Oracle Cloud Infrastructure Documentation

Add-on Reconciliation

On Compute Cloud@Customer, the OKE Service includes a reconciliation process that periodically evaluates the state of the add-on and updates the add-on if necessary.

Note

You should not install, configure, update, or delete add-ons manually. Use the OKE Service installation, configuration, update, and delete interfaces.

The reconciliation process behaves differently depending on the state of the add-on.

Active State

For add-ons in ACTIVE state, the reconciliation process runs every twelve hours. If resources have been manually deleted, the process detects the change and attempts recovery. Full recovery isn't guaranteed.

If full recovery is successful, the add-on state returns to ACTIVE.

If recovery is partial or fails, the add-on state changes to NEEDS_ATTENTION.

Needs Attention State

For add-ons in NEEDS_ATTENTION state, the reconciliation process runs every few minutes, not every twelve hours, while the add-on remains in the NEEDS_ATTENTION state. The process checks whether all deployments associated with the add-on are ready, whether all pods are healthy. The interval between reconciliation process runs is a little longer each time.

Some issues, such as unschedulable nodes, might resolve during the reconciliation process. Other issues, such as configuration problems, require user intervention to fix the problem. If the add-on remains in NEEDS_ATTENTION state after the reconciliation process has run, try to identify the issues.

Recovery actions that might be appropriate for a user to take include the following:

  • Ensure that node pools in the cluster have at least one node available. If no nodes are available, the add-on pods can't be scheduled, the add-on can't be deployed.

  • Ensure the configuration values and other settings are correct.

  • Update the add-on as needed. Updating the add-on will trigger another reconciliation process.

  • Disable and reinstall the add-on.

If the add-on recovers, the add-on is transitioned back to ACTIVE state.

If the add-on remains unhealthy, the system schedules the next check.

The reconciliation process continues to run for approximately 12.5 hours if the add-on remains in NEEDS_ATTENTION state. After the reconciliation process stops running, the add-on remains in NEEDS_ATTENTION state indefinitely. After approximately 30 minutes, the work request moves to FAILED state.

The following example shows an add-on in NEEDS_ATTENTION state.

$ oci ce cluster get-addon --cluster-id ocid1.cluster.unique_ID --addon-name WeblogicKubernetesOperator
{
  "data": {
    "addon-error": {
      "code": "409",
      "message": "Incorrect state for CR",
      "status": "IncorrectState"
    },
    "configurations": [
      {
        "key": "numOfReplicas",
        "value": "0"
      },
      {
        "key": "weblogic-operator.ContainerResources",
        "value": "{'limits': {'cpu': '500m', 'memory': '200Mi'}}"
      },
      {
        "key": "weblogic-operator-webhook.ContainerResources",
        "value": "{'limits': {'cpu': '200m', 'memory': '300Mi'}}"
      }
    ],
    "current-installed-version": "",
    "lifecycle-state": "NEEDS_ATTENTION",
    "name": "WeblogicKubernetesOperator",
    "time-created": "2025-02-26T01:41:52.020696+00:00",
    "version": null
  },
  "etag": "5f3eef22-eb32-5c2c-774c-c7a98836a13a"
}