Dne 08. 09. 23 v 13:32 Christian Brauner napsal(a):
I'd say there are several options and we should aim towards the variant
which is most usable by normal users.
None of the options is sufficiently satisfying to risk intricate
behavioral changes with unknown consequences for existing workloads as
far as I'm concerned.
I'm not convinced anyone has the 'fsfreeze' + 'unmount' as a regular workload.
Thus solving this unusual race case shouldn't be breaking anyones else
existing workload.
ATM there is no good solution for this particular case.
So can you please elaborate which new risks are we going to introduce by
fixing this resource hole ?
Zdenek
