Fix disk/memory alarms

This change restores the correct order for the memory/disk triggers. It
also applies 'group_by: hostname' for those alarms because an alarm on
a single node has a direct impact on the cluster.
1 file changed