FC2 to FC3 upgrade, e2fsck cannot continue

[Mark Murphy]

I upgraded a file server from FC2 to FC3 using the FC3 install CDs. After 
the upgrade, the server will no longer boot without administrative 
intervention, and nothing in yum update alleviated the problem.

The system has a two-disk software RAID1 array, /dev/md0. On normal boot, 
shortly after the interactive-boot message, I get: "/dev/md0 is mounted. 
e2fsck: Cannot continue, aborting", followed by the recovery console. If I 
do a shutdown -fr now at the console, it reboots and starts up fine.

I have:
-- Run fsck /dev/md0 from the recovery console, which came back clean, but 
then problem then returned after reboot
-- Modified /etc/fstab to put 0 in the last column, trying to avoid whatever 
e2fsck invocation is going on

The relevant line from /etc/fstab is:

/dev/md0 / ext3 rw,acl 1 0

The relevant line from /etc/mtab (which I have not modified) is:

/dev/md0 / ext3 rw,acl 0 0

I am booting with the following GRUB profile:

title Fedora Core (2.6.11-1.14_FC3)
root (hd0,1)
kernel /vmlinuz-2.6.11-1.14_FC3 ro root=/dev/md0
initrd /initrd-2.6.11-1.14_FC3.img

It feels like something is mounting /dev/md0 read-write before that fsck 
call happens, but I'm unclear as to why the fsck is being called there, and 
I don't grok rc.sysinit enough to understand what's going on.

Any suggestions? Thanks!

-- 
Mark Murphy  |  Information Systems Specialist
Hanover Engineering Associates
mmurphy at hanovereng.com | (610) 691-5644
cell: (610) 349-5557