This is likely a gitops issue, but worth noting here and start keeping track of it. We see this quite rarely but it has happened 1 (out of 20) where the cluster-wide argo on the spoke is just stuck in yellow syncing state and does not proceed. iirc we also had a gitops bug open about this, I need to search for it and rehash things since this is still seemingly true on gitops-1.17