Jarry wrote: > But even if it is so, if you resize partition by lvm, this advantage > could be lost. And if it even is possible to keep some partition > continuous, than resizing partition in lvm would be very long process: > if I resize 1st partition (the fastest, on the most outer cylinders) > and want to keep it continuous, lvm would have to move all other > partitions...
But LVM is so useful, that even THAT would be possible with *NO* downtime AT ALL! This is possible, if you've got multiple "physical volumes". In text books, a pv is a complete harddrive (eg. /dev/sda). But that's not necessary. Instead, you could also use a partition (/dev/sda1) and there's also nothing stopping one from having multiple PVs on one drive. Now, if there are multiple PVs in one VG, it's easy to do a "pvmove", which will move logical volumes to another phyisical volume. And all that's /possible/ while the filesystem is still in use! Granted, I'd not do this at prime time... :) But how do you do that with the legacy style of partitioning? And also, how do you *control* exactly which data is at the beginning (or wherever) of a drive, if you're going to have only one grossly oversized partition on a drive? Alexander Skwar -- As famous as the unknown soldier. Ö -- gentoo-user@gentoo.org mailing list