Thanks for the info. Its information like this that helps
me understand how to go about debuggin.
I hadn't changed the type to "fd" which was the real problem.
But with the info you've given me, I'm going to do a little
poking around and see what I can learn from it. I've been
listening to you guys for a couple of years, and ran raid0 on a
news machine with great success, but I'm rather new to raid5.
I've made a couple of dumb mistakes like this one.
Everything is pretty clean now and I'm feeling confident with
the setup. I'm about ready to put the box into production.
The only other mystery right now is why I can't raidhotremove
a disk. It says it is busy (it is). I was expecting
raidhotremove to be able to yank it out anyway.
Stephen
On Sun, 28 Mar 1999, Piete Brooks wrote:
> > My problem is that whenever I boot now, the array comes
> > up without sda2. I can raidhotadd sda2 and things are happy
> > until the next reboot.
>
> Hmm -- sounds to me like "raidhotadd" isn't re-writing the SB -- sorry, I don't
> know how that code works :-(
>
> What does a "mkraid --debug" give you in /var/log/messages ?
>
> If you can grab a copy of the on-disk SB and send me an "od -t x4" (or an
> encoded binary) I can unpick it and see what is *actually* there ...
> (failing that I can send you a copy of my SB debugging script ...)
>
> > How can I make sda2 act like the other disks?
>
> ========================================================
> *** IF THERE IS REAL DATA ON IT, DO NOT TRY THIS !!! ***
> ========================================================
>
> (READ NO FURTHER UNLESS YOU ARE A HACKER !!)
>
> My very uneducated guess (I've only recently started using RAID) is that the
> following might work:
>
> raidhotadd it, and wait til it's in sync.
> raidstop the device.
> Now it is known that all data on disk is "correct".
> Whatever "resync" subsequently happens, the result should be correct.
> Update raidtab to have it as a raid-disk
> Run "mkraid -R /dev/md0" (if you are *really* sure you want to).
> Contrary to my interpretaion of what I have read, my current
> belief is that this will ZAP the RAID SB, but not the data.
>
> With the new SB in place, it should all work ...
>
Stephen Denny mailto:[EMAIL PROTECTED]
Hex.Net Superhighway http://www.hex.net
