In one of our partners environment, they have been trying the scale in/out worker node procedure in this https://docs.openshift.com/container-platform/4.14/scalability_and_performance/ztp_far_edge/ztp-advanced-install-ztp.html#ztp-deleting-node-siteconfig_ztp-advanced-install-ztp . It seems that during same timing with scale-in out process, bmc-credentials secret which was used by this cluster have been recreated automatically and BMH resources of other nodes became in "Registration Error" state. When we now check BMHs of spoke cluster in hub cluster, almost all of the BMH resources are in "registration error" (only scaled in/out one is not in error state).

When we check the details of these BMHs it shows that "Error Message:  BMC CredentialsName secret doesn't exist The BMC secret fi-914/fi-914-bmc-credentials does not exist". However, this secret already exists in the hub cluster (also credentials in this secret are correct).

Even after recreated the secret, the error persists.

    4.14.3

    bmc-credentials secret of spoke cluster can be deleted from hub cluster and this "registration error" is shown in BMH resources of spoke cluster in hub cluster. Even after recreating this bmc-credentials of spoke cluster in hub cluster, this "registration error" persists.

    1.Delete bmc-credentials secret of spoke cluster can be deleted from hub cluster
    2. Check error message in BMH resources of spoke cluster in hub cluster
    3. Recreate bmc-credentials secret of spoke cluster
    4. Check the error message in BMH resources

    "Registration error" in BMH resources persists even bmc-credentials are recreated

    After bmc-credentials secret is recreated, it was expected for BMH resources to update theirself and get rid of the error message.