Booting - Enterprise Volume Management System

[Thiers Botelho]

Hi Alex.

Quoting / snipping is getting even trickier . . .


On 8/14/06, Alexander Skwar <listen at alexander.skwar.name> wrote:
> >
> > On 8/12/06, Alexander Skwar <listen at alexander.skwar.name> wrote:
> >>
> >> > On 8/11/06, Alexander Skwar <listen at alexander.skwar.name> wrote:
> >> >
> >> > - A triple-boot system with WinXP, Ubuntu and, say, Fedora 5 ;
> >> >
> >> > - one [or more] VFAT logical partitions for data exchange among the 3
> >> > systems (I know, no data security at all here) ;
> >>
> >> Because of that, I'd recommend ext2. There are drivers for Windows,
> >> which allow accessing ext2 from Windows.
>
> > Therefore I see no real advantage in following your recommendation
> > (i.e., ext2 for data exchange instead of vfat). Did I miss something ?
>
> Well - at least on the Linux side, the access permissions are
> preserved ;) Granted, you lose this on the Windows side - but
> with FAT, you NEVER have access permissions.
>
> OTOH: What about NTFS? With ntfs-3g, you can access NTFS
> easily and, so they say, reliably from Linux. This way,
> you keep the access perms on the Windows side but lose them
> when accessing from Linux ;)

I had forgotten about ntfs-3g. Time to refresh old paradigms . . .

However it seems we lose some security whichever solution gets chosen.
So we can't ever have our cake and eat it too. :(

>
> > One problem I see with making the PV a primary partition would be,
> > since all primary partitions come _before_ the logical partitions,
>
> No, they don't. At least not in terms of cylinders:
>
> Disk /dev/hda: 40.0 GB, 40007762432 bytes
> 255 heads, 63 sectors/track, 4864 cylinders
> Units = cylinders of 16065 * 512 = 8225280 bytes
>
>   Device Boot      Start         End      Blocks   Id  System
> /dev/hda1            2001        2062      498015   83  Linux
> /dev/hda3               1        2000    16064968+   5  Extended
> /dev/hda5               1         123      987934+  83  Linux
>
> Partition table entries are not in disk order
>
>
> There's no requirement stating how those partitions are ordered. Those
> partitions could be accessed with no problems.
>

Now that's v-e-r-y interesting. So one possible partition order
(cylinder-wise) might be as follows (hipothetical sizes & cyl
numbers):

1 4000 hda1 primary NTFS - WinXP C: drive
4001 45000 hda4 extended
4001 6000 hda5 logical NTFS - WinXP D: drive for pagefile.sys
6001 6500 hda6 logical ext2 - /boot_ubuntu
6501 7000 hda7 logical ext2 - /boot_fedora
7001 22000 hda8 logical ext3 or ntfs or vfat - data storage & exchange
22001 41000 empty chunk for copying / moving / adding / resizing partitions
41001 43000 hda10 logical swap - /swap_ubuntu
43001 45000 hda11 logical swap - /swap_fedora
45001 150000 hda2 physical - Linuxen PV for LVM use
150000 - end-of-disk

I'd keep /swap(s) as logical partitions to remain at the conservative
side regarding suspend / suspend2 . In practice, if what is said on
suspend2's site works for real, I could have only one single smaller
swap partition, shared for both Linuxen, plus 2 RAM-sized dedicated
files under LVM - each one for suspending each system.

> Alexander Skwar

Any comments warmly welcome . . .

Cheers

Thiers