Done.

On Mon, 2002-09-23 at 11:38, Brian Finley wrote:
> On Mon, 2002-09-23 at 10:29, Sean Dague wrote:
> > On Mon, Sep 23, 2002 at 10:18:32AM -0500, Brian Finley wrote:
> > > On Fri, 2002-09-20 at 12:14, Sean Dague wrote:
> > > > On Fri, Sep 20, 2002 at 10:49:45AM -0500, [EMAIL PROTECTED] wrote:
> > > > > At this point I rebooted the system and it came up with the wrong node
> > > > > name.
> > > > > The "golden" client was n06, the client I had updated was n07, but the node
> > > > > name in the login prompt was n06. The file /etc/sysconfig/network had not
> > > > > been changed to reflect n07.
> > > > 
> > > > This is probably because the default 'static_dhcp' option was chosen, which
> > > > doesn't set the hostname.  If dynamic_dhcp or static ip-method is selected,
> > > > the hostname would have been set.
> > > 
> > > Do we want to try and fix this prior to the 3.0 release?
> > > 
> > > The proposal for the fix is two-fold:
> > > 
> > >         * The first bit is to change static_dhcp and dynamic_dhcp to
> > >           simply be "dhcp" (and to not mess with the hostname at all).
> > > 
> > >         * The second bit is "how do we deal with hostnames?".  Certainly
> > >           the default will be to assign them.  
> > >     
> > >     The real question is how to deal with them if users don't want them
> > >     assigned.  Do we add a "-blank-hostname" option to getimage?  Or a
> > >     "-set-all-hostnames-to NAME" option (I can see some people wanting a
> > >     common hostname across a farm especially with an UltraMonkey type
> > >     solution)?  Or do we simply put a comment into the master
> > >     autoinstall script that looks like:
> > 
> > Currently, there is no provision for blank hostnames on nodes anyway, as
> > what happens is the hostname of the golden-client will propogate to all the
> > nodes.  
> 
> Ok.  I think I know when this happened and why.  I think it was because 
> this feature wasn't well understood when that critical bit was removed
> from the code (the hostname not being blanked out).
> 
> > Making all the nodes have the same hostname doesn't make any sense
> > to me.
> 
> It doesn't in most environments.  But in certain LVS configurations,
> where redirection is to to MAC addresses of machines in a farm behind
> the load balancer, and the destination IP address is changed to a
> loopback address (such as 127.0.0.2), it can be useful to have all of
> the machines set to have the same hostname.
> 
> > My proposal is:
> >  * drop static_dhcp all together
> >  * rename dynamic_dhcp to dhcp
> >  * don't try to solve the blank hostname issue unless it actually arrises
> >    (i.e. someone needs it on the list).
> 
> Well, we know that some people are using this -- Dave Lombard for
> example (using SystemImager 1.5).
> 
> > If someone is doing something really esoteric with SIS dealing with blank
> > hostnames, then we'll deal with it when it comes up.  
> 
> It's been a feature in SystemImager for a long time (ever since there
> was an -ip-assignment option).  
> 
> I understand your proposal and I agree.  
> 
> Heres a recap:
> 
> o Get rid of static_dhcp and dynamic_dhcp in favor of simply "dhcp" (and
>   don't mess with the hostname at all).
> o Add these comments to autoinstallscripts.
>   ##################################################################
>   #
>   # Uncomment the line below to leave your hostname blank.  Certain
>   # distributions use this as an indication to take on the hostname
>   # provided by a DHCP server.  The default is to have
>   # SystemConfigurator assign your clients the hostname that
>   # corresponds to the IP address the use during the install.
>   #
>   #HOSTNAME=""
> 
> 
> 
> 
> -Brian
> 
> 
> 
> > Otherwise we are doing
> > the wrong things for thousands of people, on the off chance that some
> > ancient Red Hat function a) still works b) is used by anyone.
> > 
> >     -Sean
> > 
> > -- 
> > _______________________________________________________________________
> >                             
> > Sean Dague                [EMAIL PROTECTED]               http://dague.net
> > 
> > There is no silver bullet.  Plus, werewolves make better neighbors than
> > zombies, and they tend to keep the vampire population down.
> > _______________________________________________________________________
> -- 
> ---------------------------------------------------
>  Brian Elliott Finley  http://baldguysoftware.com/
>   phone: 214.280.8188  http://systemimager.org/           
> ---------------------------------------------------
-- 
---------------------------------------------------
 Brian Elliott Finley  http://baldguysoftware.com/
  phone: 214.280.8188  http://systemimager.org/           
---------------------------------------------------



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Sisuite-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/sisuite-devel

Reply via email to