Gparted mystery

Greetings All,

For a while now, I have been using an older version of gparted (0.4.5) from an older bootable
Linux disk to format my hard drives. The version included with Suse 11.4 (0.8.0) has given me a puzzle.

I tend to create a number of partitions (3 primary, an extended, and a number of logical) on
my disks. For some reason, version 0.8.0 seems to requires 1 mb of space between each partition, and 2-3mb of space at the end of the drive. With the older gparted, I could create partitions with no unused disk allocation.

Is there some reason for this new behavior? Is there some way to format a drive with the newer gparted without unused space? I realize that 10-15mb of disk is fairly small, but I have this dislike of wasted space. The drives being formatted are SATA drives in the range of 250gb -750gb.

Thanx in advance

Rich

Was the partitioning created with the old gparted, or the new?

I don’t think I have tried the newest gparted, but I have seen similar behavior with “fdisk”.

The older partitioning tools use the cylinder/head/sector way of identifying partitions. The newer partitioning tools just use relative sector number.

If a logical partition was created in the cylinder/head/sector model, then the partition table for that partition is at the start of a cylinder, and the entire cylinder is consider to be unavailable. When you list that with the newer partitioning tools, it shows only the one sector for the logical partition as used, and the remainder of that space (what is left of the old cylinder) as free unallocated disk space.

What you are seeing might be an artifact of changing standards.

On 2011-04-27 20:36, richardrosa wrote:

> I tend to create a number of partitions (3 primary, an extended, and
> a number of logical) on
> my disks. For some reason, version 0.8.0 seems to requires 1 mb of
> space between each partition, and 2-3mb of space at the end of the
> drive.

I think that it is the alignment on megabyte boundaries. I have read this
mentioned elsewhere.


Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 “Emerald” at Telcontar)

Its because we know a side by side WinVistaSeven may eat your data, if these alignments are not followed. You can format a new usb Stick under WinSeven and will loose all not fitting partitions silently erased in the background.

Ok, So it seems the wasted space is due to Windows 7 requirements.
Since I don’t run Windows 7 (nor do I intend to) is there some way to bypass this? I’ve tried selecting the various alignment options (Cyl/None) but gparted is quite insistent about the 1mb space.

It strikes me as being a bit strange that a Linux tool is forcing options to assist Windows…

Should I hang on tight to my older gparted copy?

Rich

fdisk

Alignment is not only about Win7 but also flash based harddisks.

There are also alignment constraints for performance on 4kB sector disks.

BTW, any reported CHS is a fiction since disks don’t have fixed number of sectors per cylinder these days. So there is no advantage to trying to align to a “cylinder”. It’s best to think of a disk as a large array of blocks.

On Wed, 27 Apr 2011 21:36:02 GMT, richardrosa
<richardrosa@no-mx.forums.opensuse.org> wrote:

>
>Ok, So it seems the wasted space is due to Windows 7 requirements.
>Since I don’t run Windows 7 (nor do I intend to) is there some way to
>bypass this? I’ve tried selecting the various alignment options
>(Cyl/None) but gparted is quite insistent about the 1mb space.
>
>It strikes me as being a bit strange that a Linux tool is forcing
>options to assist Windows…
>
>Should I hang on tight to my older gparted copy?
>
>
>
>Rich

This is driven in part by an old DOS/MSWin hangover from CHS days.
Partitions done by MS tools have to be on cylinder boundaries, about 8MB
being a full real cylinder at the outer edge in modern high capacity
drives (around 4096 sectors by 4 heads). MS never has learned how to
correctly deal with bad hangovers from early decisions.

I recommend hanging on to the older copy. It seems to be more compatible
with MS foolishness. I have had issues with partition tables created by
new tools not being recognized by old hardware.