Hi Ted:

I think Erich checked something in to that effect, but if you have a
patch against mkdhcpconf please feel free to post it.

FYI the latest source code is available here:

http://svn.oscar.openclustergroup.org/trac/oscar/browser/pkgsrc/systemin
staller-oscar/trunk/bin/mkdhcpconf

Cheers,

Bernard 

> -----Original Message-----
> From: Ted Powell [mailto:[EMAIL PROTECTED] On Behalf 
> Of Ted Powell
> Sent: Tuesday, June 06, 2006 16:16
> To: Neil Costigan; Bernard Li; Erich Focht
> Cc: [email protected]
> Subject: ISC DHCP Server
> 
> On Mon, Jun 05, 2006 at 09:48:29PM +0100, Neil Costigan wrote:
> > 
> > a solution was emailed to me direct...
> > thanks to Igor N.
> > 
> > >I am not sure if you have figured out the problem you had with  
> > >etherboot and dhcpd in FC5.
> > >Just in case, I had a similar problem - etherboot stopped 
> accepting  
> > >dhcpd offers after upgrade to FC5.
> > >
> > >After banging my head against the wall for a couple of 
> days, I have  
> > >found a solution.
> > >For some reason etherboot wants to see the "next-server" 
> option in  
> > >the DHCP offer. And for some reason the new version of 
> dhcpd in FC5  
> > >does not send that option.
> 
> FC5 uses the ISC (Internet Systems Consortium) DHCP server. In any
> version after 3.0.2, the following release note applies:
> 
> - The siaddr field was being improperly set to the 
> server-identifier when
>   responding to DHCP messages.  RFC2131 clarified the siaddr field as
>   meaning the 'next server in the bootstrap process', eg a 
> tftp server.
>   The siaddr field is now left zeroed unless next-server is 
> configured.
>        (/usr/share/doc/dhcp-3.0.3/RELNOTES)
> 
> The relevant paragraph of RFC2131 is:
> 
>    DHCP clarifies the interpretation of the 'siaddr' field as the
>    address of the server to use in the next step of the client's
>    bootstrap process.  A DHCP server may return its own address in the
>    'siaddr' field, if the server is prepared to supply the next
>    bootstrap service (e.g., delivery of an operating system executable
>    image).  A DHCP server always returns its own address in 
> the 'server
>    identifier' option.
>        (http://rfc.net/rfc2131.html)
> 
> The change in the ISC server is appropriate, because unless 
> the server's
> administrator tells it so, it has no business claiming 
> capabilities of the
> host computer that it does not itself provide. If a client 
> machine has no
> other default to try, it can always use the 'server 
> identifier' option to
> try the machine hosting the DHCP server. The PXE code in my 
> host machine
> does so, and successfully downloads pxelinux.0, but 
> pxelinux.0 does not,
> and so it was trying to download its config file from 0.0.0.0. Once I
> added "next-server 192.168.150.101;" to /etc/dhcpd.conf, 
> things worked.
> 
> I was going to make a patch for mkdhcpconf, but I see that Bernard is
> already working with that file.
> 
> BTW, unlike the "routers" option (e.g. option routers 
> 192.168.150.101;),
> "next-server" is a _statement_ (e.g. next-server 192.168.150.101;),
> contrary to the quote from Igor N.
> 
> 
> -- 
> Ted Powell <[EMAIL PROTECTED]>   http://psg.com/~ted/
> "If you don't look, you don't know."
>     Dr. Sam Ting, Nobel laureate experimental physicist.
> 


_______________________________________________
Oscar-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/oscar-devel

Reply via email to