Linux Software Raid 1 Recovery Question

[kentborg at borg.org (Kent Borg)]

On Wed, Oct 02, 2002 at 02:49:10PM -0400, Christoph Doerbeck A242369 wrote:
> 
> I can't remember everything... but when we played EIDE RAID last March
> at BLU we had several problems.
> 
> 	+ If you loose a disk, the system will likely hang (but at least
> 	your data is reasonably protected)  I never did determine if this
> 	was because we had DMA enabled, but...
> 
> 	+ If you configure the disks on the 2 independent EIDE busses, 
> 	then each disk can be a master.  Then use the cheap hot-swap
> 	trays. If the master fails, simply swap the disks and you're done.
> 	Just make sure lilo is installed on the mirror too.

I've already done better than that.

  + I did something to my CD burner at /dev/hdb and it somehow hosed a
  DMA that was in progress on /dev/hda, but the RH 7.2 system kept
  running smoothly on /dev/hdc--in fact the only way I knew there had
  been a problem is the cron job I have that every night checks for
  changes in /proc/mdstat e-mailed me the news.  This is cool!

  + On a Mandrake 9.0 install I did a software raid 1 through the
  installer, manually made a lilo.conf.hdc, changed one byte from
  lilo.conf (an "a" to a "c"), ran lilo on it, and that machine was
  perfectly happy to boot off of a half array /dev/hdc.  This is cool!

What I have not figured out is how to get a RH 7.3 (and presumably
7.2) to boot off of a half-array /dev/hdc.  This is something I need
to figure out...and I think I need to put RH 7.2 on my test bed
machine to do so.


-kb, the Kent who doesn't want to bring down his currently one-eyed
7.2 server without significant confidence he can bring it back up.