Thomas Backlund wrote:
From: "John Danielson, II" <[EMAIL PROTECTED]>OK, point made, because I did that(I had thought the extended part table was a special case).
theAny system that has EVER had a file system with Windows or DOS on it has
|following part structure.
Part #s 1-4 can be primary.
Part #5 is always an extended part table to hold logical drives.
Parts 6 and up can be logicals.
| And Windows allows for 4 primaries, not 3. So a mixed system or a
| migrators system will have a mess.
|
You have misread the partition tables, or how it works...
Windows/dos (or actually the partitioning scheme) does allow
for 4 primaries, BUT the extended partition counts as one primary.
You dot believe me? If so, it's easy to test ...
1. Create 4 primary partitions. (make sure you leave some empty space
on the disk, for example 4 x 1GB partitions on a 10 GB disk leaves
you with 6 GB of free space on the disk...)
2. Now try to make a extended partition... IT WON'T WORK
3. Remove one of the primary partitions. Now you can create an
extended partition that uses the free space on the disk,
that also allows you to make logical partitions...
Thomas
However, some of what I was trying to say is that what linux sees as how Windows numbers things is that the extended table is always one up from that last created Primary, also. therfore, what happens when a windows tool is used to make Ext3 or Ext2 is that:
If primary 1 and 2 exist, then extended part table is 3, first logical is 5, etc.
If primary 1 exists, then extended part table is 2 and the logicals start at 5.
diskdrake numbers it 4-- so long as only and forever Diskdrake works on parts, this is fine, but when it errors it does need something else to work on it to fix or the /etc/fstab manually revised after an e2fsck failure at boot.
If a Windows or DOS based tool, including Ranish, sees a single primary, an extended numbered 4 and a logical numbered 5 it will says the chain is invalid. most GUI'd Windows tools will then invalidate the drive insofar as access is concerned.
Linux Logical parts can live within a Win95 Ext'd LBA defined if they themselves each bound on CHS boundaries. User cannot calc those boundaries, new current HD have variable sectors per track-- either diskdrake talks to drive controller or it needs to use nearest pure cylinder bound less than user assigned space (calcing with avg. sectors\track will give a CHS bound, good enough). What does w newbie do when asking fro size of part??? 8,000,000 Kbytes is roughly 8Gig-- so he says 8,000 MB. Asks for that. PM 8.0 lets you drag, and shows you valid CHS boundary increments. Nice-- if use then part numbers change versus what /etc/fstab was written to reflect, things get mashed AGAIN until manual edit performed.
Windows based tools that are backward compatible do this, and I also get different sized parts out of what diskdrake tells me while the windows tools tell exactly what the bound is by changing the size dynamically to show the revised size bounded to CHS. Newbie favors this info level, WILL use these tools if was a Windows power user tired of paying for Windows junk that crashes and has security holes that Linux does not (prime business motivator, and where Mandrake tends to get support funds, and where I KNOW RedHat gets support and training income).
Easiest fix is to unhook the mount link by commenting it, then reboot and try to fix. Diskdrake erred, so newbie is GONNA try something else. Newbie likes GUIs-- is he gonna go for a CLI based thing that has no visual feedback and little validate before write?? No, he is not-- is he gonna pay for something obviously broken??? Half won't-- our market increase potential for Linux Mandrake as a whole just halved itself. I gotta sell Windows users on this thing, and if must say here is how you do this with windows tools to handle a diskdrake prob, they are not gonna use diskdrake, tell friends they can't cuz its broke, and the FUD starts, and Mandrake's rep suffers. The core user tools HAVE to be clear, clearly doced, with fallback recovery procedures in that doc-- procedures tuned for EASY.
Two major areas picking up Linux use in my area-- Small businesses and end users, and both need things they can do for themselves, not things having to pay someone for tech support. I have a school district interested in the LTSP, but if a Parapro cannot maintain a HD with GUI'd Linux tools he is gonna use GUI'D windows tools that the district has on hand. Resulting conflicts between Mandrake's tools and other tools that can make valid Ext3 parts that linux can use will result in increased resistance to Linux rather than a ground swell, and where schools go there go our kids here for the most part -- another resistance to overcome.
John.
