> This sounds just perfect. Good. It's done.
> > > Or should I just convert to disklabel? > > > > I don't understand how that's relevant. > > Suppose I have these logical partitions: > > hd0s5 = some crap > hd0s6 = paging > hd0s7 = important filesystem > > and then I remove hd0s5. Now the paging area becomes hd0s5 and > the important filesystem becomes hd0s6. If I still have > paging=hd0s6 in some configuration file, I lose. Ah, I see what you mean. So don't do that then. > I was under the impression that BSD-style disklabels could avoid > this by letting me tag partitions so that their letters don't > change if I add or remove other ones. Don't they work that way, > then? Well, yes they do, and so do MSDOS (linux) partitions. It sounds like you are just using some really annoying tool to frob your partition table. In both formats, there are numbered slots and each slot specifies a physical location and size on the disk. There is no requirement that the physical spans of consecutively numbered partitions must be laid out consecutively. With tools like fdisk, you frob individual partitions by number, so no renumbering happens unless you do it yourself. With cfdisk, you add and delete partitions and it numbers them automagically to be consecutive, but it leaves unused gaps rather than renumbering when you delete a partition. I guess if you split one partition into two, it will renumber the later ones. Hmm, but I think BSD systems might check the type field in the disklabel and refuse to swap on partitions not marked for swap (which is a much more intelligent technique that doesn't eat a page on disk).
