postid wrote:
I'm tempted to reinstall, this time making all four partitions primary partitions.These logical partitions seem illogical to me. Have I donesomething wrong or is it just that way with logical partitions? What are the advantages/disadvantages of logical and primary partitions?. I've read about them on the web, but it's still as clear as mud to me.
The advantage of using logical partitions is that you can have more than four. If you will never need more than four, you can stick with primaries. Windows *must* have a primary partition as its boot drive, even if most of it is installed on a logical one. Early Windows versions needed the boot partition to be the first primary partition they could recognise.
Apart from the Windows issues, there's no real difference. 'Logical' partitions simply have partition table entries in an 'overflow' area away from the space used by IBM's original four entries ('nobody will ever need more than four partitions...'). The actual partitions on the physical disc are indistinguishable as regards 'logical' and 'primary', and primary partitions (numbered one to four) may even be placed after the extended partition which contains the logical ones (numbered five upwards). Generally the extended partition can be totally ignored, as it is never referred to by anything. Only a slightly brain-dead utility will automatically generate an icon or link to it. Yes, I've seen it done. You would use fdisk or one of its successors such as cfdisk to see the physical details of the partitions.
As Lisi has pointed out, the *nix operating systems use a single filesystem, regardless of how many drives, partitions or network shares it may be actually composed of. The /etc/fstab file shows the operating system how to mount the various drives and shares, i.e. at which branch of the filesystem to attach them. The mount command, issued without parameters, will show the current state of play and can be used to manually mount and dismount drives and partitions. The *nix filesystem is quite structured compared to that of Windows, which puts log files in a subdirectory of the system DLL library.
The advantage of this method is that if a drive is filling up, you can fit and install an additional drive. You can then take one or more of the sub-branches of the part of the filesystem located on the nearly-full drive, move their contents to new empty partitions, then adjust /etc/fstab and dismount and remount the entries (or reboot the machine) and from now on those sub-branches will live on the new drive, freeing space in the branch from which they were moved (just copying them will not empty the space). This is a bit trickier if the 'system' parts of the filesystem need to be moved, and is probably best done offline.
If you've ever moved data to a new drive under Windows, you will know that the utility you use must locate all references (shortcuts, registry entries etc.) to the moved files and change the drive letter associated with every one of them. This is impossible when references are embedded within other files. Under *nix, the moved files still have exactly the same pathname, even if they now live on a share on a different computer. What used to be even worse under Windows was that adding a new drive containing one or more primary partitions would alter the existing drive letter assignments above C:\. Great fun, and at least that ended with XP. Individual drive and partition designations which are visible to the user are a legacy of DOS which Windows, nearly twenty years on, still hasn't left behind.
-- Joe