On 08/19/2009 05:15 PM, Bastian Blank wrote: > On Wed, Aug 19, 2009 at 04:50:09PM +0200, Marc Haber wrote: >> Since the last update, my crypto devices are recognized by the system as >> /dev/dm-xx, for example in fsck and in df's output: >> /dev/dm-15 12G 11G 565M 96% /mnt/home >> /dev/dm-13 7,9G 7,0G 532M 94% /mnt/usr >> /dev/dm-14 3,0G 1,4G 1,5G 50% /mnt/var >> >> lrwxrwxrwx 1 root root 8 19. Aug 09:06 home -> ../dm-15 >> lrwxrwxrwx 1 root root 8 19. Aug 09:06 usr -> ../dm-13 >> lrwxrwxrwx 1 root root 8 19. Aug 09:06 var -> ../dm-14 >> >> I'd like them to show up as /dev/mapper/home, /dev/mapper/usr and >> /dev/mapper/var again. > > This means that the real devices needs to be named this way as mount > always dereferences symlinks. > > Patch which does this against the Debian package is attached. > > Alasdair: Does Red Hat solve that problem somehow? > > Bastian >
Just to explain why we decided to do it this way in LVM upstream: The (very) early version of the rules were discussed with udev team so it would be correct from udev point of view as well. The discussion led us to a solution where we had to use /dev/dm-* as nodes and /dev/mapper/* as symlinks, citing Kay Severs: "In general we do not want any unneeded disconnect from kernel names and /dev names, and dm block devices should stay as /dev/dm-* device nodes." "Please do not rename kernel devices, they should match the kernel names. Only create SYMLINK+= to the kernel names..." (you can find the whole discussion at http://markmail.org/message/bj4zkjo2peeocnhq) We're just trying to comply with those udev requirements... Yes, there are some problems associated with the utilities using /dev/mapper nodes, but, as Alasdair says, this should be corrected there directly. We have no other way, either we break 'udev laws' or we break a few utilities. But we would like to do this correctly, so I'm voting for the update of those utilities (the grub2 problem mentioned somewhere here in this bz -- I'm already preparing a proposal for the grub team to deal with this issue). It's quite painful now, I know, but once done correctly and having all depending things fixed, we will have a proper solution that everybody will be happy with. Peter -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org