niles at scyld.com
Mon Jan 22 22:31:17 PST 2001
> After doing a "beofdisk -w", you have to do a reboot for the partition
> table to be written. On scyld software, you can reboot with
> "bpctl -S <node_number> -s reboot"
> That should get it to work.
Hmmm... I'm not sure I agree. You shouldn't have to reboot to have
the partition table written.
Let go though what I think is happening...
> beoboot -q
> beoboot -w
First off I sure hope you mean "beofdisk -q" and not "beoboot -q".
Second, if you slave nodes have some previous partitioning scheme
(i.e. this disks have been used for something else before) and you
want the default Scyld scheme...do "beofdisk -q -d" which will write
out the default partition table (to a /etc/beowulf/fdisk file) that
will include a partition for the beoboot image. Then do "beofdisk -w"
to write it out.
After this, "beoboot-install -a /dev/hda" should work...
Scyld Computing Corporation
("author of beofdisk")
More information about the Beowulf