> Isn't the real problem that cancel does not actually mean cancel, > but rather also implies "..and maybe continue"? IMHO cancel should cancel > (and say how much work was performed), while the intention to resume should > be called e.g. "pause". This makes the behaviour clear and prevents > accidental semantic overlap. > IMHO a better choice of wording might also be "interrupt" and "abort", but maybe that's just being pedantic (cancel could then stay in place for backwards compatibility). In any case the current man page for btrfs-progs makes it somewhat clear what's supposed to happen and my confusion was mostly caused by the bug. BTW: Thanks to everyone who helped tracking down the cause of this bug. Haven't had a chance to test the proposed patch yet, but am looking forward to schedule large scrubs across multiple nights.
