I was using gparted to restructure my partitions and I noticed that on my drive there was a 7.84 MiB on the front end and a 2.47 MiB on the back end which were unallocated. The disk was originally new and I believe I never got Windows 98 to install because of it not being able to handle the new hardware. I installed OpenSuse 11.1 and did not intentionally leave areas unallocated. It does not show in the partition manager of 11.2. Are those unallocated areas some special boot or recovery areas which I should not mess with or are they “rounding” errors and can be used?
I see the same “remnants” when using Gparted. The openSUSE partition manager gracefully does not show these. The small fragments (such as between partitions) are sometimes due to partition allocation with cylinder rounding: The next partition starts on a rounded boundary, but the prior partition was allocated in un-rounded units. If you wish, you can use Gparted (or another similar tool) to expand the partition to include the fragment. This seems to work for ext4 and NTFS. I have not tried this technique on FAT/VFAT or ext2/3 partitions.
NOTE: The ~2.47 Mib at the end of the volume seems to be related to physical volume management (and used by NTFS as well). The one time I grabbed it with Gparted, the MBR and GRUB were clobbered, requiring recovery with Parted Magic.
Suggestion: leave the small end-of volume remnant alone.
The one time I grabbed it with Gparted, the MBR and GRUB were clobbered, requiring recovery with Parted Magic.
That’s scary! I’ll be sure and leave that alone. Anyone else experience that?
Another question I notice with gparted is with extended partitions. In restructuring, I was wanting 3 primary partitions and the rest logical. My current system already has 3 primary, but I mistakenly blocked those out of my mind as I was looking at the future goal when I eventually combined those into one or two. I had deleted some partitions near the beginning of the extended one and so then resized the beginning of extended unallocated past it. That is, I moved the extended beginning to the right. So now I had the unallocated one in the primary area where I eventually wanted it. I went to install and noticed that it was giving a sdb6 choice. I expected sdb3. I stopped and then realized I already had 3 partitions.
It looked to me that the install was going to put the unallocated partition in the primary area as a logical partition part of the extended partition. Gparted said I already had 3 primaries but the install acted like it was no problem. I went back and moved the beginning of the extended partition back where it was and will keep that plan for the future. What would have happened if I went and installed it anyway? That is, sdb6 is ordered after my sdb5 home but physically is before? Would that be confusing to the system if not just me?
By the way, is there a way to get the partitions to be sda instead of sdb?
I have my main drive as a SATA drive (I think it is) and then after original install, I added an old IDE drive for testing. After this install of 11.3 it has named them sdb instead of sda. Is there a way to switch them and if so, where would I do it and what things would I need to worry about? I am mounting by label.
IDE+SATA are a problem because the BIOS may assign positions arbitrarily and capriciously. I’d avoid mixing. It is a BIOS problem.
udev loads the kernel modules asynchronously to save half a second at boot time. As a consequence hard disks get named ‘randomly’ in the order they are found. This becomes particularly obvious when you mix IDE and SATA. What you should worry about: Never trust device names! Don’t be surprised if you install on sda and it becomes sdb after rebooting. If you mount by label, it’s fine. Pay attention not to use device names in /boot/grub/menu.lst either. Mount your swap partition by UUID. Use blkid to see the UUIDs. They don’t change (unless you reformat the partition). Generally speaking, mixing IDE and SATA is not fun.