I used ebuilds provided at bugs.gentoo.org versioned device-mapper-0.96 and lvm2-user-1.95. Also were provided two patches for /etc/init.d/[checkfs|halt.sh] and they whole bunch worked fine. Recently these ebuilds were replaced by sys-libs/device-mapper 1.00.05 and sys-apps/lvm2 2.00.06. I installed these but I have to mount at boot time by hand. The patches (provided fir 1.95 and 0.96) are still applied, so something other must have changed. I figured out, that I can mount partitions after issuing lvchange -a y /dev/VG/vgXX. I have to type it TWO times! First devfsd is complaining about "No such module /dev/Nikki_Tyler/l_mucke0 (I _think_ it is devfsd...) second time it is quite and only then I can mount /usr, /home and so on... I don't know exactly yet where the error lies. May be issuing this bugreport helps to find out of init scripts have to be adapted or changed or so... I also do not know exactly ATM if there is a easier or other way than do an lvchange -a y /dev/VG/LV PS.: I am running 2.6.0-test4-mm1 and devfsd-1.3.25-r4 Konsti Reproducible: Always Steps to Reproduce: 1. Install beta dev-mapper and lvm and 2.6.0-test4-mm1 -> runs fine 2. Install lvm2-2.00.06.ebuild and device-mapper-1.00.05.ebuild 3. boot Actual Results: lvm2 volumes was not initialized properly, they must be lvchanged and mounted by hand now. Expected Results: initialize, mount, boot Portage 2.0.49-r3 (default-x86-1.4, gcc-3.3.1, glibc-2.3.2-r1,2.3.2_pre1-r0, 2.6.0-test4-mm1) ================================================================= System uname: 2.6.0-test4-mm1 i686 Intel(R) Pentium(R) 4 CPU 2.40GHz ACCEPT_KEYWORDS="x86 ~x86" AUTOCLEAN="yes" CFLAGS="-march=pentium4 -O3 -pipe -fomit-frame-pointer -ffast-math -fexpensive-optimizations -funroll-loops -frerun-cse-after-loop -frerun-loop-opt -fforce-addr -falign-functions=4 -msse2 -mmmx -mfpmath=sse" CHOST="i686-pc-linux-gnu" COMPILER="gcc3" CONFIG_PROTECT="/etc /var/qmail/control /usr/kde/2/share/config /usr/kde/3/share/config /usr/X11R6/lib/X11/xkb /usr/share/texmf/tex/generic/config/ /usr/share/texmf/tex/platex/config/ /usr/share/config" CONFIG_PROTECT_MASK="/etc/wget /etc/rc.d /etc/make.conf /etc/hosts /etc/conf.d/net /etc/profile /etc/fstab /etc/lilo.conf /etc/rc.conf /etc/gconf /etc/env.d" CXXFLAGS="-march=pentium4 -O3 -pipe -fomit-frame-pointer -ffast-math -fexpensive-optimizations -funroll-loops -frerun-cse-after-loop -frerun-loop-opt -fforce-addr -falign-functions=4 -msse2 -mmmx -mfpmath=sse" DISTDIR="/usr/portage/distfiles" FEATURES="sandbox ccache autoaddcvs" GENTOO_MIRRORS="http://gentoo.oregonstate.edu http://distro.ibiblio.org/pub/Linux/distributions/gentoo" MAKEOPTS="-j2" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/export/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/export/portage" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="x86 oss apm crypt cups gif libg++ mad ncurses nls pdflib spell gdbm berkdb readline X gpm pam libwww python cdr 3dnow aalib alsa -arts -bonobo avi dvd encode -esd -gtk -gnome2 gtk2 -gnome imlib jpeg -kde maildir oggvorbis mpeg mmx mikmod nas -mysql opengl perl png -qt -qtmt -ruby sdl -svga xml2 xv zlib -tcltk sse ssl -slang -quicktime tetex -java -threads dga -doc -nptl -lirc imap gimp-print xinerama truetype -xfs -mozcalendar -mozaccess -mozinterfaceinfo -mozp3p -mozxmlterm moznoirc moznomail moznocompose -motif tiff -xmms cjk -foomaticdb -tcpd"
While updating from linux-2.6.0-test5-mm1 to linux-2.6.0-test5-mm1 I checked the "ioctl interface version 4" option, now the bootup process initializes the lvm2 volumes correctly and mounts them now :) Has the, now checked, option something to do with the issue? I though the version of the ioctl interfaces is regardeless because device-mapper is always compiled against the actual header files... It is not adapted (recompiled against new kernel tree) now btw... Konsti
Yes, device-mapper-1.00.xx and LVM2-2.00.xx expect ioctl-4 support in the kernel. I'll mark this bug invalid though -- all newer 2.6 kernels and the latest patches from sistina for 2.4 have ioctl-4 version of device-mapper.
AAAaah! OK, I did not know that device mapper-1 and lvm2 _expect_ ioctl version 4, I thought it is distinguished at compile time. Ok, so that was my fault, sorry for the inconvenience! Regards, Konsti