rootfs fsck check - too early?

[Adam Gołębiowski]

On Fri, Apr 13, 2007 at 10:35:04AM +0200, Jacek Konieczny wrote:
> On Thu, Apr 12, 2007 at 07:08:23PM +0200, Adam Gołębiowski wrote:
> > Isn't it too early? Akcyza - with rootfs on lvm built on top of soft
> > raid5 - was down for a few hours today after power failure, and the
> > bootup process stoped on the abovementioned fsck due to reiserfsck
> > complaining about missing /dev/sys/rootfs device. 
> > 
> > We are assembling soft raid a little bit later (line 550). Shouldn't we
> > move fsck there? 
> 
> I guess root filesystem RAID was already assembled (rc-scripts won't run
> otherwise). It was LVM scan not done and LVM device nodes not created. 

Exactly.

> I had the same problem on my machine. I "fixed" that by using "/dev/root"
> (or something like that) in /etc/fstab. But that broken initrd
> generation (so the machine didn't boot after kernel upgrade).

Hmm... and what can go wrong by moving LVM initialization before the
problematic fsck?

Right now I am using 'fastboot' as an ugly hack, but would love to see
general solution...

-- 
 http://www.mysza.eu.org/ | Everybody needs someone sure, someone true,
   PLD Linux developer    | Everybody needs some solid rock, I know I do.