Hello, On Sun, 10 Apr 2011, Anton Aylward wrote:
David C. Rankin said the following on 04/09/2011 04:33 PM:
KISS - is the best philosophy. Use only the number that you need.
And that's the wonderful thing about LVM.
KISS is exactly why I don't use LVM. $ grep 'sd.[0-9]$' /proc/partitions 8 17 31457280 sdb1 8 18 1048576 sdb2 8 19 31457280 sdb3 8 20 1 sdb4 8 21 1889550304 sdb5 8 1 5253223 sda1 8 2 1060290 sda2 8 3 31463302 sda3 8 4 1 sda4 8 5 5253223 sda5 8 6 31463271 sda6 8 7 413890596 sda7 8 49 1953514552 sdd1 8 65 1953513560 sde1 8 81 1953514552 sdf1 8 97 1044188 sdg1 8 98 10490448 sdg2 8 99 1941977362 sdg3 8 113 1953514552 sdh1 8 33 1465136001 sdc1 8 129 488384001 sdi1 How do you sanely handle this? When disks die? Oh, and there's also another 6 disks (or more) imported via NFS when needed, and 3 "good" external (and a couple "scrap"[3] ones). I mostly use a tree of symlinks (including the NFS-mounts when available) basically starting in ~. Let's say I'd use LVM, what if e.g. /dev/sde shows defects (seen in the SMART data). Can LVM show me what dirs and files(!) I have on the PV(s) on that probably soon-failing drive?
If I need to compile a package I can create /usr/src and give that space back when I'm finished then next week use that same space as /data to do some analytics ...or to take a snapshot of other partitions in turn for backup.
For stuff like that, I've got a bunch of GB of scrap-space (e.g. in /data/~user or /home2/~user/[4]) where I can create a dir, if needed symlink or 'mount --bind' it to anywhere I want, use it and then remove the whole stuff again. For small stuff, /dev/shm/~user suffices. No need to add a level of complex software (AHS, ASS![1] and Murphy![2]) that can go wrong. -dnh, hand-selected, translated sig for this occasion [1] All Hardware Sucks, All Software Sucks, [2] anything that can go wrong, will. At the most inconvenient time. Wreaking as much havoc as possible [e.g. taking the most important data with it]. [3] i.e. ones that have a few defective sectors, but still work and don't rapidly get more defects [4] containing big, but less important stuff, think thumbnails, savegames giggle-mudball cache and suchlike. -- RAID: One more disk fails than can be recovered by the redundancy. -- Andreas Dau -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org