[ipxe-devel] Implicit SAN+PXE?

[Jarrod Johnson]

Ok, patch for review at:
https://xcat.svn.sourceforge.net/svnroot/xcat/xcat-dep/trunk/xnba/ipxe-registersan.patch

<https://xcat.svn.sourceforge.net/svnroot/xcat/xcat-dep/trunk/xnba/ipxe-registersan.patch>This
patch was 'good enough' for my purposes.

It does request another 256 byte set of chars.

There is an amount of code duplication between 'boot_root_path' and
'reg_root_path' in this approach.

One way to fix both would be to have the hook & describe before trying to
boot anything, then either filename or san_boot, so that both have a common
hook&describe.  I was cautious and also was unsure if we should consider
hook, parse, and describe errors as not particularly fatal in the boot
filename whereas in the sanboot case it would warrant more pessimistic
output...

Thankfully, the hook/describe/boot is nice and segregated now, so a patch
like this is trivial.


On Thu, Nov 11, 2010 at 6:51 PM, Michael Brown <mbrown at fensystems.co.uk>wrote:

> On Thursday 11 Nov 2010 22:27:03 jarrod.b.johnson at gmail.com wrote:
> > I do, will post tomorrow for review.  It is conservative and may be
> >  better done through refactoring boot_root_path in half...
> >
> > One objection I could see.  IIRC, freebsd is a case where the pxe
> >  payload will interpret root_path for nfs root and ipxe trying to
> interpret
> >  could be odd..
>
> I'm assuming that we should handle an existent but incomprehensible
> root-path
> (e.g. an NFS root-path) by noisily failing to hook the SAN device and then
> continuing with the normal filename boot.
>
> Michael
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ipxe.org/pipermail/ipxe-devel/attachments/20101112/9ddc0eb7/attachment.htm>