If you are certainly sure, you have restarted the instance and the OS space and new size  after startup (as seen in system management) portal don't match, a 'truncate' will release the OS space immediately. I've encountered same issue  and this is what i've done.

Alternatively, you can create your own separate database that you are able to manage yourself and mark the setting of Journal to 'No'. Then map all not so important globals to that DB.

Thanks Mark - am also looking at the profiles of these blades - and depending on the workload - the HPE Synergy 660 Gen10 is more on database side with need for more cores/larger memory.