[dm-devel] RE:Fwd: [linux-lvm] 2.4.20 + xfs +lvm2 = raid0_make_request bug

[Svetoslav Slavtchev]

>On Wed, Jan 29, 2003 at 08:32:44PM +0100, Svetoslav Slavtchev wrote:
>> useing lvm2(with lvm1 everything is working perfect for the last 6 months)
>>  when i try to mount a xfs filesystem wich is on 
>> lv over soft raid-0, i'm getting just like in
>> 2.5.2x - 2.5.4x (i didn't try later kernels) :
>> 
>> ############################
>> raid0_make_request bug: can't convert block across chunks or bigger than 16k
>> 25181311 4
>> raid0_make_request bug: can't convert block across chunks or bigger than 16k
>> 25181343 4
>> ############################
>> 
>> is this know problem, is it fixed ?
>This is an XFS problem, search the list for info on using the new
>journal format.  The reason it has been working for LVM1 is that XFS
>has some hard coding in that recognises the LVM1 major number, and
>drops back journal to to 512byte chunks.

i think i misunderstand something, i'm trying to mount my old lvm1 volumes,
if xfs crated them with 512byte chunks cause of this hard coded major number,
why should i have troubles when i try to mount them using lvm2 


> i thought that the problem exist only in 2.5 kernel
>This shouldn't be a problem with 2.5.  If you have seen similar
>problems please send me details.
>
i'll  drop a line as soon as i try it out

best,

svetljo

PS.
on more question,
in case i find a way to make a back-up to convert 
to other chunk size for the log( not very likely:( )
which chunk size should i use ?
will it impact performance ?