how to ensure leap second validity for long running processes

As of the latest version, an expiration check happens during every call to access the leap second data, and a warning is issued if the leap data is expired. Would we want to do anything more severe than that? options:

• throw an exception (too severe since the expiration doesn't mean the data isn't necessary current still valid)
• attempt to update the user cache in situ
• spin off a background task to update the user cache
• ...