kern/54383: NFS root configuration without dynamic protocols: dhcp, bootp, etc ...

[Steve Shorter]

On Fri, Nov 28, 2003 at 04:35:25PM +0000, omestre wrote:
> On Tue, 25 Nov 2003, Steve Shorter wrote:
> 
>  Hello!
>  Well, when i did and post that patch, i have mentioned about "It's not a
> good/normal diskless solution"... but "it's still a solution! :))
> alternative solution...
>  I have 6 (six) production diskless servers (4.7) and some ones (5.0) in
> "tests". And
> are working fine!
>  But now, my diskless servers are booting from CDROM/RW (5MB) images. When
> i want change the kernels, well... in 30 minutes this task is done.
>  Thanks. (Just the fact of the freebsd hackers are "looking" my patch...
> it is just great).
>  Think, maybe other crazy guy want my solution. To many "Options", this is
> what OpenSource is all about. At least to me. With six or seven lines in
> the loader.conf file and a kernel+mypatch, you have a diskless server.
>  Thanks again.
> 

	I used to boot from 16 meg solid state M-SYS IDE Pro
devices. But abandoned that when I discovered how good
PXE support was in STABLE and how simple and reliable it was.
Your solution is OK for what you are doing.

	-steve



>  Leal.
> 
> > Date: Tue, 25 Nov 2003 13:09:36 -0500
> > From: Steve Shorter <steve at nomad.tor.lets.net>
> > To: freebsd-net at freebsd.org
> > Subject: kern/54383: NFS root configuration without dynamic protocols:
> >     dhcp, bootp, etc ...
> >
> > Howdy!
> >
> >
> > 	I only boot from Intel cards that support PXE.
> >
> > 	This greatly simplifies the administration
> > of large numbers of diskless machines. It makes kernel
> > changes trivial. If your going to go diskless you may
> > as well go all the way for the base/root system and kernel.
> >
> > 	If you need RAID disks for data then you can
> > throw some swap on them, but otherwise with memory
> > cheap you can usually get away without swap.
> >
> > 	The problem with hardrives isn't cost. Its a
> > source of failure and administrative overhead.
> > Avoid them. If you can't, then use RAID for data and
> > still PXE boot over the network.
> >
> > 	Upgrading kernel on each machine is a major drag
> > if there are lots of machines.
> >
> > 	In general a solution that is not scalable
> > is not a good one.
> >
> >
> > 	-steve
> > _______________________________________________
> > freebsd-net at freebsd.org mailing list
> > http://lists.freebsd.org/mailman/listinfo/freebsd-net
> > To unsubscribe, send any mail to "freebsd-net-unsubscribe at freebsd.org"
> >
> 
> omestre at sdf.lonestar.org
> SDF Public Access UNIX System - http://sdf.lonestar.org