Client is using the cli to disable as service (custom httpd). Every time they do so, the ssh client they are logged in on get killed. They are able to re-log in immediately and the service is disabled sucessfully.
All processes with open references on a cluster-managed file system get killed when the service is disabled. This means if you 'cd' to there, your home directory was on it, or your shell was started in it, or sshd was using something on it (for instance, I have custom sshd services which start in the cluster mount point for each service), those processes will get killed in order to clear the references on the file system so that it can be unmounted.