As described in [Hypershift issues at scale - OCPBUGS-52256|https://issues.redhat.com/browse/OCPBUGS-52256], the memory usage of hypershift quickly grows causing issues. 

One persistent error that shows up in the logs is:

{"level":"error","ts":"2025-03-17T14:14:17Z","msg":"Reconciler error","controller":"hostedcluster","controllerGroup":"hypershift.openshift.io","controllerKind":"HostedCluster","HostedCluster":{"name":"cluster1","namespace":"control"},"namespace":"control","name":"cluster1","reconcileID":"e785ff72-6253-48bc-ba86-8bc63da26c10","error":"failed to determine if release image multi-arch: failed to retrieve manifest us.icr.io/am/ocp-release:4.18.1-x86_64: toomanyrequests","stacktrace":"sigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller).reconcileHandler\n\t/hypershift/vendor/sigs.k8s.io/controller-runtime/pkg/internal/controller/controller.go:324\nsigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller).processNextWorkItem\n\t/hypershift/vendor/sigs.k8s.io/controller-runtime/pkg/internal/controller/controller.go:261\nsigs.k8s.io/controller-runtime/pkg/internal/controller.(*Controller).Start.func2.2\n\t/hypershift/vendor/sigs.k8s.io/controller-runtime/pkg/internal/controller/controller.go:222"}

This leads to a steady memory leak which is caused in part by the lookups when the image reference is a named tag (for example, us.icr.io/am/ocp-release:4.18.1-x86_64) instead of a canonical name. Since the cache always uses the image reference ID (canonical name), the cache check will fail when the image reference has a named tag.

This is true for:
* GetManifest - https://github.com/openshift/hypershift/blob/00164ec3e6df7c140283872bb6fe6d524f76cc75/support/util/imagemetadata.go#L221
* ImageMetadata - https://github.com/openshift/hypershift/blob/00164ec3e6df7c140283872bb6fe6d524f76cc75/support/util/imagemetadata.go#L72

This appears to have been introduced in: https://github.com/openshift/hypershift/commit/1af27dbff1cc063c8bb29ac4c6543eb466d1ebe6

    1.
    2.
    3.