Our current solution is simple enough. We run a script on the target to see if any filesystem is too full, and report WARNING or ERROR based on a local config file. The filesystem name and usage is propagated to the action, and our Operations staff uses the notes from the action email to determine the escalation.