Looks like their database choked due to the load: https://github.blog/2022-03-23-an-update-on-recent-service-d... With a single primary taking all the writes for that cluster, it would be tricky to tackle this. #hugops