with our very first 3500 SNO scale test run for ACM 2.16/OCP 4.21,   all the cluster are deployed but only 500 are cgu complaint.  
The cgu log has 574 lines similar to below2026-01-16T12:49:52Z	ERROR	Reconciler error	{"controller": "clustergroupupgrade", "controllerGroup": "ran.openshift.io", "controllerKind": "ClusterGroupUpgrade", "ClusterGroupUpgrade": {"name":"vm02847","namespace":"ztp-install"}, "namespace": "ztp-install", "name": "vm02847", "reconcileID": "6c701ca6-085a-4e9e-a780-d4b0e1f74081", "error": "storage is (re)initializing: failed to list apps.open-cluster-management.io/v1, Kind=PlacementRule: unable to determine group/version/kind: cbor: exceeded max number of elements 1024 for CBOR array"}
All other cgu without the error  look like just stuck at new state
oc get PlacementRule cmd returns the same error
# oc get placementRule -AError from server: unable to determine group/version/kind: cbor: exceeded max number of elements 1024 for CBOR array
Before this we hit another issue with gitops application controllar complaints about the similar cbor array exceed max mumber of 1024 and we have to work around it by reducing the clusters per gitops app from 1000 to 250. refer details in bug OCPBUGS-73812 and slack thread : https://redhat-internal.slack.com/archives/CMP95ST2N/p1768253414929519
so looks like this is more like a OCP 4.21.0-rc1 fundamental error, I’m  not sure which component, bit looks like apiserver is a good one to start 

    1.
    2.
    3.