Ronald G Minnich <[EMAIL PROTECTED]> writes:
> On Thu, 14 Jun 2001 [EMAIL PROTECTED] wrote:
>
> > I was curious about that myself. Okay, so it's kind of like the way LILO
> > boots the kernel, right? It does the hardware initialization and triggers
> > the MBR to get everything running, and then a command can be executed
> > remotely in a standard shell prompt to update the BIOS. Is that about
> > right?
>
> The way we update our Alpha DS10 nodes is pretty instructive. The nodes
> come up with Linux, then mount an init ram disk and join a scyld cluster.
>
> If we need to update flash we run a little script called 'flange' that
> does the deed.
>
> In fact the need to update flash is rare. Part of the Scyld boot process
> involves downloading a "real" kernel. So you only run that flash-based
> kernel for a second or so.
Ron. How is the palcode handled. Unless things are dramatically different
from how I set them up. The first kernel loads palcode, and then you are using
scyld to boot a second kernel. How do you prevent overwriting your palcode?
Not that this isn't a good process but if you have this problem solved
somewhere I would like to know. Currently that is my show stopper but
with my alpha code. Booting a kernel from a kernel that has loaded palcode.
Eric
