Re: lvm vs traditional partitioning
On Fri, Dec 29, 2006 at 11:06:38AM -0400, E0x wrote:
> i asking it because i was thinking in use lvm in desktop setup , and i can
> live with a harddisk lose and the data on it , but not with all data lost
>
> pd: i have some small HD
>
> On 12/29/06, Roberto C. Sanchez <roberto@connexer.com> wrote:
> >
> >On Fri, Dec 29, 2006 at 10:50:53AM -0400, E0x wrote:
> >> a question about lvm , if i have 3 harddisk in a lvm setup for save
> >data ,
> >> and dont have any raid setup , just lvm for make a big virtual HD , now
> >on
> >> of the 3 HD goes damage i can start with the other 2 left and only
> >missing
> >> the data that was copy in the 3 HD area ?
> >>
> >That is only if you are very lucky. When you create a volume group or a
> >logical volume, you can specify which physical volumes it should use for
> >those, but that sort of defeats the purpose of LVM, which should handle
> >those sorts of things for you transparently.
> >
> >Without RAID, you are really relying only on luck to keep your data
> >safe. Do yourself a favor setup a RAID5 and run your LVM over that.
> >
> >Regards,
> >
> >-Roberto
> >
> >--
> >Roberto C. Sanchez
> >http://people.connexer.com/~roberto
> >http://www.connexer.com
> >
I think (as Roberto implied) the answer in general is no. Losing
one drive in a multi-drive LVM is a bit like losing a group of
cylinders in a single drive. It depends on your partitioning scheme
and the filesystem format - if you are unlucky all of your inodes could
be on the failed disk and you lose everything. Not much better
would be to have holes appearing randomly throughout your files.
Only partitions which you know were not allocated any space on
the failed drive would be safe and could be trusted.
The one good thing is that unlike the situation with a single damaged
or corrupted drive, in this case you would be able to determine fairly
unambiguously which files were intact and which were damaged.
Regards,
DigbyT
--
Digby R. S. Tarvin digbyt(at)digbyt.com
http://www.digbyt.com
Reply to: