e2fsck -y says "yes" to "Abort?"

e2fsck -y says "yes" to "Abort?"

[Eric Sandeen]

I've got this bug filed against Fedora:

Bug 484913 -  Change verbage of e2fsck questioning to make the "-y"
switch more useful

and I think he has a point:

sh-3.2# e2fsck -y /dev/VolGroup00/VolVol02
e2fsck 1.41.3 (12-Oct-2008)
The filesystem size (according to the superblock) is 14090240 blocks
The physical size of the device is 8847360 blocks
Either the superblock or the partition table is likely to be corrupt!

Abort? yes

sh-3.2#

I'm reluctant to invert the logic of the Abort? question as suggested
("Are you sure you want to continue?") because this is a significant
enough problem that we probably should really pause for consideration.

But it seems like perhaps stopping at "Abort?", allowing the user to say
"n" to that and then let the "-y" flag take over from there would be
reasonable.

If this sounds ok I'll whip up a patch, something like a way to flag the
really serious questions (?) as unaffected by -y.

Thanks,
-Eric

Re: e2fsck -y says "yes" to "Abort?"

[Theodore Tso]

On Sat, Apr 18, 2009 at 09:10:50AM -0500, Eric Sandeen wrote:
> I've got this bug filed against Fedora:
> 
> sh-3.2# e2fsck -y /dev/VolGroup00/VolVol02
> e2fsck 1.41.3 (12-Oct-2008)
> The filesystem size (according to the superblock) is 14090240 blocks
> The physical size of the device is 8847360 blocks
> Either the superblock or the partition table is likely to be corrupt!
> 
> Abort? yes
> 
> I'm reluctant to invert the logic of the Abort? question as suggested
> ("Are you sure you want to continue?") because this is a significant
> enough problem that we probably should really pause for consideration.
> 
> But it seems like perhaps stopping at "Abort?", allowing the user to say
> "n" to that and then let the "-y" flag take over from there would be
> reasonable.
> 
> If this sounds ok I'll whip up a patch, something like a way to flag the
> really serious questions (?) as unaffected by -y.

Seems reasonable to me; we'll have to update the documentation to
explain that -y really doesn't mean yes to _all_ questions, but that
seems like the best approach.

					- Ted

Re: e2fsck -y says "yes" to "Abort?"

[Ric Wheeler]

Theodore Tso wrote:
> On Sat, Apr 18, 2009 at 09:10:50AM -0500, Eric Sandeen wrote:
>   
>> I've got this bug filed against Fedora:
>>
>> sh-3.2# e2fsck -y /dev/VolGroup00/VolVol02
>> e2fsck 1.41.3 (12-Oct-2008)
>> The filesystem size (according to the superblock) is 14090240 blocks
>> The physical size of the device is 8847360 blocks
>> Either the superblock or the partition table is likely to be corrupt!
>>
>> Abort? yes
>>
>> I'm reluctant to invert the logic of the Abort? question as suggested
>> ("Are you sure you want to continue?") because this is a significant
>> enough problem that we probably should really pause for consideration.
>>
>> But it seems like perhaps stopping at "Abort?", allowing the user to say
>> "n" to that and then let the "-y" flag take over from there would be
>> reasonable.
>>
>> If this sounds ok I'll whip up a patch, something like a way to flag the
>> really serious questions (?) as unaffected by -y.
>>     
>
> Seems reasonable to me; we'll have to update the documentation to
> explain that -y really doesn't mean yes to _all_ questions, but that
> seems like the best approach.
>
> 					- Ted
> --
>   

The only down side is when you try to automate this (say in an 
appliance) and you don't have a human reading the output. In this case, 
you might just want to invert the logic but in general, it does seem 
dangerous to invert the logic for a long standing option,

Ric

Re: e2fsck -y says "yes" to "Abort?"

[Eric Sandeen]

Ric Wheeler wrote:

> The only down side is when you try to automate this (say in an 
> appliance) and you don't have a human reading the output. In this case, 
> you might just want to invert the logic but in general, it does seem 
> dangerous to invert the logic for a long standing option,
> 
> Ric
> 

Maybe make something like "-yy" automatically answer "n" to anything
that would stop the fsck, and answer "y" to anything that it proposes
fixing?

-Eric

Re: e2fsck -y says "yes" to "Abort?"

[Ric Wheeler]

Eric Sandeen wrote:
> Ric Wheeler wrote:
>
>   
>> The only down side is when you try to automate this (say in an 
>> appliance) and you don't have a human reading the output. In this case, 
>> you might just want to invert the logic but in general, it does seem 
>> dangerous to invert the logic for a long standing option,
>>
>> Ric
>>
>>     
>
> Maybe make something like "-yy" automatically answer "n" to anything
> that would stop the fsck, and answer "y" to anything that it proposes
> fixing?
>
> -Eric
>   
That would be useful for scripting users - it is already assumed to be 
wildly dangerous to run it in "yes" mode I would assume in any case :-)

ric

Re: e2fsck -y says "yes" to "Abort?"

[Andreas Dilger]

On Apr 20, 2009  14:38 -0400, Ric Wheeler wrote:
> Eric Sandeen wrote:
>> Ric Wheeler wrote:
>>> The only down side is when you try to automate this (say in an  
>>> appliance) and you don't have a human reading the output. In this 
>>> case, you might just want to invert the logic but in general, it does 
>>> seem dangerous to invert the logic for a long standing option,
>>
>> Maybe make something like "-yy" automatically answer "n" to anything
>> that would stop the fsck, and answer "y" to anything that it proposes
>> fixing?
>   
> That would be useful for scripting users - it is already assumed to be  
> wildly dangerous to run it in "yes" mode I would assume in any case :-)

Well, all of the questions are designed to choose a safe option when
run with "-y", and to avoid changing the filesystem when run with "-n".

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.