Hello, I have a problem with rancher after the upgrade to 2.6.6 : impossible to make a

kubectl

using the kubeconfig in another computer for two specific clusters (olders EKS cluster saws as ). Both clusters are in 1.21 version. I have this error after all `kubectl`/`helm` commands :

Error from server (InternalError): an error on the server ("unable to create impersonator account: ClusterUnavailable 503: ClusterUnavailable 503: cluster not found") has prevented the request from succeeding

One other weird thing : I can't use execute shell or logs directly in rancher GUI for this two clusters. It works fine before the 2.6.6. I decide to upgrade to rncher 2.6.8. But, only for this two clusters, the cluster-agent upgrade is not propaged (I'm still in clter-agent 2.6.6 for them) After navigate in rancher console, I saw, one of this clusters in error in cluster management page. It said the cluster is unavailable (But I can navigate and manipulate it in rancher interface). I research for it and I found this command :

kubectl patch <http://clusters.management.cattle.io|clusters.management.cattle.io> <REPLACE_WITH_CLUSTERID> -p '{"status":{"agentImage":"dummy"}}' --type merge

I try this for the "unavailable" cluster but no changes. I decide to watch the logs in the rancher deployment and I get this error :

[ERROR] [secretmigrator] failed to migrate service account token secret for cluster c-wrh8l, will retry: Operation cannot be fulfilled on <http://clusters.management.cattle.io|clusters.management.cattle.io> "c-wrh8l": the object has been modified; please apply your changes to the latest version and try again

Somebody can help me ?