how can we check if a bcachefs mirror is consistent and not affected by kernel bugzilla #99171 ?

Roland <[email protected]> Sat, 11 Oct 2025 13:18:07 +0200
Newsgroups org.kernel.vger.linux-bcachefs
Message-ID <[email protected]>
hello,

i'm playing with bcachefs , as it's nice to see that there is another=20
powerful cow filesystem in development.

i would like to make sure that it is not affected by=20
https://bugzilla.kernel.org/show_bug.cgi?id=3D99171 , which is about=20
drbd/mdraid/btrfs/lvm suffering from raid mirror inconsistencies under=20
special circumstances when written to using O_DIRECT.

at least i can see that development of bcachefs has focus on O_DIRECT

https://patchwork.kernel.org/project/linux-fsdevel/patch/20230509165657.17=
[email protected]/

one question to make sure that my testing gives reliable result:

does "bcachefs data scrub /fs" scrub both sides of the mirror in=20
parallel AND while doing that, does it compare, if data blocks on each=20
device really are the same / have the same checksum , so that we would=20
see if O_DIRECT writes go wrong , probably reaching only one side of the=
=20
mirror ?

root@debian13:~# bcachefs=C2=A0 data scrub /bcachefs
Starting scrub on 2 devices:=C2=A0 sdb sdc
device=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0checked=C2=A0=
 =C2=A0corrected uncorrected=C2=A0 =C2=A0 =C2=A0 =C2=A0total
sdb=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 518 MiB=
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A00 B=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A00 B=
=C2=A0 =C2=A0 =C2=A0518 MiB 100%=C2=A0=20
complete
sdc=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 518 MiB=
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A00 B=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A00 B=
=C2=A0 =C2=A0 =C2=A0518 MiB 100%=C2=A0=20
complete

Roland
sysadmin

p.s.:
i'm not subscribed to this list, so please keep me CC'd.