I have installed a root raid1 following the wonderful HOWTO from Jakob,
and it is so much easier now with the autodetect code, than one year ago.
Almost everything is perfect now, except...
I am running into a bizarre problem getting the mirror to boot
when the first disk has 'failed' (i.e. been unplugged with the power off,
and the termination reconfigured yadda yadda).
I got the lilo-21 source rpm from redhat 6, that has the lilo.raid1 patch
to allow booting off of any physical disk in the array, and built it on my
RedHat 5.2 system, with the latest 2.2.13 kernel + raid patches, and the
latest raidtools. But, it doesn't work. When both drives are ok, it
works. If I powerdown, and disconnect the 'second' disk (higher SCSI id)
it boots. If, instead, I unhook the lower ID drive (and correct
termination of course), it gives me:
LILO:
Error 0x80
which indicates (quoting the lilo README): '"Disk timeout". The disk or
the drive isn't ready.' I also notice the floppy light has blinked. The
LILO prompt returns after this error, so I hit enter a few more times: lo
and behold, I can actually hear my pathetic floppy drive doing a seek each
time.
So I try again: after the LILO: prompt (before pressing enter, and before
the timeout) I stick a (random) floppy into the drive, and I get:
LILO:
Error 0x04
which indicates: '"Sector not found". This typically indicates a geometry
mismatch.' according to LILO documentation, confirming that the device
that LILO is accessing is the wrong one, my floppy instead of my hard
drive.
So the question is, why is the boot map on the second disk mangled so as
to attempt to load the kernel from my floppy! I don't know... so I'm
asking you all.
Also, does anyone know who wrote the lilo.raid1 patch? Ie. can I email
him/her for info?
--
/==============================\
| David Mansfield |
| [EMAIL PROTECTED] |
\==============================/
