Disk usage discrepancy?

Absolutely the same suggestions:

  1. Forget untrusted satellites using the --force flag: How To Forget Untrusted Satellites
  2. You need to enable the scan on startup, if you disabled it (it’s enabled by default)
  3. Restart the node
  4. Periodically check your logs for errors related to the databases and/or filewalkers (search for error and database, error and filewalker)
  5. If you have errors, you need to fix them, otherwise usage will remain the same or reset on the restart.
  6. Wait for several days (maybe weeks - depends on how slow your disks are responds) for finishing of the used-space-filewalker for each trusted satellites, to track you can search for used-space-filewalker and started|completed in your logs.