Ben Russo wrote:
Well, I guess that might be a bug, open a bug in RedHat's Bugzilla, you might get a free T-shirt out of it.
ok.. I haven't done that before, some morning around 4am when I have some time to myself, might give it a try ;-)
I wouldn't use IDE disks in a resiliant system... I'm old school
But I think that their
answer will be that software RAID is intended to be used as a protection against data loss and unscheduled downtime from disk sector
failures during runtime, I don't think that software RAID can be used *reliably* for resiliency at boot time because an IDE disk could
fail in such a way that the PC BIOS detects the 1st bootable hard disk as present, and having a valid boot sector, but that cannot finish the full boot sequence.
and nothing but scsi disks on separate scsi controllers for me...
and if one bios doesn't boot, then the next will.. but of course
as you say below, it may only boot part way and that's why this is
just one node in a cluster... so if it goes down and doesn't come
back up cleanly, the STONITH device gets him anyway and he won't come back up. It's really more for complete disk failure so that I don't have to rebuild the system which takes a long time, but just shove a new disk in and let it rebuild.... but if I can't get it to boot from the good disk... well, I'm really backto where I started.
Anyway, Mr. Flory showed me how to list what is initrd linuxrc file
and I think that's the hacking I needed to be able to see how it works... I hope, though I am still having a lot of trouble getting
grub to boot the system with only the second scsi hard disk installed...
but I imagine I'll get it working at some point. To be honest, I probably should have used lilo, but... for some reason, lilo won't
over write grub in the MBR... I tried installing lilo repeatedly
and grub kept coming up, yes I ran lilo and it didn't complain,
but upon reboot, still get grub ... guess I'll have to fdisk /mbr
it, but everything I read on the grub web site and lilo howtos
says that lilo, should just overwrite grub... go figure.
Thanks again,
Bill
For example, let's say that the Head movement arm mechanism breaks so that the upper
cylinder range is unreachable, but the MBR, and the /boot partition
are reachable while the / partition is not. The PC BIOS would
start the boot, the kernel would load and then panic because
even though it could find the initrd image it couldn't read the
/etc/raidtab file. Having LILO or GRUB installed on both drives
is not a guarantee, it will only work in some limited circumstances.
Maybe RedHat doesn't offer to set up your box that way is because
they don't want to set false expectations that won't be lived up
to.
If you really want that type of redundancy you may have to go with
a hardware RAID solution.
-- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED]?subject=unsubscribe https://listman.redhat.com/mailman/listinfo/redhat-list
