FC5 RAID 1 - booting off either drive

Craig White craigwhite at azapple.com
Sat Apr 1 06:24:03 UTC 2006


On Fri, 2006-03-31 at 23:34 -0600, Laurence Vanek wrote:
> Craig White wrote:
> > On Tue, 2006-03-28 at 08:05 -0700, Craig White wrote:
> >   
> >> On Mon, 2006-03-27 at 21:43 -0600, Laurence Vanek wrote:
> >>     
> >>> Greetings -
> >>>
> >>> To my surprise, fresh install of FC5 with RAID 1 configured at install 
> >>> time went without a hitch.  I have /dev/md0 defined on /boot 
> >>> partitions.  It boots fine with this setup.
> >>>
> >>> I had my FC4 system setup this exact way except I found that I had to 
> >>> manually install grub on one of the disks to permit booting off either 
> >>> one if the other is down.
> >>>
> >>> Does anyone know of a way to check for the presence of grub on each 
> >>> drive (obviously its on hda since it boots presently) to see if this 
> >>> needs to be done for FC5 also without powering down each drive in turn & 
> >>> attempting to boot?
> >>>       
> >> ----
> >> thought I should post against the right thread
> >>
> >> just an untested thought...
> >>
> >> bb if=/dev/sda of=/tmp/sda-bootblock.bin bs=512 count=1
> >> bb if=/dev/sdb of=/tmp/sdb-bootblock.bin bs=512 count=1
> >>
> >> diff /tmp/sda-bootblock.bin /tmp/sdb-bootblock.bin
> >>
> >> Thinking...
> >>
> >> - substitute different values for sda/sdb as fits
> >> - the first 512 bytes on each drive are the boot (perhaps less, someone
> >> will surely correct me...it might just be the first 256 or 384 bytes)
> >> - if they are the same (i.e. grub has been installed on both), there
> >> will be no diff
> >>
> >> otherwise...
> >>
> >> grub-install /dev/sdb
> >>
> >> Of course, the only way that you'll ever KNOW for sure that it's going
> >> to work is to do a real simulation, i.e. disconnecting one drive, then
> >> the other drive...
> >>     
> > ----
> > not only wrong thread but as Paul so correctly points out.
> >
> > dd - NOT bb
> >
> > I am rapidly ingesting coffee but haven't figured out a mainline method
> > so I am left to normal ingestion methods which apparently haven't quite
> > caught up to my fingers or my brain.
> >
> > duh.
> >
> > Sorry
> >
> > Craig
> >
> >   
> Craig -
> 
> results using bs=466
> ============
> [root at localhost tmp]# dd if=/dev/hda of=/tmp/hda-bootblock.bin bs=466 
> count=1
> 1+0 records in
> 1+0 records out
> 466 bytes (466 B) copied, 7.6e-05 seconds, 6.1 MB/s
> [root at localhost tmp]# dd if=/dev/hdc of=/tmp/hdc-bootblock.bin bs=466 
> count=1
> 1+0 records in
> 1+0 records out
> 466 bytes (466 B) copied, 7.7e-05 seconds, 6.1 MB/s
> [root at localhost tmp]# diff hda-bootblock.bin hdc-bootblock.bin
> Binary files hda-bootblock.bin and hdc-bootblock.bin differ
> ================
> 
> files also differ is bs=256, 128
> BUT for bs=64 we get no complaint from diff:
> 
> ================
> [root at localhost tmp]# dd if=/dev/hda of=/tmp/hda-bootblock.bin bs=64 count=1
> 1+0 records in
> 1+0 records out
> 64 bytes (64 B) copied, 7.8e-05 seconds, 821 kB/s
> [root at localhost tmp]# dd if=/dev/hdc of=/tmp/hdc-bootblock.bin bs=64 count=1
> 1+0 records in
> 1+0 records out
> 64 bytes (64 B) copied, 7.6e-05 seconds, 842 kB/s
> [root at localhost tmp]# diff hda-bootblock.bin hdc-bootblock.bin
> ================
----
well - you sort of blew my theory right out of the water - thanks.

the theory being, that the grub boot blocks on both drives should be the
same - I guess that something is different on both drives but in theory,
I was thinking that if it were installed...

grub-install /dev/hda
grub-install /dev/hdc

then they should both be the same boot blocks. On yours, the boot blocks
on /dev/hda were written by the installer (anaconda) and on /dev/hdc
were written by you manually by issuing 'grub' command and stepping
through it manually as I recall.

;-(

Well, I'm glad it worked...sorry I was really of little help.

Craig




More information about the fedora-list mailing list