${JWT_TOKEN}" \
https://api.${CLUSTER_FQDN}/v1/projects/${PROJ_NAME}/compute/os
4. Update the EN instance's ``desired_os`` to point to the OS profile:
.. code-block:: bash
curl -X PATCH \
-H "Accept: application/json" \
-H "Authorization: Bearer ${JWT_TOKEN}" \
--data '{"osId": "${OSPROFILE}"}' \
--header "Content-Type: application/json" \
https://api.${CLUSTER_FQDN}/v1/projects/${PROJ_NAME}/compute/instances/${INSTANCE}
5. Observe that the ``OS upgrade available`` note appears on the host page.
6. Schedule a maintenance window for your EN by following the `guide on maintenance window creation `_.
7. Monitor the ``Update`` status to confirm the update was completed. The ``No new updates available`` status message indicates a successful update.
Perform Version Rollback
------------------------
In the event of an unsuccessful upgrade, the Orchestrator operator should be able to revert manually via the Argo CD UI. Follow these steps:
1. Log into the Argo CD UI.
2. Search for and select the application you want to roll back.
3. Open the ``History and Rollback`` tab.
.. image:: images/rollback-tab.png
:alt: Rollback example
4. Select the entry with the previous version of your application
5. Select ``Rollback`` from the three dot menu.
.. image:: images/rollback-example.png
:alt: Rollback example
6. Ensure the application status is ``Healthy`` and that it reflects the previous version.