Problem Description

When running with agents in the managed cluster.

addon-controller is deploying drift-detection manager in every managed cluster irrespective of profile configuration.

This causes drift-detection to go in crash loop back off, because ResourceSummary CRD is deployed only when there is a profile with SyncMode: ContinuopsWithDriftDetection

Work-around: delete the drift-detection deployment in the managed cluster.

Fix: upgradeDriftDetectionDeploymentsInManagedClusters must verify whether the drift-detection deployment is present and only then upgrade it.