same as for any other error at that point in the existing code. We
considered keeping the item and leaving the balance paused so the
resume could be retried, but chose to stay consistent with the existing
-error handling. Nothing is lost by ending the balance: a failed
-write-back at resume happens before anything is relocated, and after a
-failed limit update the chunks already relocated stay relocated. The
+error handling. Ending the balance undoes no completed work: a failed
+write-back at resume happens before the resumed balance relocates
+anything, and after a failed limit update the chunks already relocated
+stay relocated. The remaining work is not done, and the user has to
+start a new balance for it, as after any other balance error. The
failed write-back at resume is logged, because the mount-time resume
thread would otherwise drop the error silently.