From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1) Description of problem: I set up a logical volume, tested it by mounting it with the mount command and doing a df, so no problems there. The problem is that the test at line 310: if [ -e /proc/lvm -a -x /sbin/vgchange -a -e /etc/lvmtab ] fails, leading to the system not booting the rest of the way and going into a failsafe prompt environment, even though those files exist. When I commented out that line (and the fi command at line 312) everything worked fine. The problem is with the test. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1.create an logical volume 2.add it to /etc/fstab 3.reboot Actual Results: system went into a maintenance shell Expected Results: system should have booted happily Additional info:
You probably need to re-run "mkinitrd". Kernel modules needed at boot time get loaded along with the initial kernel image, and the installer sets all that up so that the initrd includes the lvm.o module if needed. Please reopen the bug if that doesn't get things going.
The lvm problem--really it is not a problem with lvm itself, which works fine, but I was not able to determine where else to report this--is with the rc.sysinit file. When I start lvm manually--vgscan, vgchange -a y, and so on-- it works fine, no module errors or anything. The fix is easy: remove the -e /proc/lvm part of the test on line 310 of rc.sysinit. After that is done, all works fine. This leads me to believe that the proc filesystem is not testable at this stage of loading, thus the error. This error also occurs if lvm is set up during the installation.
The rules in Red Hat Linux are simple --- the initscripts only set up the LVM volumes if the kernel has been set up with LVM configured. That's NOT a bug, it's simply the design of the initscripts. And you *need* to rerun mkinitrd AND (unless you're using grub) lilo if you enable LVM after the initial setup. Otherwise, the initial kernel boot won't load with the LVM modules, and the init scripts won't see /proc/lvm present, and you won't get the volumes initialised. I have many systems here using LVM perfectly happily using this. The existing 8.0 rc.sysinit works with this scheme just fine --- IF you enable LVM in your kernel boot (via the initrd). If you have a reproducible case where setting up LVM in the initial install doesn't result in LVM being enabled at run time, then please file that bug against the installer ("anaconda").
*** Bug 79018 has been marked as a duplicate of this bug. ***
Incidentally, this is NOT a problem (not having the LVM kernel moduled loaded in the initial ram disk) in Red Hat Linux 9. Red Hat Linux 9's rc.sysinit file will load the LVM kernel module if the initial ram disk didn't. Here is the relevant portion. # LVM initialization if [ -f /etc/lvmtab -a ! -e /proc/lvm ] ; then modprobe lvm-mod >/dev/null 2>&1 fi if [ -e /proc/lvm -a -x /sbin/vgchange -a -f /etc/lvmtab ]; then action $"Setting up Logical Volume Management:" /sbin/vgscan && /sbin/vgchange -a y fi