RAID & HDD failure recovery

David G. Miller dave at davenjudy.org
Sat Nov 18 00:19:57 UTC 2006


Laurence Vanek <lvanek at charter.net> wrote:

> David -
>
> My experience is identical to yours from steps 1) --> 3). At step 4) my 
> experience diverged.
>
> My step 4) went like this:
>
> [4] new drive arrives, I install & attempt to boot, like I had several 
> times while waiting for new hdd when failed drive (hda) was powered down.
>
> [4a] Boot hangs, cant find valid partition(s) (i.e. its only looking at 
> newly installed drive). For some reason good drive I had been running on 
> isnt seen.
> ...
>   
Unfortunately, its been long enough ago that I don't remember the exact 
steps for introducing the replacement drive "back" into the array.  I 
think I had to boot in rescue mode and partition the drive so that it 
already had the correct partitions. 

The other variable is I still run lilo on my server.  lilo quite simply 
just works with Linux software RAID while grub can only barely be 
coerced to work with it the last time I looked into the problem.  There 
are lots of arguments as to why grub is superior to lilo for a 
development machine (e.g., all of the grub command line stuff you can 
do) but these don't apply to a server environment.  The way I look at 
it, if a server is so screwed up that the only way you might be able to 
recover is to go into the grub command line and diddle with things then 
you're hosed already.

BTW, to anyone reaching for their flamethrower, I don't want to start a 
grub vs. lilo flame war so please don't.  If you have a *foolproof* and 
*native* (doesn't rely on external commands) way to get grub to support 
Linux software RAID, I'm more than willing to listen.  Hint, anything 
that includes an instruction like "use dd to copy the MBR..." is a 
non-starter.

Cheers,
Dave

-- 
Politics, n. Strife of interests masquerading as a contest of principles.
-- Ambrose Bierce




More information about the fedora-list mailing list