[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [lvm-devel] [PATCH v3 13/18] fsadm: remove -y (YES) option



Dne 4.10.2011 08:29, Lukas Czerner napsal(a):
On Mon, 3 Oct 2011, Zdenek Kabelac wrote:

Dne 3.10.2011 18:39, Lukas Czerner napsal(a):
On Tue, 27 Sep 2011, Zdenek Kabelac wrote:

Dne 27.9.2011 15:42, Lukas Czerner napsal(a):
There is some confusion in using -y (YES) and -f (FORCE) options in
fsadm. In some cases we are asked for yes/no question which can be
override by -f option, but not by -y option. Usually most of the
questions
tools ask for are yes/no and it can be overridden by forcing it with -f
(e.g. fsck.(extN|xfs), lvm and others...) so it make sense to get rid of
-y option and use only -f instead.

Also I do not think it is wise to use -y option in fsck.extN since
people using fsadm would probably not know how it works, so we should
NOT provide them with that option, but rather let them use "real" fsck
instead (and let them read man page if needed). Also running fsck with
-y when you have corrupted file system is probably not a good idea from
multiple reasons. This is also fixed by this commit.

This commit removes '-y' option and use '-f' instead. With exception of
fsck.


NACK

-f  and  -y  are different.

While  fsck will proceed with -f  on mounted file system (leading to
certain
damage) -y  option will stop here and just answer -y  to  question about
umount.

I am not saying that -f should be used instead of -y.

That is why having both options for the fsadm does not make sense,
because it is not just fsck which fsadm is using internally. Force means
force, we are trying to simplify things here not map every argument of
every tool into fsadm.

When user force things, he should know what is he doing, since he really
is using "force" :). -f option is quite generic and most of the tools
does have it, however -y options is specific, that is why fsadm does not
have -n option, or even -p option.

Generally I am in favour of changing check command to only check for
file system consistency and if problems are found report to the user
that he should run proper fsck by itself, since it is a bit delicate
situation and the user should really know what is he doing if he does
not want to lose data.

Finally I think that having both '-f' and '-y' option which are really
inconsistent among the tools does not make sense.

Well think more about the difference between -y and -f.

If something does not make sense to you, the first solution is to remove it,
the second (harder) is to try to think about it deeply and try to understand
for multiple reasons of this NACK - and from your other thread replies, it
also applies to your other  'does not make sense to me' comments...

Zdenek


I think I gave you multiple reasons why it does not make sense, if you
do not want to listen, I do not really care :) but it does to change the
fact that it is not consistent and it is confusing. No talking about the
fact that doing fsck with -y might be really bad idea if you do not know
what you're doing, or if you at least do not have metadata backup.


Well simply take it as a fact -y  and -f  will stay.

And please try to read what has been written (or even said) about the reasons these options have to stay supported - like all other options which are already part of fsadm.

Zdenek


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]