I like how easy it is to install from scratch or migrate an existing system by uploading configuration files (as opposed to having someone do this).
It's not great at logging errors / exceptions in some cases where things are going wrong - especially when there isn't much useful information logged out of those events. This would be helpful if we could get more details about what happened during these kinds of issues but they don't seem very forthcoming here either! We have been able to monitor our Kubernetes clusters better than ever before now thanks to coscale which has allowed us to identify nodes running slowly without needing manual inspection/reporting etc.