On 4/9/02 at 8:06 AM, Charles Steinkuehler <[EMAIL PROTECTED]> wrote:
> > > > Packages will be backed up to whatever disk is in > > > > the drive - make sure you put the appropriate disk > > > > in the boot drive before backing up. > > > > > I have a small request that the backup scripts write > > > to the drive from which the package was loaded. Would > > > that be a major rewrite? > > > > Not a major rewrite, but a major project nonetheless. > > > > You have to add the following capabilities to the system: > > > > * Tracking where files came from - including storing > > data, additions, deletions, and everything else - a new > > database really. MySQL anyone? > > How about a flat text file per-package? Simple but inefficient - but in any case, would occupy a LOT of space - especially if you load 200 packages (as I've been known to do :) More things to consider: these files would have to be excluded from package creation (including root.lrp in old systems). Removing packages would have to remove these files too. > > * Unique identifiers for disks - including checking for > > the right one and error handling for the case when it > > isn't. > > > > This would be a big project to get right, requires a > > database with full database accessability, and > > identifiers for disks that are guaranteed to be right. > > What's more, what if you can't back the package up to > > the right disk but want to back up to a new disk - more > > functions. > > I've got this functionality in my backup scripts for > Dachstein already (and it was a fairly major re-write). > The system "remembers" where the package was loaded from, > and defaults to backing up to the same location. You can > also manually change the backup destination, if desired. However, you didn't account for ALL of the possibilities: 1. User backs up a package - to the right medium (/dev/fd0u1440 for instance), but the WRONG disk (oops). Then what? 2. User backs up a package - to the right medium, but a NEW (different) disk. Then what? To do this right, I'd think you'd need an identifier for each and every disk, and a routine to refuse writing to a disk that didn't match - as well as the ability to write to a NEW disk. Another thing: Define The Problem. I don't see backing up to this disk or that a problem. What Problem does all this extra code solve? -- David Douthitt UNIX Systems Administrator HP-UX, Unixware, Linux [EMAIL PROTECTED] _______________________________________________ Leaf-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/leaf-user
