[luau] NFS ate my raid superblock

[hattenator at imapmail.org]

I have two questions.  One is my situation, and one is a generalized
question relating to it.  First, I am in a bad situation here.  I have
been using all the new beta versions of packages updated off of cooker
for some time now.  I was always amazed by the fact that even though
these were all barely tested packages, they actually worked quite well on
my system together.  I'm not running a server or anything, so I was just
grabbing these new packages for fun to test them out.  I was amazed at
how well Mandrake was working, and I never had one problem (I had plenty
of little bugs using stable RedHat).  But one day, when setting up a
boot-net server, I guess all my months of stability caught up with me and
I ran into quite a problem.

I was working on making my system into a net-boot device for my old sun. 
So I tried to recall how to do this, referred to some docs, and got to
work.  I installed bootparamd, rarpd, tftpd, and nfs.  It took me a few
hours, but I got all the boot stuff working all the way to the point that
I could watch the debian splash screen and a few kernel messages on my
sun.  Now it was time for the NFS.  I created an exports file and started
up nfs.  However, the service locked, and I had to break it.  I figured
out something was wrong (I no longer remember what), fixed it, and tried
it again.  Now I got nfs debug errors and big FAILED messages for the
service.  But I felt one step closer.  I tried again, and it hardlocked
the system with a kernel oops.  I rebooted and tried again.  I
uninstalled nfs, reinstalled, and tried one more time.  Still another
hardlock.  So I booted again, and got a message to the effect of "invalid
superblock, please use 'root='".  So this kind of sucked.  I tried
booting off an old mdk 8.2 cd I had lying around, but it couldn't
auto-load my software raid-0 array in order to fsck it.  So I have two
questions:

1.  Can I easily fix my problem by booting a resue CD/disk, creating a
propper /etc/raidtab, then running mkraid and fsck?
2.  How does the system know what my raid array is.  For instance, the
kernel can boot my raid array without reading /etc/raidtab, because /etc
is on the array.  Certain modern distro CD's can figure out where my /
partition is, load up the raid array, and read it no problem, without
needing /etc.  But I suppose they can't do this without a valid
superblock.  Can someone explain this process to me?  Also, ext3 creates
redundant superblocks throughout the partition, right?  

-Eric Hattemer
-- 
  
  hattenator at imapmail.org

--
http://fastmail.fm - Same, same, but different...